Re: [blfs-support] blfs networking
On 04/16/2013 03:50 PM, lux-integ wrote: > On Tuesday 16 April 2013 19:57:24 Bruce Dubbs wrote: >> Whether to use ipv4 or ipv6 really depends on the ISP. Generally an >> office only needs one address and then uses NAT (network address >> translation) internally. NAT multiplexes a single address into many. >> Any office can use private IP addresses internally (192.168.x.x, >> 10.x.x.x, 172.16,x,x) and one ipv4 (or ipv6) externally. > > the problem is some useful applications such as IPsec and VoIP do not like > NAT . Who says? Only poor implementations don't deal well with it, but single NAT should work fine for almost everything now days. > Also one NAT-router behind another and NAT becomes problematic. > AND one has to resort to all sorts of convolutions such as stun-servers, > nat-traversal, port-forwarding etc to attempt to use these. > This is referred to as double NAT and is likely to be really supported or supportable in very few scenarios. Don't double NAT! > here is a far better ( than the one I posted previously ) utube video on ipv6 > http://www.youtube.com/watch?v=9UHZ7eX_f30 > Many of the people who were conscious of networking in the early '90s still consider NAT a poor hack at best. The security provided by NAT is only a side effect of having broken everything. Double NAT is broken in the same way that NAT is, only doubly, because IPv4 in its inception was not intended to be used as it is today. Though I've not inadvertently dealt with double-NAT in several years, it was my experience that it never worked exactly as intended (at least not with the cheaper devices that I could afford at the time). With that said, NAT has made us quite lazy WRT security. It was quite a shock at fist, but my traffic rules increased 4 fold on the v6 side compared to my v4 rules when I handed my previously internal only home boxes routable v6 addresses (my take was that I had better do it on my overly complex home network before tackling it in the field). Now, I must state that everything _is_ dual stack, and was from the very beginning, because I can only actually use v6 at a handful of places. Auto-IP works flawlessly with Linux (and has for at least the past few years)-- IOW, no need for an IPv6 DHCP server as long as your router supports router advertisement, but you must still provide static DNS entries manually if you should want to address your DNS servers via v6 (your v4 DNS servers will happily return addresses if requested so just handle it via DHCP). Windows and Android devices also work flawlessly (preferring v6 to v4 when possible), no idea about OS X and iOS (not to be confused with IOS which has worked just fine with v6 for several years), though I suspect they too should be fine at this point in time. If you do intend to implement v6, the big thing is that you must be conscious of the fact that you are exposing your entire internal network to the live internet (if that term still means anything). Don't be afraid to be overly cautious. Basically begin by denying everything in and out, then add to the top of the stack and work from least traffic to most traffic, then most restrictive to least restrictive as you go up the chain, being as explicit with your traffic rules as is possible. For instance, IP/proto:port in->out goes at the very top, followed by ICMP and ICMPv6 rules in the middle, then IP/proto:port out->in near the bottom, then deny everything else at the very bottom. You can always insert rules to allow more traffic, but if you leave a gaping hole from the very beginning...well, you are likely to have a gaping hole when you think you are finished. You will also likely need to run dual stack for quite some time because v6 adoption is moving very slowly. My advice, having done it myself: Don't do it yet! At least, not until you have a real world need to do so. Unless you are doing it for training, or have a real need that cannot be addressed with existing v4, is just a headache for very little benefit at the moment. HTH --DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] SSL certificates
On 01/28/2013 05:44 PM, DJ Lucas wrote: > > I get the same behavior with openssl s_client unless I explicitly set > CApath, which makes me wonder if our OpenSSL installation is slightly > broken. I'll look into that quick. > Oops, already been though this once before. I had forgotten, but this is expected behavior. -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Stripping
On 01/28/2013 10:54 AM, Randy McMurchy wrote:
> Fernando de Oliveira wrote these words on 01/28/13 10:41 CST:
>> Forwarded from the "BLFS Book Maintenance List" list.
>>
>> Sorry for top posting.
>>
>> Thanks, Randy.
>
>
> Though essentially the same thing Bruce said, here is what I do at the
> completion of LFS. Simply modify the log file locations and include any other
> directories you wish and you may like the results:
>
>
> du -sch {,usr/}{sbin,bin} \
> >home/rml/build/Logs/LFS_System/Post-Installation/strip-bin.log 2>&1
> echo \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-bin.log
> find sbin bin usr/sbin usr/bin -type f -exec strip --strip-all
> --preserve-dates {} \; \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-bin.log 2>&1
> echo \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-bin.log
> du -sch {,usr/}{sbin,bin} \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-bin.log 2>&1
> cat home/rml/build/Logs/LFS_System/Post-Installation/strip-bin.log | \
> grep -v "File format not recognized"
>
>
> du -sch {,usr/}lib \
> >home/rml/build/Logs/LFS_System/Post-Installation/strip-lib.log 2>&1
> echo \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-lib.log
> find lib usr/lib -type f -exec strip --strip-debug --preserve-dates {} \; \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-lib.log 2>&1
> echo \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-lib.log
> du -sch {,usr/}lib \
> >>home/rml/build/Logs/LFS_System/Post-Installation/strip-lib.log 2>&1
> cat home/rml/build/Logs/LFS_System/Post-Installation/strip-lib.log | \
> grep -v "File format not recognized"
>
Just FYI, I ran across this comment in Arch's PKGBUILD for glibc (which
I've not verified, and blindly followed):
# Do not strip the following files for improved debugging support
# ("improved" as in not breaking gdb and valgrind...):
# ld-${pkgver}.so
# libc-${pkgver}.so
# libpthread-${pkgver}.so
# libthread_db-1.0.so
Now, as to how useful that actually is in practice, I can't honestly see
a need being that everything else is already lacking...
Anyway, --strip-debug for static libs, --strip-unneeded for shared libs,
and --strip-all for executable files is what is default for makepkg (and
coincidentally is what I have stuck with for a while).
-- DJ
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] SSL certificates
On 01/28/2013 01:47 PM, Jean-Philippe MENGUAL wrote:
> Hi,
>
> Thanks very much, this helps a lot. I didn't experience any problems following
> the process. But I still get:
>
> fetchmail: No mail for moderate...@absolinux.net at imap.1and1.fr
> fetchmail: Server certificate verification error: self signed certificate in
> certificate chain
> fetchmail: This means that the root signing certificate (issued for
> /C=SE/O=AddTrust AB/OU=AddTrust External TTP Network/CN=AddTrust External CA
> Root)
> is not in the trusted CA certificate locations, or that c_rehash needs to be
> run on the certificate directory. For details, please see the
> documentation of --sslcertpath and --sslcertfile in the manual page.
> fetchmail: Warning: the connection is insecure, continuing anyways. (Better
> use --sslcertck!) fetchmail:
> No mail for te...@aaui.eu at mail.accelibreinfo.eu fetchmail: Server
> certificate verification error: unable to get local issuer certificate
> fetchmail: This means that the root signing certificate (issued for
> /CN=actux.eu.org) is not in the trusted CA certificate locations, or that
> c_rehash needs
> to be run on the certificate directory. For details, please see the
> documentation of --sslcertpath and --sslcertfile in the manual page.
> fetchmail: Server certificate verification error: certificate not trusted
> fetchmail: Server certificate verification error: unable to verify the first
> certificate
> fetchmail: Warning: the connection is insecure, continuing anyways. (Better
> use --sslcertck!)
>
> Is some certificate missing? I don't understand why fetchmail still complains.
>
> Thanks for your answer.
>
> Regards,
>
> JPM
>
>
> On Sunday 27 Jan 2013 à 19:27:26 (-0600), DJ Lucas wrote:
>> On 01/27/2013 06:18 AM, Jean-Philippe MENGUAL wrote:
>>> Hi,
>>>
>>> Thanks very much for the information. I probably didn't understand
>>> everything
>>> in the process, anyway. Indeed, in make-ca.sh, I replaced
>>> BUNDLE="BLFS-ca-bundle-${VERSION}.crt" with AddTrustExternalCARoot.crt.
>>> Then I ran script. I also updated mozilla's certs, through the proces
>>> described
>>> in the book and also with mozilla-root.crt?.
>>>
>>> So .pem are all updated and generated. Is it enough? Should the
>>> ca-bundle.crt
>>> be updated itself? Because with such process, fetchmail displays the same
>>> thing.
>>>
>>> Did I misunderstand something in this process of certificates?
>>>
>>> Thanks very much and sorry to disturb but I've to say that this security
>>> concepts
>>> are not natural for me.
>>>
>>> Best regards,
>>>
>>
>>
>> The steps should be:
>>
>> certhost='http://mxr.mozilla.org' &&
>> certdir='/mozilla/source/security/nss/lib/ckfw/builtins' &&
>> url="$certhost$certdir/certdata.txt?raw=1" &&
>> wget --output-document certdata.txt $url &&
>> unset certhost certdir url &&
>> make-ca.sh &&
>> remove-expired-certs.sh
>>
>> Those update to the latest Mozilla certs, and the following adds your
>> new CA root to the trusted certs:
>>
>> keyhash=$(openssl x509 -noout -in AddTrustExternalCARoot.crt -hash) &&
>> cp AddTrustExternalCARoot.crt \
>> /etc/ssl/certs/${keyhash}.pem &&
>> c_rehash &&
>> unset keyhash
>>
>> To update the bundle, with what is currnetly in /etc/ssl/certs, run the
>> following command at any time:
>>
>> cat /etc/ssl/certs/*.pem > /etc/ssl/ca-bundle.crt
>>
>> -- DJ
>>
>> --
>> http://linuxfromscratch.org/mailman/listinfo/blfs-support
>> FAQ: http://www.linuxfromscratch.org/blfs/faq.html
>> Unsubscribe: See the above information page
Not a fetchmail user, so this is second hand, but I think you'll need to
set sslcertpath in fetchmailrc.
I get the same behavior with openssl s_client unless I explicitly set
CApath, which makes me wonder if our OpenSSL installation is slightly
broken. I'll look into that quick.
-- DJ
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] SSL certificates
On 01/27/2013 06:18 AM, Jean-Philippe MENGUAL wrote:
> Hi,
>
> Thanks very much for the information. I probably didn't understand everything
> in the process, anyway. Indeed, in make-ca.sh, I replaced
> BUNDLE="BLFS-ca-bundle-${VERSION}.crt" with AddTrustExternalCARoot.crt.
> Then I ran script. I also updated mozilla's certs, through the proces
> described
> in the book and also with mozilla-root.crtŒ.
>
> So .pem are all updated and generated. Is it enough? Should the ca-bundle.crt
> be updated itself? Because with such process, fetchmail displays the same
> thing.
>
> Did I misunderstand something in this process of certificates?
>
> Thanks very much and sorry to disturb but I've to say that this security
> concepts
> are not natural for me.
>
> Best regards,
>
The steps should be:
certhost='http://mxr.mozilla.org' &&
certdir='/mozilla/source/security/nss/lib/ckfw/builtins' &&
url="$certhost$certdir/certdata.txt?raw=1" &&
wget --output-document certdata.txt $url &&
unset certhost certdir url &&
make-ca.sh &&
remove-expired-certs.sh
Those update to the latest Mozilla certs, and the following adds your
new CA root to the trusted certs:
keyhash=$(openssl x509 -noout -in AddTrustExternalCARoot.crt -hash) &&
cp AddTrustExternalCARoot.crt \
/etc/ssl/certs/${keyhash}.pem &&
c_rehash &&
unset keyhash
To update the bundle, with what is currnetly in /etc/ssl/certs, run the
following command at any time:
cat /etc/ssl/certs/*.pem > /etc/ssl/ca-bundle.crt
-- DJ
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] For {,anti}systemd folks
On 01/26/2013 04:07 PM, Bruce Dubbs wrote: > Armin K. wrote: >> http://0pointer.de/blog/projects/the-biggest-myths.html >> >> I am just sharing this. Please don't start any arguments why something >> he said is true or not. > > Not an argument, a discussion. > > There are a few things that he left out. I don't particularly care for systemd, but I've found it to be usable enough to support it if the majority is moving in that general direction. Standardization is a good thing IMO, so it gets my vote (even if Debian and its derivatives can't or are unwilling to adopt it). > > systemd uses binary log files that can't use standard tools like grep. > Myth or fact? Myth. This was covered in the referenced blog post. It brings additional functionality, and yes, the systemd executable itself does log to a binary format via the journal, but does not replace, nor hinder operation of a traditional syslog (and actually adds functionality by way of the early logging). We'd still have to do manual parsing of the binary log to get our boot.log file if it is still warranted, but that's less than what we do in rc today to get it. I find it kind of ironic that we attempted to partially solve our own early logging problem by modifying our init system (and using quite a few more tools, although, in our defense, those tools were already present on the system). :-) > > systemd does not support building udev without the rest of systemd. > Myth or fact? Fact. And that is one very annoying fact given the simplicity of rectifying it, but again, I also _understand_ the developers' take on it as well. It is one configuration that will only be used by a handful of its consumers. To get only the basics, we are talking dbus, glib, and libcap2. > > systemd solves problems encountered by many/most LFS users. Myth or fact? > Fact. LFS already uses part of systemd (udevd), BLFS will require more of it if it intends to continue to provide Gnome and KDE (logind), and security is everyone's continual problem (cgroups, though the security advantage is entirely mute if you happen to be an SE user already). Early logging also just got a little bit better to boot as it covers more than just boot script output when combined with a syslog daemon and one dump at the end of init (if you want/need systemd's logs to go to a standard syslog format). > systemd boots significantly faster than LFS on the same hardware. Myth > or fact? > Unable to answer. The question is too subjective because there are far too many variables. If you want a real answer, please define both 'boot' and 'LFS'. If LFS is inclusive of a complete BLFS system where you are nearing the 100 startup items of a typical distro, then yeah, it should be decidedly faster (especially if you consider 'boot' to end at the instant a login prompt is displayed). A noticeable speed difference could be realized if you happen to be running several instances of say Apache httpd, Tomcat, SQL, or any number of other services that take a considerable amount of time, say 100ms, to pass a return value to the shell that called it. Whether the time savings is significant is entirely subjective (even my suggestion of 100ms being considerable is subjective). I'll accept that it _might_ even be slower in some cases, but I've not looked for evidence one way or the other. But then again, if speed is the only concern, the rc script could be replaced with a single script with 7 cases and each called script followed by a single ampersand character if parallelism is desired. The output wouldn't be pretty with current scripts, but it'd be functional none the less. > systemd has/needs over 100 pages of documentation. Myth or fact? > Fact/Myth You asked two questions and I answered both. Yes it has great documentation! Whether the average user needs all of the API documentation is questionable. I will second Simon's earlier response, however. :-) HTH to maybe paint it in at least a partially more positive light. -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] SSL certificates
On 01/26/2013 09:46 AM, Jean-Philippe MENGUAL wrote: > Hi, > > How could I get rid of this warning/error from fetchmail: > http://sprunge.us/YIUJ > > This seems to make a output code different from 0, which prevents a script > from > running properly, so I'd like to fix the problem. I didn't find any way to > make fetchmail pass without problem (even quiet displays this). How can it > accept > these certs? > > I could add them to CA-certs, following: > http://www.linuxfromscratch.org/blfs/view/svn/postlfs/cacerts.html > but I have not understood how I can fetch the non-trusted certificate. > Finally, I must write not properly my search, I don't find any tutorial for > this on the Internet. You won't, other than what is in the BLFS book because everyone does it differently. Being BLFS, you are expected to do it manually. The root you seek is available here: https://support.comodo.com/index.php?_m=downloads&_a=viewdownload&downloaditemid=87 1st link from a Google search for "AddTrust external ca root". Anyway, steps are exactly what is in the book now. Probably best to first update with current Mozilla certs, then download the one above and drop it into /etc/ssl/certs, run c_rehash, and finally recreate /etc/ssl/ca-bundle.crt from the newly installed certs. Might not even be a bad idea to create a directory for other trusted certs in /etc/ssl and include them in the original script from the book. -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] XORG 7.7 libraries
On 09/30/2012 11:01 AM, Carl Thorn wrote: > I have recently completed build LFS-7.2 and have installed GPM, Lynx, > Screen, along with other programs and libraries in the current stable > BLFS book. I am trying to install XORG-7.7. XORG_PREFIX=/usr. > Util-macros, XORG protocol headers, makedepend, libXau, libXdmcp, > libpthread-stubs, xcp-proto, and libxcb all installed with no problems. > When install XORG libraries xtrans-1.2.7 installs without error. But > libX11-1.5.0 fails during configure with this error:/checking keysym > definitions... configure: error: Cannot find keysymdef.h/. How do I > locate what is causing the problem and how do I fix it? > > I haven't seen that particular error, but it looks like at least the package xproto wasn't installed correctly. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] CUPS problem: lp: Unsupported document-format "text/plain".
On 06/25/2012 04:44 PM, Jeremy Henty wrote: > > I wrote: > >> CUPS refuses to work on my latest LFS system - any attempt to print >> produces 'lp: Unsupported document-format "text/plain".' . > > Fixed! I had to configure ghostscript-9.05 with --with-install-cups . > This installs the gstoraster filter program and the gstoraster.convs > configuration file that tells CUPS how to convert PostScript to raster > (I should have said that I don't have a PostScript printer.) > Ghostscript-8x didn't need this parameter. FYI, I went out and bought a $40 HP 3050 AIO to scan with, though I haven't used it in some time, I figured I might possibly need printing sometime too. Scanning worked perfectly, but I was having difficulties printing via HPLIP, every page was stretched and spanning four pages. I re-installed GS with the option above and all is well. > > Perhaps this is worth mentioning on the BLFS Ghostscript page? > Being that it causes no ill effects for users of PS printers, it should probably be default (circular dependency for CUPS?). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Network Manager configuration
On 07/08/2012 02:06 PM, Armin K. wrote: > On 07/08/2012 07:44 PM, DJ Lucas wrote: >> On 07/08/2012 10:36 AM, DJ Lucas wrote: >>> On 07/08/2012 08:37 AM, DJ Lucas wrote: >>>> On 07/08/2012 02:37 AM, Armin K. wrote: >>>>> On 07/08/2012 04:53 AM, DJ Lucas wrote: >>>>>> Anybody have an idea of how to allow a regular user (as a member of a >>>>>> dedicated group) to modify a wired connection's configuration? With the >>>>>> config file I've added to the BLFS Wiki (well, similar to that one, I've >>>>>> divided it up as I had suggested), >>>>>> http://wiki.linuxfromscratch.org/blfs/wiki/NetworkManager I'm able to >>>>>> enable and disable the connections, but I cannot edit settings using the >>>>>> gnome nm applet on wired connections. I'm sure I'm missing something >>>>>> simple, but can't for the life of me find it. I've tried various changes >>>>>> to policies with no change. Also, clicking network settings from the >>>>>> tray does nothing...I have to open nm-connection-manager manually. >>>>>> >>>>>> -- DJ Lucas >>>>>> >>>>> >>>>> Does it work? It looks to me like something I found in Debian not so >>>>> long ago, but I've noticed that network manager will work without that. >>>>> However, you need to make sure about two things: polkit-gnome MUST be >>>>> running. Probably the reason why you can't edit connections, because it >>>>> would require polkit authentification agent to be running. Another main >>>>> reason is that your session needs to be registered with consolekit and >>>>> also marked as active (I suggest using Linux PAM there, since I don't >>>>> know any other way). >>>>> >>>>> $ ck-list-sessions >>>>> Session2: >>>>> unix-user = '1000' >>>>> realname = 'Armin K' >>>>> seat = 'Seat1' >>>>> session-type = '' >>>>> active = TRUE >>>>> x11-display = ':0' >>>>> x11-display-device = '/dev/tty7' >>>>> display-device = '' >>>>> remote-host-name = '' >>>>> is-local = TRUE >>>>> on-since = '2012-07-08T06:55:14.236487Z' >>>>> login-session-id = '2' >>>>> >>>>> My session is started via GDM, so x11-display-device is /dev/tty7. You >>>>> might get different output there, but you need to make sure that active >>>>> = TRUE and is-local = TRUE is really present, or you will be unable to >>>>> do any "administrative" stuff. >>>>> >>>> >>>> There it is. I'm not using gdm so the x attributes aren't set. active = >>>> FALSE. Gnome only needs to be started by gnome-session in .xinitrc, >>>> thanks to the PAM configuration you added (dbus-lauch >>>> --exit-with-session is no longer required), however, I do not see >>>> polkit-gnome-authentication-agent-1 in ps output. That'll give me >>>> something to work from, however. >>>> >>> >>> Argh! You have this covered in the book already, I just didn't catch it >>> when updating my gnome scripts to latest. I didn't get the autostart >>> file installed. >> >> BTW, still not fixed. I'm unable to get an active ck session. Been >> staring at this all day and just now realized that the login-session-id >> was not set. Just rebuilt kernel with CONFIG_AUDITSYSCALL=y (I have no >> idea why I would have even messed with auditing support. Still not >> fixed, but I suspect it has to do with dbus-launch now. Will report back >> when fixed. >> >> -- DJ >> >> > > You need to use ck-launch-session dbus-launch --exit-with-session > gnome-session > > dbus-launch is needed if gnome-session is run from command line afaik ... > > http://www.linuxfromscratch.org/blfs/view/svn/gnome/gnome-session.html > Yep, did all of that, amongst about 100 other options, including separating dbus-launch and ck. No matter what happened, I was unable to obtain either local or active session with XDG attributes. > as you see there is full command in the configuration section. > > To get valid session id, you need to use pam_loginuid and > pam_ck_connector in system session OR in login configuration file (the > consolekit pam adition) > Unfortunately, the pam modules don't really make a difference here as they were enabled in the configuration and have already done their job at login. I was using startx from a terminal. There are literally thousands of threads on this topic, and eventually some users are able to make it work, but I (among many others) was able to gain no clear understanding of how. I just got T'd off and let it die. I installed GDM as I couldn't make ConsoleKit play nicely, and all is fine when starting from GDM (as you said earlier). It just works! Other minor issues brought up on -dev for discussion. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Network Manager configuration
On 07/08/2012 10:36 AM, DJ Lucas wrote: > On 07/08/2012 08:37 AM, DJ Lucas wrote: >> On 07/08/2012 02:37 AM, Armin K. wrote: >>> On 07/08/2012 04:53 AM, DJ Lucas wrote: >>>> Anybody have an idea of how to allow a regular user (as a member of a >>>> dedicated group) to modify a wired connection's configuration? With the >>>> config file I've added to the BLFS Wiki (well, similar to that one, I've >>>> divided it up as I had suggested), >>>> http://wiki.linuxfromscratch.org/blfs/wiki/NetworkManager I'm able to >>>> enable and disable the connections, but I cannot edit settings using the >>>> gnome nm applet on wired connections. I'm sure I'm missing something >>>> simple, but can't for the life of me find it. I've tried various changes >>>> to policies with no change. Also, clicking network settings from the >>>> tray does nothing...I have to open nm-connection-manager manually. >>>> >>>> -- DJ Lucas >>>> >>> >>> Does it work? It looks to me like something I found in Debian not so >>> long ago, but I've noticed that network manager will work without that. >>> However, you need to make sure about two things: polkit-gnome MUST be >>> running. Probably the reason why you can't edit connections, because it >>> would require polkit authentification agent to be running. Another main >>> reason is that your session needs to be registered with consolekit and >>> also marked as active (I suggest using Linux PAM there, since I don't >>> know any other way). >>> >>> $ ck-list-sessions >>> Session2: >>> unix-user = '1000' >>> realname = 'Armin K' >>> seat = 'Seat1' >>> session-type = '' >>> active = TRUE >>> x11-display = ':0' >>> x11-display-device = '/dev/tty7' >>> display-device = '' >>> remote-host-name = '' >>> is-local = TRUE >>> on-since = '2012-07-08T06:55:14.236487Z' >>> login-session-id = '2' >>> >>> My session is started via GDM, so x11-display-device is /dev/tty7. You >>> might get different output there, but you need to make sure that active >>> = TRUE and is-local = TRUE is really present, or you will be unable to >>> do any "administrative" stuff. >>> >> >> There it is. I'm not using gdm so the x attributes aren't set. active = >> FALSE. Gnome only needs to be started by gnome-session in .xinitrc, >> thanks to the PAM configuration you added (dbus-lauch >> --exit-with-session is no longer required), however, I do not see >> polkit-gnome-authentication-agent-1 in ps output. That'll give me >> something to work from, however. >> > > Argh! You have this covered in the book already, I just didn't catch it > when updating my gnome scripts to latest. I didn't get the autostart > file installed. BTW, still not fixed. I'm unable to get an active ck session. Been staring at this all day and just now realized that the login-session-id was not set. Just rebuilt kernel with CONFIG_AUDITSYSCALL=y (I have no idea why I would have even messed with auditing support. Still not fixed, but I suspect it has to do with dbus-launch now. Will report back when fixed. -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Network Manager configuration
On 07/08/2012 08:37 AM, DJ Lucas wrote: > On 07/08/2012 02:37 AM, Armin K. wrote: >> On 07/08/2012 04:53 AM, DJ Lucas wrote: >>> Anybody have an idea of how to allow a regular user (as a member of a >>> dedicated group) to modify a wired connection's configuration? With the >>> config file I've added to the BLFS Wiki (well, similar to that one, I've >>> divided it up as I had suggested), >>> http://wiki.linuxfromscratch.org/blfs/wiki/NetworkManager I'm able to >>> enable and disable the connections, but I cannot edit settings using the >>> gnome nm applet on wired connections. I'm sure I'm missing something >>> simple, but can't for the life of me find it. I've tried various changes >>> to policies with no change. Also, clicking network settings from the >>> tray does nothing...I have to open nm-connection-manager manually. >>> >>> -- DJ Lucas >>> >> >> Does it work? It looks to me like something I found in Debian not so >> long ago, but I've noticed that network manager will work without that. >> However, you need to make sure about two things: polkit-gnome MUST be >> running. Probably the reason why you can't edit connections, because it >> would require polkit authentification agent to be running. Another main >> reason is that your session needs to be registered with consolekit and >> also marked as active (I suggest using Linux PAM there, since I don't >> know any other way). >> >> $ ck-list-sessions >> Session2: >> unix-user = '1000' >> realname = 'Armin K' >> seat = 'Seat1' >> session-type = '' >> active = TRUE >> x11-display = ':0' >> x11-display-device = '/dev/tty7' >> display-device = '' >> remote-host-name = '' >> is-local = TRUE >> on-since = '2012-07-08T06:55:14.236487Z' >> login-session-id = '2' >> >> My session is started via GDM, so x11-display-device is /dev/tty7. You >> might get different output there, but you need to make sure that active >> = TRUE and is-local = TRUE is really present, or you will be unable to >> do any "administrative" stuff. >> > > There it is. I'm not using gdm so the x attributes aren't set. active = > FALSE. Gnome only needs to be started by gnome-session in .xinitrc, > thanks to the PAM configuration you added (dbus-lauch > --exit-with-session is no longer required), however, I do not see > polkit-gnome-authentication-agent-1 in ps output. That'll give me > something to work from, however. > Argh! You have this covered in the book already, I just didn't catch it when updating my gnome scripts to latest. I didn't get the autostart file installed. Thanks. -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Network Manager configuration
On 07/08/2012 02:37 AM, Armin K. wrote: > On 07/08/2012 04:53 AM, DJ Lucas wrote: >> Anybody have an idea of how to allow a regular user (as a member of a >> dedicated group) to modify a wired connection's configuration? With the >> config file I've added to the BLFS Wiki (well, similar to that one, I've >> divided it up as I had suggested), >> http://wiki.linuxfromscratch.org/blfs/wiki/NetworkManager I'm able to >> enable and disable the connections, but I cannot edit settings using the >> gnome nm applet on wired connections. I'm sure I'm missing something >> simple, but can't for the life of me find it. I've tried various changes >> to policies with no change. Also, clicking network settings from the >> tray does nothing...I have to open nm-connection-manager manually. >> >> -- DJ Lucas >> > > Does it work? It looks to me like something I found in Debian not so > long ago, but I've noticed that network manager will work without that. > However, you need to make sure about two things: polkit-gnome MUST be > running. Probably the reason why you can't edit connections, because it > would require polkit authentification agent to be running. Another main > reason is that your session needs to be registered with consolekit and > also marked as active (I suggest using Linux PAM there, since I don't > know any other way). > > $ ck-list-sessions > Session2: > unix-user = '1000' > realname = 'Armin K' > seat = 'Seat1' > session-type = '' > active = TRUE > x11-display = ':0' > x11-display-device = '/dev/tty7' > display-device = '' > remote-host-name = '' > is-local = TRUE > on-since = '2012-07-08T06:55:14.236487Z' > login-session-id = '2' > > My session is started via GDM, so x11-display-device is /dev/tty7. You > might get different output there, but you need to make sure that active > = TRUE and is-local = TRUE is really present, or you will be unable to > do any "administrative" stuff. > There it is. I'm not using gdm so the x attributes aren't set. active = FALSE. Gnome only needs to be started by gnome-session in .xinitrc, thanks to the PAM configuration you added (dbus-lauch --exit-with-session is no longer required), however, I do not see polkit-gnome-authentication-agent-1 in ps output. That'll give me something to work from, however. Thanks! -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
[blfs-support] Network Manager configuration
Anybody have an idea of how to allow a regular user (as a member of a dedicated group) to modify a wired connection's configuration? With the config file I've added to the BLFS Wiki (well, similar to that one, I've divided it up as I had suggested), http://wiki.linuxfromscratch.org/blfs/wiki/NetworkManager I'm able to enable and disable the connections, but I cannot edit settings using the gnome nm applet on wired connections. I'm sure I'm missing something simple, but can't for the life of me find it. I've tried various changes to policies with no change. Also, clicking network settings from the tray does nothing...I have to open nm-connection-manager manually. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Java (JDK, OpenJDK and IcedTea-Web) questions
On 07/03/2012 01:17 PM, alex lupu wrote:
> I understand this is a confusing subject for the layman
> (and the preamble of OpenJDK-1.7.0.5 build instructions only
> helps to emphasize this)
> so not surprisingly, I have a few questions:
>
> 1. If one chooses a "Binary Installation" of OpenJDK,
> what's the difference (technical and legal) from installing
> the Oracle's "jdk-7u5-linux-i586.tar.gz" (~79MB) directly?
>
Legal: You don't get to use the coffee cup logo.
Technical: The Oracle build is statically linked.
Moral: The plugin and javaws is closed source. Personally, I prefer not
to use closed software, but to each his own.
> 1.1. I understand that for the BLFS option one has to install
> "IcedTea-Web" as well (to get the plugin) but that's a minor
> issue IMHO (i.e., you live by BLFS, you die by BLFS :).
>
> BTW, inserting the word "both" somewhere in the sentence
> "The browser plugin and webstart implementation have been
> split off into a separate project."
> might help the neophyte (again, IMHO).
>
It is an odd case. In either seemingly appropriate location, it provides
little bennefit to the reader, and actually sounds a little strange when
read aloud. It could also be that since I wrote it...but together, they
are one package (going a little deeper, the plug-in is actually a
library and provides all of the functionality to javaws).
Short of a grammatical error, I'm inclined to disagree. I haven't found
anything on the OWL to give advice either way. That doesn't mean that
there isn't a rule someplace, I just didn't look for very long.
> I see a relative size difference between the "BLFS option" (OpenJDK)
> and Oracle (JDK). Most OpenJDK file sizes are larger. For example,
>
> 5654 jdk1.7.0_05/bin/java
> 8792 OpenJDK-1.7.0.5-bin/bin/java
>
My binary OpenJDK builds are not stripped. On x86_64, original size is
10,957, stripped it is 4904. The larger size of the stripped Oracle
binary likely owing again to static linking (different strip options are
also a possibility).
> 144272 jdk1.7.0_05/jre/lib/i386/libnpjp2.so
> 1337442 OpenJDK/jre/lib/IcedTeaPlugin.so
>
IcedTea plugin is a completely new implementation and contains most of
the functionality of javaws as well, and again, it is not stripped.
> but (as an exception to the rule),
>
> 79688 jdk1.7.0_05/jre/bin/javaws
> 1116 OpenJDK-1.7.0.5-bin/jre/bin/javaws
See above.
>
> 1.1. Has the OpenJDK software been reverse-engineered?
No, most of it is actually provided by Oracle. There are a few remaining
parts of the proprietary JDK that are not open sourced as those parts
are under license from another entity. Audio/MIDI libs, the plugin, and
web-start are a few examples (they are actually the only ones left IIRC).
> 1.2. Is it fully equivalent or does it even provide extra functionality?
>
It should be fully compatible, but you can't say that or use the coffee
cup logo. :-) Other distros use the same code and similar build method,
and do test against the TCK regularly to claim compatibility.
> 2.1. In the OpenJDK text:
> Where is "PATH_HOLD" in
>export PATH_HOLD="${PATH}"
> used?
>
You found an error. Old instruction that didn't get removed with the
latest two updates. :-)
> 2.2 Where is the in
> ln -s ${JAVA_HOME}/jre/lib//IcedTeaPlugin.so ?
>
> On my "IcedTea-Web" build (i686-pc-linux-gnu),
> the plugin shows up in
> OpenJDK-1.7.0.5-bin/jre/lib/
>
> Note: there _is_ an OpenJDK-1.7.0.5-bin/jre/lib/i386/
> subdirectory though, but with some other stuff in it.
>
And another bug. It appears to have changed, but the book was not
corrected with update. I'll tend to both errors shortly.
Good questions and observations. Thanks!
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
[blfs-support] SVG conversion program? (Was: CUPS problem: lp: Unsupported document-format "text/plain".)
On 06/25/2012 11:19 PM, William Tracy wrote: > > On Jun 25, 2012 8:28 PM, "DJ Lucas" <mailto:d...@linuxfromscratch.org>> wrote: > > Way OT: Coincidentally, I never bothered to ask here, but what is the > > best way to convert from BMP to SVG? > > I would be surprised if there were an automated way to do that--bmp is a > bitmap format (bmp is actually an abbreviation for bitmap) and svg is a > vector format (that's what the v stands for). > Wow, give me a little credit...I'm not a graphics guy by any means, but I'm not quite that lost in the woods! ;-) But bummer. I find it difficult to believe that there is not already a tool for that. My simpleton logic would suggest a way to find contiguous regions in any given bitmap image (or other decompressed image) and assign recognizable shapes or straight lines (similar to a line art drawing maybe). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] CUPS problem: lp: Unsupported document-format "text/plain".
On 06/25/2012 05:12 PM, Ken Moffat wrote: > I *assume* what is > currently in the book works for *someone* ? Perhaps someone with a > real postscript printer ? Ya know, the laughing fit I had prior to posting this message was probably not warranted, just one of those silly things that kinda caught me off guard. I'm pretty sure that I haven't printed a single page on my inkjet since back in November of 2010 when I did the designs for the new singlets. I wonder how thick the dried ink layer is on those cartridges now? Way OT: Coincidentally, I never bothered to ask here, but what is the best way to convert from BMP to SVG? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Thoughts on Xorg-7.7+
On 06/25/2012 03:08 PM, alex lupu wrote: > Hi, > > First and foremost: Many thanks and a great debt of gratitude > for the huge effort and outstanding quality of this BLFS release > (Ch. 24 "Introduction to Xorg-7.7-1"). > I suspect DJ has (again) been instrumental here. > > -- > > The following couple of thoughts are mostly directed to > existing 7.6 and future 7.7 BLFS users. > > 1. > While stable on 7.6, early in the game I subscribed to > the Xorg Archives list, > > Among many more or less interesting posts, you receive "ANNOUNCE" > announcements issued by various Xorg developers there. > As an example: > > << > [ANNOUNCE] xcb-util 0.3.9 > > Arnaud Fontaine ... > Wed May 30 20:57:02 PDT 2012 > ... > Download > > http://xcb.freedesktop.org/dist/xcb-util-0.3.9.tar.bz2 > ... > sha256: c611259c0ab ... e8b3636315c18c4 > ... >>> > > For some reason (mostly ADD I suppose), I've made a habit to follow each > and every [ANNOUNCE] and act on it. Whatever they threw at me, > I downloaded, built and installed (successfully I might add; the proper > dependencies were those provided by the 7.6 related software - makes sense). > > So, all of a sudden, in checking with the book as well, I realized this way > one > can end up perched on the 7.7 "release", without much fuss (and/or muss) - > or as the Germans say, sans vraiment se fatiguer. > > As an aside, I even noticed that with at least two packages > I (accidentally) exceeded the book level: > > xkeyboard-config-2.6 > xcb-util-0.3.9 Yes, I'm currently working on a little different presentation of the Xorg packages for the book, which will also allow us to update packages individually. For the -1 release version, I wanted specifically to match katamari. I'd spend a bit more time describing that, but really, it won't matter any more (and really hasn't for the past few years) after the packages are split up (as BLFS will no longer tack on a release version). Hopefully that means that updates will happen much quicker in the future...I've just got to get it to a point where it looks presentable. And don't worry, I'll leave the build as a group functionality for those of us not packaging builds. As far as the updates you've mentioned, I haven't reviewed the change logs yet, but I would suspect that they should be OK. Pretty much as long as the major version doesn't change, it's _probably_ OK to just blindly install it with Xorg. Minor version changes should still be reviewed prior to blindly installing, however. While they are usually OK, there have been instances of software not working with each other (libX11, libxcb, libdrm, Mesa, drivers, and of course the server are packages that you should be especially cautious with regard to minor version changes). Same thing for protocol headers, though they rarely changes much, they should match their counterparts in the both the libraries and running server. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] World's worst seder
On 06/23/2012 11:59 AM, DJ Lucas wrote:
> On 06/23/2012 11:27 AM, DJ Lucas wrote:
>> what is the correct syntax
>> for [a-z0-9]{32}?
>>
>> -- DJ Lucas
>>
>
> Never mind...had to escape the braces. Ultimately, the substitution I
> was looking for was this:
>
> '/^[a-z0-9]\{32\} [a-z]/s@^[a-z0-9]\{32\}@\&-md5sum;@'
>
>
It's not fair to ask a silly question without revealing the goal. In an
effort to make this thread not completely useless (and just for S&Gs),
the whole process for the md5sums file looked like this...maybe it'll be
useful for somebody (or not):
Open the new xml file in vim and use :r "x7-app-7.7-1.md5" to read the
md5sums file into the xml (in a "cat >
applications.md5 << "EOF"\nEOF" block) and
save file, then perform the first substitution command:
sed '/^[a-z0-9]\{32\} [a-z]/s@^[a-z0-9]\{32\}@\&-md5sum;@' \
-i applications.xml
Perform the second substitution command:
sed '/^\&-md5sum/s@[0-9]\.[0-9]\.[0-9]\.tar@\&-version\.tar@' \
-i applications.xml
And finally, fix the entities with a third substitution:
for listitem in `grep "^\&-md5sum;" applications.xml | sed -e
's@\&-md5sum; @@' -e 's@\&-version;\.tar\.bz2@@'`
do
sed -e "/${listitem}/s@\&-md5sum@\&${listitem}md5sum@" \
-e "/${listitem}/s@\&-version@\&${listitem}version@" \
-i applications.xml
done
All of that to avoid the monotonous, not to mention error-prone, work of
copy and paste! It'll all be reused to add the entities on the next one
and package instructions for all of them to create the template for
adding the descriptions and forward dependency information.
-- DJ
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] World's worst seder
On 06/23/2012 11:27 AM, DJ Lucas wrote:
> what is the correct syntax
> for [a-z0-9]{32}?
>
> -- DJ Lucas
>
Never mind...had to escape the braces. Ultimately, the substitution I
was looking for was this:
'/^[a-z0-9]\{32\} [a-z]/s@^[a-z0-9]\{32\}@\&-md5sum;@'
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
[blfs-support] World's worst seder
Well I thought I was well versed, but I needed to import the xorg
md5sums into the book. After futzing with it for a few minutes, instead
of asking for help on this pattern:
/^[a-z0-9]{32} [a-z]/
I got ticked off and used a ridiculously long command with 32 dots
(twice) just to get it done. :-D
So anyway, now that I don't actually need it, what is the correct syntax
for [a-z0-9]{32}? Is there a better way to identify an md5sum?
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] 回复： libwww-perl-6.04
On 06/20/2012 01:53 AM, Omar wrote: > >> The following appears when the command "perl Makefile.PL" is carried > out. > >> Can't connect: IO::Socket::INET: Bad hostname 'lfs:41344' > >> Test server timeout > >> The can-we-talk-to-ourself test failed. > >> Warning: prerequisite Encode::Locale 0 not found. > >> Warning: prerequisite File::Listing 6 not found. > >> Warning: prerequisite HTTP::Cookies 6 not found. > >> Warning: prerequisite HTTP::Daemon 6 not found. > >> Warning: prerequisite HTTP::Date 6 not found. > >> Warning: prerequisite HTTP::Negotiate 6 not found. > >> Warning: prerequisite HTTP::Request 6 not found. > >> Warning: prerequisite HTTP::Request::Common 6 not found. > >> Warning: prerequisite HTTP::Response 6 not found. > >> Warning: prerequisite HTTP::Status 6 not found. > >> Warning: prerequisite LWP::MediaTypes 6 not found. > >> Warning: prerequisite Net::HTTP 6 not found. > >> Warning: prerequisite WWW::RobotRules 6 not found. > >> Writing Makefile for LWP > >> Writing MYMETA.yml > >> > >> I don't know what to fix. I haven't set up DNS or HTTP which may be > >> possible reason I'm not sure. Anyone give me any suggestions. Thanks in > >> advance. > >> Omar > >> > > > >Fix (or add) the entry in /etc/hosts for lfs. > > > >-- DJ Lucas > > This is the content of /etc/hosts which is completed in the LFS 7.1. How > to fix it? Is it with some errors? > > # Begin /etc/hosts (network card version) > 127.0.0.1 localhost > 192.168.0.100 lfs.example.org lfs1 > # End /etc/hosts (network card version) > > Omar > Yes, what you have above is incorrect (there is no 'lfs' entry in the file). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] libwww-perl-6.04
On 06/19/2012 10:07 PM, Omar wrote: > Hi, all, > > The following appears when the command "perl Makefile.PL" is carried out. > Can't connect: IO::Socket::INET: Bad hostname 'lfs:41344' > Test server timeout > The can-we-talk-to-ourself test failed. > Warning: prerequisite Encode::Locale 0 not found. > Warning: prerequisite File::Listing 6 not found. > Warning: prerequisite HTTP::Cookies 6 not found. > Warning: prerequisite HTTP::Daemon 6 not found. > Warning: prerequisite HTTP::Date 6 not found. > Warning: prerequisite HTTP::Negotiate 6 not found. > Warning: prerequisite HTTP::Request 6 not found. > Warning: prerequisite HTTP::Request::Common 6 not found. > Warning: prerequisite HTTP::Response 6 not found. > Warning: prerequisite HTTP::Status 6 not found. > Warning: prerequisite LWP::MediaTypes 6 not found. > Warning: prerequisite Net::HTTP 6 not found. > Warning: prerequisite WWW::RobotRules 6 not found. > Writing Makefile for LWP > Writing MYMETA.yml > > I don't know what to fix. I haven't set up DNS or HTTP which may be > possible reason I'm not sure. Anyone give me any suggestions. Thanks in > advance. > Omar > Fix (or add) the entry in /etc/hosts for lfs. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] xorg-7.7-1 circular dependency
On 06/16/2012 01:10 PM, John Burrell wrote: > > You don't appear to give any suggestion in the book. I haven't tried to > upgrade yet but > it would be good to get some guidance from the book, since I assume you > have done this > successfully. > > Thanks > > jb. Yeah, I copied the xbitmaps page from another when I separated it and left that in there by mistake. It does not require the apps to install it. I'll get that fixed momentarily. -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] glibc-2.15 make failure
On 06/15/2012 09:32 PM, Bruce Dubbs wrote: > alex lupu wrote: >> I'd like to offer a very strong (but polite) dissenting voice about the >> subject of NO-NO updating/upgrading "in place" which came up here once again. > >> IRONICALLY, the one and only problem that could possibly be ascribed to >> working "in place" is _exactly_ (and coincidentally) the one described >> in the (original) subject of this thread. > > I don't know where anyone said that you shouldn't update packages in > place. Except for glibc, I've been doing it when required for years. > > As for updating gcc, the better instructions for gcc are in BLFS, not LFS. > > BTW, this entire thread should be in LFS support, not BLFS support. > > -- Bruce I've done many glibc updates in place in the not so recent past. As a matter of course, I would drop to runlevel 1 before make install, but have done it in the past as an example within a gnome-terminal in runlevel 5 without issue (search archives). I do immediately reboot afterward. Going way back now...I used to keep a statically linked copy of what is now know as coreutils laying around just in case (I can't even remember what the 3 packages were that combined to create coreutils now). At one point, there was a problem with the glibc install target using the install command, but I'm pretty sure that problem went away before linuxthreads did. That was a long time ago wasn't it? At any rate, I too build a new system when it is time, but tend to keep them around way longer than they need to be...I get attached and too lazy (not to mention lack of time), though I am about ready to give up this dinosaur (still running gnome-2.32). -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] X unsolved and networking solved
On 05/01/2012 11:51 PM, Walter Webb wrote: > The first problem I encountered was that there was > no acceleration in the X server with LFS-7.1. The problem appears > to be with something installed before X. I am still working on > that. What video card? If newer ATI, Intel, or NVidia device, you'll need to use KMS now days. You'll also need to build in the firmware for newer ATI devices. Unfortunately, I don't have a firm understanding of "newer" yet. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
[blfs-support] OT: Windows for work (Was: Re: DHCP-4.2.3-P2 Client Configuration)
On 04/01/2012 04:55 PM, Bruce Dubbs wrote: > Ken Moffat wrote: > >> I was going to reply earlier, then I remembered that I do have one >> windows machine (my netbook - it came with some form of windows7, >> but is totally unusable with it (insufficient memory), so I've not >> been able to update my satnav [ the download is win only ]). It >> also made even long-gone redhat installers look state-of-the-art in >> comarison to the win installer :) > > Going a bit off topic (e.g. DHCP), let me make a comment. When I was > teaching, I sat through a course on Windows for comparison. This was > around 2002. One of our tasks was to install it on a bare drive. When > people complain about installing Linux, it only means that they never > installed Windows, but bought it pre loaded. It was a PITA. When > complete, I looked at what came with it. My mind boggled -- I asked "Is > that all there is?" The answer was yes. Anything useful is an add on > at extra cost. > > It makes you wonder if the world wide water system has been contaminated. > IDK. I use whatever is required for the job at hand. My laptop does dual boot, and I do have cross-over office installed so that I can use Outlook in Linux (which was really a waste as I haven't used it, I use my phone for just about everything now). A clean Windows 7 installation is not that bad (maybe 6 or 7 prompts now days), nor is Windows 8 (but 8 will likely be a flop along the lines of Windows ME). Now, that's not to negate "extra cost" because my time is not free and there are likely drivers to add, and additional software packages to install, user migration, etc. Now imaging does come into play for businesses to offset some of those costs, but from a management perspective, Windows is quite a bit easier for both admins and end users. I want something done, say a new network printer installed, I set a policy using a simple and fairly intuitive point and click interface, nothing need be done by the end user, and it is done on *all* machines that I request it be done on. I need info on a machine, I personally clob together a simple WMI query using vbscript, but my counterparts usually employ something point and clicky from the myriad of tools available to automate this task. As far as my experience on Linux, nothing simple like the two examples above exist. Am I capable of writing an LDAP query to push something to my machines over ssh/scp, sure. Login scripts, certainly. I can ssh into any machine and get whatever I need, but again, well over half of my counterparts at my current place of work are incapable of these tasks. While the additional features certainly make Windows less secure (ports open for WMI, additional code that can potentially harbor security vulnerabilities, etc.), there is an old adage that still holds true: Speed, reliability, low cost, pick two. In just about any corporation, you can expect low cost to be one of the two selected variables, and good unix admins don't come cheap. Until we are commoditized (not a real word) by mindless point and click interfaces and kiddos fresh out of tech school with a Linux ataboy certificate that says nothing more than they know how to pass a test, Windows will likely be around. I find it best to be able to work with it. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Xorg
On 02/12/2012 08:57 AM, Fernando de Oliveira wrote: > Hi, > > I have been able to run Xorg in LFS SVN-20111210, using /etc/X11/xorg.conf.d > instead of /etc/X11/xorg.conf. > > For the first time the server successfully run in my first attempt, at about > Feb 10 22:50 BRT!!! > > I had delayed it for some time, until I decided to create the xorg.conf.d > files, as did not know how to name them and how to adapt from xorg.conf. One > line in "Xorg-7.6-2 Testing and Configuration" with comparison of deprecated > xorg.conf and how to make the adaptation could help., or may be I have missed > it. > > []s, > Fernando Few actually need xorg.conf.d files now days. The only things that make sense now days are keyboard configurations for non-US English keyboard layouts, and driver configurations for proprietary video drivers or multi-monitor setups where hardware does not match logical layout, and those only when the DE does not have tools to control the keyboard or video layout, or if you use graphical login tools (xdm, kdm, gdm, etc.). There are examples for two of the three at the bottom of the configuration page here: http://www.linuxfromscratch.org/blfs/view/svn/x/xorg-config.html I would like to add one for nVidia cards as well, but don't have the hardware currently. For the most part, the /etc/X11/xorg.conf.d directory should be empty save for the default 10-evdev.conf and 50-synaptics.conf files that ship with those drivers. Most things just work out of the box. Another thing that was common in the past was to set default resolution in the xorg.conf. The Xorg log file contains the available modelines, but likely all you will need is an xrandr statement in .xinitrc, and again, only if your choice of desktop does not include a tool to control it or you use graphical login. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] make shadow fails after installation of Linux-PAM
On 02/04/2012 07:43 AM, Ronnie van Aarle wrote:
> $ if [ -L /lib/libpam.so ]; then
> for LINK in libpam{,c,_misc}.so; do
> ln -v -sf ../../lib/$(readlink /lib/${LINK}) /usr/lib/${LINK} &&
> rm -v /lib/${LINK}
> done
> fi
>
> I do have my doubts on this last if-statement for-loop, because I don't
> realy think it does what it is supposed to do. In my situation, I am
> building shadow from
>
> /sources/security/shadow-4.1.3/
>
> So ../../lib does not exist. I suppose that I can just drop the ../../
> because I think the /lib/libpam{,c,_misc}.so are the files that are
> required. While I'm writing this email I just decided that that is also
> the next thing I want to try, however, I would realy appreciate some
> confirmation from someone who has more skills and routine in bash than I do.
Glad you got it fixed, but I thought I'd point out the misunderstanding
above as well. When using relative links, they are relative to the link
target, not the current directory (a very common misconception). IOW:
/usr/lib/../../lib, not /sources/lib. The difference between ../../lib
and /lib in the link becomes very important if you were to mount that
partition elsewhere.
HTH
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] libxcb 1.0 bug with dependencies xcb-proto
On 02/01/2012 04:05 AM, Simon Geard wrote: > On Wed, 2012-02-01 at 09:26 +, Firerat wrote: >> On 1 February 2012 08:54, Dmitry Blum wrote: >>> X86 >>> LFS 6.8 >>> BLFS 6.3 >>> X`s prefix - /usr >> /usr? >> or /usr/X11 ? >> >> you are better off having X in its own subdir and not 'mixed in' with >> the rest of /usr > > Why do you say that? Unless it's part of your package management > strategy, putting everything in /usr is the least painful approach, > avoiding the usual problems with $PATH, $PKG_CONFIG_PATH, etc. > > Simon. > Yes, using alternate paths is a PITA. While X isn't really horrible, I still do it for the book on my dev VMs, but on my daily system I use /usr. Perhaps I should make that more clear in the book? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] krb5 client question(s)
On 01/13/2012 08:25 PM, luxInteg wrote: > > I have looked at the blfs-bootsctipts and I do not see any for krb5-client. > So I would be grateful to know:-. > --Does the nfs4-server and client both run kdc's? > --should krb5.login be running as a daemon ? > If you really want to separate them, that is fine. There is no reason that you should have to put authentication services on the storage server itself, saving for the obvious case of the authentication server being down rendering all boxes useless without having real local users. There is no "client" as you put it, but kerberized tools. In your case you are probably looking at PAM modules (PAM_KRB5) to handle authentication, and NFS/NIS I really can't help with any longer because it's been so long since I messed with them. My assumption is that NSS is still handled internally for NIS (is there any NSS_NIS?) but I recall nothing about how the mappings actually work in that case. My preferred testing ground now days is Kerberos for authentication, LDAP for address book and UID/GID mappings, and Samba for network storage. My setup of course requires NSS_LDAP and PAM_KRB5. There are plenty of tools and tutorials available to get it working correctly in a matter of hours (days? weeks? ;-) ). All of the above I am a little rusty on however...so little play time any more. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] Problems with a (partial) gnome3 install
On 12/07/2011 03:22 PM, Ken Moffat wrote: > > 1. 'help' no longer works in applications. I had this in the past, > and the solution was to add gvfs. Searching google only finds that > recommendation. But, I've *got* gvfs (1.10.1). Looking at the > build log for an older version on my previous system, GConf support > has been entirely removed [ I guess this used to be the important > part with gnome2/GConf2 ] and FTP/HTTP/WebDAV support is now > enabled. > > Curiously, if I open yelp directly, I can get to all of the > installed application help, but 'Help -> Contents' gives me > Could not display help: Operation not supported It's been a couple of years, but something similar has happened before in either Gnome 2.28 or 2.30 (seems to me it was this exact problem, but I'm not absolutely positive on this). It turned out to be a change in xulrunner or webkit (I forget which, but that was in addition to missing gvfs and gnome-doc-utils). I looked through my old logs, but I only have 2.30.2 now for yelp, with some random sub-set of a 2.32 desktop environment. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] BLFS-2011-12-08, MesaLib
On 12/09/2011 02:53 PM, Dr.-Ing. Edgar Alwers wrote: > btw. what is the matter with MesaGLUT ? MesaGLUT is not free software. The Meaa developers have permission to distribute it, free of *charge*, in source form. We elected to remove it and use FreeGLUT. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] nouveau to radeon transition help needed
On 12/06/2011 09:58 AM, Ken Moffat wrote: > On Tue, Dec 06, 2011 at 07:35:09AM +, luxInteg wrote: >> >> I still have resolutions limited to 1024 x768 and yes I have monitors capable >> of higher resulitions ) and I have keyboard errors. For instance for >> keyboard tweaking I tried to put stuff in /etc/X11/xog.conf.d using this >> guide >> ( http://www.gentoo.org/doc/en/xorg-config.xml ) for example >> >>#--- /etc/X11/xorg.conf.d/30-keyboard.conf >> >> Section "InputClass" >> Identifier "keyboard-all" >> Driver "evdev" >> Option "XkbLayout" "gb" >> Option "XkbModel" "pc105" >> Option "XkbRules" "evdev" >> Option "XkbOptions" >> "grp:alt_shift_toggle,grp:switch,grp_led:scroll,compose:rwin,terminate:ctrl_alt_bksp" >> Option "XkbVariant" ",qwerty" >> MatchIsKeyboard "on" >> EndSection >> >> but it is not digested when I run X. >> >> Any ideas? > > My current install (in /usr - ISTR you have used different prefixes > in the past - with the 1.11.2 server) uses /usr/share/X11/xorg.conf.d > although the (now old) wiki documentation I've read suggested /etc. > > If you have a 10-evdev.conf installed by the server, use the > directory that is in. > > ĸen You need the symlink for it to make it work in /etc with latest server. I'm not sure why or when that changed though. I'm pretty sure it worked without in 1.9.x. Curiousity, does 'xrandr -s 0' in an xterm have any bad or good effect? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] remove-expired-certs.sh
On 12/04/2011 12:32 PM, Bruce Dubbs wrote:
> To make a fix, we could do something like:
>
> certdate=$( date -d "${date}" +%Y%m%d 2>/dev/null )
> if [ $? != 0 ]; then
> # skip it with continue
> # or remove it with rm -f "${cert}"
> fi
Yes, assuming that this is the 2038 bug, a continue here would be much
better, not more complex logic elsewhere. Thank you for catching that Bruce.
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] remove-expired-certs.sh
On 12/04/2011 10:04 AM, Fernando de Oliveira wrote: > Hi, > > Running this script I obtain (LFS 6.5, 6.7, small difference in 6.8): > > $ remove-expired-certs.sh > date: data inválida "Jul 31 12:31:40 2038 GMT" > /bin/remove-expired-certs.sh: line 18: [: -lt: esperado operador unário > certs/ad6c2ff9.pem is expired! Removing... > certs/84009bc3.pem is expired! Removing... > date: data inválida "Jul 31 12:29:50 2038 GMT" > /bin/remove-expired-certs.sh: line 18: [: -lt: esperado operador unário > date: data inválida "Dec 31 14:20:24 2040 GMT" > /bin/remove-expired-certs.sh: line 18: [: -lt: esperado operador unário > certs/48ef30f1.pem is expired! Removing... > date: data inválida "Dec 31 14:10:36 2040 GMT" > /bin/remove-expired-certs.sh: line 18: [: -lt: esperado operador unário > certs/972672fc.pem is expired! Removing... > > > where a free translation of the messages could be: > > data inválida = invalid date > esperado operador unário = expecting unary operator > > > []s, > Fernando The error "esperado operador unário" means that the data contained in the first argument to test either has non-numerical characters, or that it is empty. I'm guessing you are running a 32bit kernel and libc and the 2038 date bug is rearing its head. I suppose we need to dump the test output to /dev/null and put the removal into a second if case based on the return value of test to work around this. 2 is the return value needed to test on. Thanks for the report. I'll get to it in a day or two (I don't have a recent 32bit system to test with right this second). OTOH, it is a good learning experience. We'll undoubtedly see more of this in the coming years (for as long as we support 32bit hosts). If you want to try and fix it yourself, feel free to do so and post the corrected script and we'll get it tested and updated in the book. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-support] bind and intranet subnets question
On 11/20/2011 07:32 AM, luxInteg wrote:
> Greetings
>
> Suppose I have two private IPv4-assigned networks with addresses
> --network 1:- 10.255.255.y
> --network 2:- 172.31.255.x
>
> can I configure bind (as per
> http://www.linuxfromscratch.org/blfs/view/svn/server/bind.html )
> to do names resolutions for hosts in both subnets
> and if so how so (i.e. would it be
> a) a box with a single NIC ( and if so where would this be placed)? or
> b) a box with two NIC's connected to two ethernet switches etc etc)
>
> advice would be much appreciated
>
> sincerely
> lux-integ
I'm not sure I understand your question. You can create as many zone
files as you need. If you followed BLFS instructions, you run BIND in a
chroot jail, and these paths are good. Simply add another zone entry to
/srv/named/etc/named.conf, and a correct zone file. If you meant as far
as how to configure bind...nothing to do if the machines have a route to
the host, just make sure 53 UDP is allowed.
Configuration is similar to the following:
= /srv/named/etc/named.conf snipit
zone "litstest.com" {
type master;
file "fz/litstest.com.zone";
};
zone "152.168.192.in-addr.arpa" {
type master;
file "pz/192.168.152.zone";
};
===
== /srv/named/etc/namedb/fz/litstest.com.zone =
[root@postal ~]# cat /srv/named/etc/namedb/fz/litstest.com.zone
; BIND db file for litstest.com
$TTL 86400
@ IN SOA ns1.litstest.com. hostmaster.litstest.com. (
2008080601 ; serial number YYMMDDNN
28800 ; Refresh
7200; Retry
864000 ; Expire
86400 ; Min TTL
)
NS ns1.litstest.com.
MX 10 maiamailguard.litstest.com.
$ORIGIN litstest.com.
ns1 IN A 192.168.152.202
litstest.com. IN A 192.168.152.202
exchangeIN A 192.168.152.202
maiamailguard IN A 192.168.152.201
www IN CNAME exchange
==
= /srv/named/etc/namedb/pz/192.168.152.zone ==
$TTL 3D
@ IN SOA ns1.litstest.com. hostmaster.litstest.com. (
1 ; Serial
8H ; Refresh
2H ; Retry
4W ; Expire
1D) ; Minimum TTL
NS ns1.litstest.com.
201 PTR maiamailguard.litstest.com.
202 PTR exchange.litstest.com.
==
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: [blfs-support] Request
On 11/19/2011 06:59 PM, Bruce Dubbs wrote: > > For my client (seamonkey), the Reply-to: is visible. Return-Path: is > not. Besides the return path is > which is slightly confusing. Yeah, you have to add it. Unfortunately, I've been using that and carrying over for so long, I really don't recall why. Hell, it might even date back to belgarath, or maybe right at the switch to quantum. :-) -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Request
On 11/19/2011 02:51 PM, Bruce Dubbs wrote: > Fernando de Oliveira wrote: >> I do not know to which list should I address this. >> >> I am having a little trouble with the messages, now that I am trying >> to follow more than one *lfs* list. Worse, when the same message is a >> notification simultaneously sent to more than one list. >> >> I would like to suggest, to the maintainers (Bruce?, Ken?), >> including, if possible and not very difficult, an automatic >> identification in the subject, such as in this message >> [blfs-support], or [BLFS Support], when distributing it to the list. > > I'll take a look and see what I can do, but I've had filters for years > on the Reply-To: address. That is: > > BLFS Support List > LFS Support List > > etc. > > -- Bruce I've always filtered on "return-path" for LFS list emails. I do not now recall the reason for that over "reply-to". -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: firefox-8.0.source
On 11/09/2011 06:07 PM, Jeremy Henty wrote: > > Bruce Dubbs wrote: >> Andrew Benton wrote: >>> >>> Yes, with wget I downloaded >>> ftp://ftp.mozilla.org/pub/firefox/releases/8.0/source/firefox-8.0.source.tar.bz2 >>> It compiled on a 3.1 kernel without patching. >> >> I'd appreciate knowing the instructions you used. I can't do it yet >> because I haven't put X on my development system yet. > > I'm not the person you replied to, but here are my notes on > Firefox-[78]* anyway... > > You must configure cairo with --enable-tee . > > If you haven't installed yasm then you must disable webm video. > > Jeremy Guys, just in case you are not aware, we should probably consider using external npapi. We can of course, fix anything that is using the incorrect version scheme, but looking to the future, that might be a bit of work (OpenJDK specifically will be a PITA, don't know about other open source plug-ins). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: blfs setup with dns and kdc??
On 11/02/2011 11:11 AM, luxInteg wrote: > On Saturday 22 October 2011 17:58:56 DJ Lucas wrote: >>> Thanks for the randon thoughts. AND more such would be most appreciated. >>> >>> sincerely >>> luxInteg >> >> # Begin /etc/pam.d/system-auth >> >> auth sufficient pam_ldap.so >> auth requiredpam_unix.so use_first_pass >> >> # End /etc/pam.d/system-auth >> >> In the example above, pam_ldap was checked first, it provided the >> passing vote so to speak, in this case the success auth token is passed >> to the required pam_unix module. The use_first_pass directive to >> pam_unix allows it to use the information obtained from the previous >> module in its own evaluation of the requirement. A simple example, but >> one that works for system I was speaking about last night. If pam_ldap >> fails to auth, then using sufficient allows pam_unix to ignore the >> previous one and fall back to /etc files (shadow). > > I have to iterations of openldap-2.4.23 (both on 64bit amd-based machines) > > --A- built with the configure script not mentioning bdb; but the > openldap-2.4.23 sources compiled and generated a test file (with make > test) of some 40k lines (2.1 Mbytes in size) > > --B- compiled with berkeley-db configured either as module or 'yes' and > generated a (post-compiled) test file of ~100kbytes. 'make test' fails > consistently on test-29 no matter if LDFLAGS is set or not set. > Any difference in the version of Berkeley DB? Is there an included DB with OpenLDAP? What is the failure output? I don't have the time right this second to build it. > I have pam_ldap and nss_ldap as per the cblfs site. > ( http://cblfs.cross-lfs.org/index.php/Pam_ldap ) installed > > > As regards login > > --A allows login whatever the status of openldap > --B only allows login if slapd is running it seems to ignore > /etc/pam.d/system-auth settings? > > > any suggestion > Unfortunately, it sounds like either a logic error in your PAM configuration, or incorrect login binary. What does /etc/pam.d/login contain on the failing system vs. the working system? Is it just the default that is included in the book? Also, which login utility is being used? There are three to choose from, you should be using the one from Shadow after it was recompiled with PAM support. Any chance it was overwritten by one of the other(s)? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: blfs setup with dns and kdc??
On 11/04/2011 06:58 AM, luxInteg wrote: > Is /etc/default/slapd important as described in the excerpts ? And if so how > what are the contents/properties of this file? > > Advice would be much appreciated. > > sincerely > > lux-integ This would equate to /etc/openldap/slapd.conf in a default installation. As best I can gather from the above, I would guess that Debian uses multiple include files to keep the configuration broken down into logical parts. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: blfs setup with dns and kdc??
On 10/22/2011 04:44 AM, luxInteg wrote: > On Saturday 22 October 2011 08:09:15 DJ Lucas wrote: >> then you'll likely want to >> use Heimdal over MIT for the cracklib integration to enforce strong >> passwords when using kpasswd. > I tried Heimdal a few years ago buta program called StrongSwan I was > compiling (afterwards) for a router would not compile. I had to switch to > MIT-krb5. > > Incidently do you or anyone else on list know of how to set up network logins > (i.e. kerberized) with SmartCards? > > I have used smart cards, transferred keys etc but only for remote access not > local login. What I would like to set up is a regime where local (i.e. non- > kerberized) login on a workstation is available without smart card presence > and kerberized (or network) login with the smart card. Is this possible? > > Thanks for the randon thoughts. AND more such would be most appreciated. > > sincerely > luxInteg > > # Begin /etc/pam.d/system-auth auth sufficient pam_ldap.so auth requiredpam_unix.so use_first_pass # End /etc/pam.d/system-auth In the example above, pam_ldap was checked first, it provided the passing vote so to speak, in this case the success auth token is passed to the required pam_unix module. The use_first_pass directive to pam_unix allows it to use the information obtained from the previous module in its own evaluation of the requirement. A simple example, but one that works for system I was speaking about last night. If pam_ldap fails to auth, then using sufficient allows pam_unix to ignore the previous one and fall back to /etc files (shadow). I hope I explained that well enough, though I'm sure the terminology is incorrect. It is spelled out a lot better in the link provided in the book for the System Administrator's guide (which unfortunately was on kernel.org). Here is a copy of it: http://debian.securedservers.com/kernel/pub/linux/libs/pam/Linux-PAM-html/Linux-PAM_SAG.html . While I'm not familiar with SmartCard login, if there is a method to get that info into the system via nss and auth via PAM (or any other nss combination you can dream up be it in LDAP or otherwise), then you would use a similar technique to get the PAM info from the smartcard module to pam_unix and thus meet the required auth. The above reads horribly and my terminology is probably wrong...I'm sorry about that, but I have to leave again and am out of time for today. I hope that gets you going in the right direction. The administrators guide is really what you need to fully understand it. I'll try and look it up myself when I have a little more time to spare and give you a more technically correct answer with proper terms and maybe some additional suggestions. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: blfs setup with dns and kdc??
On 10/21/2011 07:20 PM, luxInteg wrote: > Greetings, > > I want to set up a kerberos (KDC) on a blfs-based machine. It seems > wasteful to dedicate a whole PC to ticket granting so I am wondering if I can > hitch a ride on a dns server inside an intranet. I would thus be grateful to > know from anyone on list experienced with kerberos and bind what are the > security implications of running dns on a kdc. > I missed that in my response following Randy's comments. From your comment above, I gather this is your first Kerberos installation. If not, please disregard this reply. If so, here are a couple of random thoughts from my own experiences learned along the way. Take the time to write out and know your goals ahead of time. I've been down this path as far as a full SBS replacement in a test environment. If you are unfamiliar, SBS is Microsoft's all in one server that breaks all of their own rules. Minus a real Exchange server (though I hear OpenChange is coming along nicely), the closest equivalent I came up with was Postfix for SMTP, Dovecot for IMAP, Samba, CUPS, DHCP, DNS (I didn't ever get dynamic registration working though), Heimdal, OpenLDAP, Funambol + Apache Java HTTP Server (I forget what it's called now) for contacts and calendar sync, SquirelMail for webmail on Apache httpd, PHP-LDAP Admin for management (also on httpd), and PPP with PPTP patches...all on one box. While I'd never put something like this in a live environment, I learned a lot doing it, which led me to my more responsible incarnations across multiple VMs. If you intend to use this box for anything more than kerberized tools on every connecting host, use LinuxPAM with the LDAP and Kerberos modules. Think of a Windows user trying to use putty to change her password as opposed to the Ctrl+Alt+Delete > Change Password, or even a web mail user doing the same as opposed to a web-based admin tool. It *is* a royal PITA to setup initially, but worth it for your end users in a complex setup. My re-write of the PAM configuration files in BLFS a few years ago was based largely on my experience with that experiment mentioned above, so it is now at least a little easier to get the PAM configuration correct. If you elect to skip PAM for whatever reason, then you'll likely want to use Heimdal over MIT for the cracklib integration to enforce strong passwords when using kpasswd. Otherwise, if not using PAM, choosing between the two is just a question of licensing AFAIK...both work (and I do not recall any major performance difference in testing, though limited). MIT was the original FWIW and is the base used by Microsoft's Active Directory (though there have been many, many changes...fun story there) if that adds any weight to the decision one way or the other. I too, like Randy, use Heimdal, but not for any technical reason...only due to some form of undue partiality in that I used it first as I was working from Randy's hint years ago when I first played with Kerberos. Both licenses work fine for me. When putting in the kerberos TXT record and kerberos and ldap SRV records in DNS, make sure the response is a host that has an A and/or record, for the same reasons you do this with NS and MX records (those are actually bound by RFC, not sure about ldap or kerberos), but there is absolutely no sense increasing your DNS traffic for those look ups by 50%. If Samba is a goal (sorry, I know this is a Linux list, but I hold multi-platform interoperability in high regard) then take a look the SMB LDAP Tools and their LDAP schema suggestions as per their how-to. It accounts for both *nix UIDs and GIDs and Windows GUIDs. You can also easily extend the schema if you need to for any additional attributes you might need. Also, do not use domain.local, use internal.domain.tld if you choose to separate internal and external domains now days. OT rant: Same holds true for any Windows admins that might be reading and managed to call Ubuntu or OSX a POS because it didn't play nicely with your incorrectly named DNS domain! You cannot correctly substitute anything you want for a TLD! As for specifically the case of .local, a little history lesson: Multicast DNS (mDNS) is an approved IETF draft, and was in draft well before Windows 2000 arrived, though it wasn't approved until last year due to language deficiencies used in the latter part of the first 27 revisions or so. Microsoft and Apple both contributed to the first few revisions and Apple carried the ball from there. Anyway, just some long-winded random thoughts. Hopefully some of my jabbering above is useful to you, or anybody else who stumbles across it some day. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: blfs setup with dns and kdc??
On 10/21/2011 09:17 PM, Randy McMurchy wrote: > On 10/21/2011 7:20 PM, luxInteg wrote: >> I would thus be grateful to >> know from anyone on list experienced with kerberos and bind what are the >> security implications of running dns on a kdc. > > I have done exactly what you are looking to do. I feel that as long > as you set up Kerberos properly (I've always used Heimdal), there > should be no problem creating a DNS server on the same machine. > > Bruce is a good authority on Bind and DNS (as is DJ), and either of > them could provide additional helpful information. But from the > Kerberos side of things, just ensure that the installation is set > up properly (permissions on the database directory and the /etc > files). > Yes, bind runs locally on my authentication server, in a chroot jail as per the installation and configuration instructions in BLFS. That said, chroot is not an end all be all of security as Randy mentioned above. The more services you have on a box, the more vulnerable you are plain and simple. If for nothing else, it gives more potential for exploits and even simple denial of service attacks. That said, I know that risk exists and personally I do take that (minimal IMO) risk for the example you have posted. Of course, everything is locked down to the outside world, even internally my routing and network division probably stinks of insanity by most people's standards. OT: After watching the long BackTrack Linux demo (highly recommended if you've never studied security as it gives a nice overview of penetration testing without getting super complex - and even if you have studied, it is fun to watch the simplicity of it), I've added even more firewall rules to further isolate my VLANs. :-) -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: git and https
On 10/01/2011 01:09 PM, Andrew Benton wrote: > But I wasn't happy with that, if my git couldn't use https that wasn't > good enough. As mentioned earlier, the ssl certificate is verified by > curl and this is actually a problem with how curl was set up. When > installing curl, it's possible to generate a bundle of ssl certificates > with the command: >make ca-bundle > > That will download a file from the mozilla source and run a perl script > to extract the ssl certificates. If you already have a copy of the > Firefox source you can use that instead of downloading it again (useful > if you're installing offline): >cp /path/to/firefox/security/nss/lib/ckfw/builtins/certdata.txt . >perl lib/mk-ca-bundle.pl -n >mv ca-bundle.crt /etc/ssl/certs > Now all that is needed is to tell git about the ssl certificates: >git config --global http.sslcainfo /etc/ssl/certs/ca-bundle.crt > That will make the change in ~/.gitconfig > To do it for everyone in /etc/gitconfig as root: >git config --system http.sslcainfo /etc/ssl/certs/ca-bundle.crt > > Hope that helps someone avoid retracing my steps > > Andy Andy, be sure that the perl script properly scrubs expired certificates and the test certs. Take a look at the BLFS CA bundle currently installed with OpenSSL (which also uses Mozilla as the upstream source for certificates). I just updated it about two weeks ago, so if the mk-ca-bundle.pl script included with curl is equal to the cert setup we currently have, then we could use that instead of maintaining our own two step process in BLFS as I had done (which I borrowed from Fedora, see the auxfiles/mkblfsca.sh in svn if you have it handy). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: GTK+2.20.1 compile problems
On 08/21/2011 06:15 PM, DJ Lucas wrote: > On 08/20/2011 03:00 PM, Thomas Benko wrote: >> Hail! >> I try to compile GTK+2.20.1, but it fails with >> /usr/lib/libxcb-xlib.so: undefined reference to `_xcb_unlock_io` >> /usr/lib/libxcb-xlib.so: undefined reference to `_xcb_lock_io` >> error. I don't know, what to do, google didn't helped me, because the >> solution I found with it was to freshen libxcb to 1.4. But I use >> libxcb1.7... :( >> What can I do now? >> > > That was an ABI change in libxcb. I take it you upgraded in place from > 1.3? Unfortunately, the solution was to remove a good portion of xorg > and related libs and then reinstall (basically anything that links to > libxcb-xlib.so). I had a thread on it in BLFS-Dev a few years back, but > am unable to find it. Found it! Dan had the solution for the libtool archives here: http://www.mail-archive.com/blfs-dev@linuxfromscratch.org/msg09549.html -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: GTK+2.20.1 compile problems
On 08/20/2011 03:00 PM, Thomas Benko wrote: > Hail! > I try to compile GTK+2.20.1, but it fails with > /usr/lib/libxcb-xlib.so: undefined reference to `_xcb_unlock_io` > /usr/lib/libxcb-xlib.so: undefined reference to `_xcb_lock_io` > error. I don't know, what to do, google didn't helped me, because the > solution I found with it was to freshen libxcb to 1.4. But I use > libxcb1.7... :( > What can I do now? > That was an ABI change in libxcb. I take it you upgraded in place from 1.3? Unfortunately, the solution was to remove a good portion of xorg and related libs and then reinstall (basically anything that links to libxcb-xlib.so). I had a thread on it in BLFS-Dev a few years back, but am unable to find it. The newer libxcb does not create libxcb-xlib (but old is left in place and is scattered about through a bunch of libtool archives. Removing it does no good as you'll simply break xorg completely. I realize it isn't much help, but that is what needs to happen. Backup first so you can quickly return from where you started if something goes terribly awry. Here is a gentoo guide to assist, but I'm not sure how relevant it is now days...it's been a long time since I've had to deal with it (I'm thinking late 2007 or early 2008). http://www.gentoo.org/proj/en/desktop/x/x11/libxcb-1.4-upgrade-guide.xml In fact..it looks a little specific to gentoo. They had libxcb-xlib emulation patches for a while. Basically it amounts to removing libxcb-xlib (all archives, pc file, etc) and everything linked to it and re-install). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Libtool search path
On 08/03/2011 11:49 AM, 高飛 wrote: > Excuse me. I am trying to build Gnome 3 in /opt/gnome. I've added > /opt/gnome/usr/lib/pkgconfig to PKG_CONFIG_PATH and /opt/gnome/usr/lib > to "ld.so.conf". > I get glib 2.24.7 installed to /opt/gnome and I am sure there is > "libgmodule-2.0.la" file in /opt/gnome/usr/lib. > But when I try to build GConf, error occurs and it seems that libtool > keeps trying to find "libgmodule-2.0.la" file in /usr/lib. Here's the > error message: > > /bin/sh ../libtool --tag=CC --mode=link gcc -g -O2 -Wall > -version-info 5:5:1 -no-undefined -o libgconf-2.la -rpath > /opt/gnome/usr/lib gconf-internals.lo gconf-backend.lo > gconf-changeset.lo gconf-error.lo gconf-listeners.lo gconf-locale.lo > gconf-schema.lo gconf-sources.lo gconf-value.lo gconf.lo > gconf-client.lo gconf-enum-types.lo GConfX-common.lo GConfX-skels.lo > GConfX-stubs.lo -pthread -Wl,--export-dynamic -L/opt/gnome/usr/lib > -lgio-2.0 -lgmodule-2.0 -lORBit-2 -lgobject-2.0 -lgthread-2.0 -lrt > -lglib-2.0 > /bin/grep: /usr/lib/libgmodule-2.0.la: No such file or directory > /bin/sed: can't read /usr/lib/libgmodule-2.0.la: No such file or directory > libtool: link: `/usr/lib/libgmodule-2.0.la' is not a valid libtool archive > > How can I adjust the path? > Thanks for your help. > Tester He What version of LFS? If recent, did you remove the installed glib and pkgconfig file in /usr? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Debugging bind
On 06/16/2011 06:42 AM, Andrew Benton wrote: > On Thu, 16 Jun 2011 02:45:53 -0500 > DJ Lucas wrote: >> Hmm, latest version of Bind is working fine for me 9.8.0-P2, though I >> had to rejig the test suite patch because they added ipv6 interface tests. >> >> root@anu [ /sources/smbldap-tools-0.9.6 ]# nslookup >> > >> Server: 192.168.143.225 >> Address: 192.168.143.225#53 >> >> Non-authoritative answer: >> www.nhs.uk canonical name = www.prod.nhs.uk.akadns.net. >> Name:www.prod.nhs.uk.akadns.net >> Address: 217.64.234.65 >> > >> >> Double check the format and perms of your root.hints file. That address >> is A.ROOT-SERVERS.NET and is rather odd that it is looking at the >> host for an A record (especially if no IPv6 GW is setup). >> >> Latest hints file is dated June 8th, 2011, and added only the >> record for D, and changed the TTL for L's record from 7D to 41.6_D >> (like the rest). Book is current for all A records, but records >> should probably be added at some point (because I like that format much >> better than the one at IANA/internic). Probably also need to add ::1 to >> /etc/hosts in LFS at some point soon enough. > > I had updated to the latest root.hints file. I've since edited it to > get rid of the records but to no avail, I still get this error: > > Jun 16 12:29:21 eccles named[26648]: createfetch: nhs.uk A > Jun 16 12:29:21 eccles named[26648]: createfetch: . NS > Jun 16 12:29:21 eccles named[26648]: error (network unreachable) resolving > 'nhs.uk/A/IN': 2a01:40:1001:35::2#53 > Jun 16 12:29:23 eccles named[26648]: createfetch: nsa.nhs.uk > Jun 16 12:29:23 eccles named[26648]: createfetch: nsb.nhs.uk > Jun 16 12:29:51 eccles named[26648]: client 127.0.0.1#54742: query failed > (SERVFAIL) for nhs.uk/IN/A at query.c:6199 > > I've updated to bind-9.8.0-P2 (and worked around a bug using ssl in > chroot) but it still doesn't work. It's very annoying. It is still defaulting to v6. Sorry. Forgot about the libgost.so issue. Kernel IP Auto Configuration turned on? What is the output of 'ip addr show' and 'ip route'? Only other thing I can think of is using the real ip of the box instead of 127.0.0.1. -- DJ Lucas > > Bruce Dubbs wrote: > >> I agree with you that it appears to be an IPV6 issue, but the >> question is how would it get fixed? There would need to be some sort of >> active proxy to get from an IPV4 address to an IPV6 address. >> >> Perhaps named just inquires via the kernel. Is IPV6 enabled there? > > IPv6 was enabled in the kernel, but as I don't have an IPv6 connection > I'd only enabled IPv6 in IPv4 tunneling. I've since recompiled the > kernel and enabled a lot more IPv6 options but it hasn't made any > difference. > > Thanks for your time > > Andy -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Debugging bind
On 06/15/2011 12:13 PM, Andrew Benton wrote: > Hello World! > I've installed bind as described in BLFS and have done for years; it's > never given me any problems and speeds up browsing. It's normally the > only nameserver I need so I have only needed 1 line in /etc/resolv.conf: > > nameserver 127.0.0.1 > > However, I recently found I am unable to open pages at www.nhs.uk in > Firefox. /var/log/sys.log has lines like this: > > Jun 15 16:22:05 eccles named[911]: network unreachable resolving > 'www.nhs.uk/A/IN': 2001:503:ba3e::2:30#53 > Jun 15 16:22:07 eccles named[911]: network unreachable resolving > 'nsa.nhs.uk//IN': 2001:7fe::53#53 Hmm, latest version of Bind is working fine for me 9.8.0-P2, though I had to rejig the test suite patch because they added ipv6 interface tests. root@anu [ /sources/smbldap-tools-0.9.6 ]# nslookup > Server: 192.168.143.225 Address:192.168.143.225#53 Non-authoritative answer: www.nhs.uk canonical name = www.prod.nhs.uk.akadns.net. Name: www.prod.nhs.uk.akadns.net Address: 217.64.234.65 > Double check the format and perms of your root.hints file. That address is A.ROOT-SERVERS.NET and is rather odd that it is looking at the host for an A record (especially if no IPv6 GW is setup). Latest hints file is dated June 8th, 2011, and added only the record for D, and changed the TTL for L's record from 7D to 41.6_D (like the rest). Book is current for all A records, but records should probably be added at some point (because I like that format much better than the one at IANA/internic). Probably also need to add ::1 to /etc/hosts in LFS at some point soon enough. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Fwd: blfs-support Digest, Vol 2321, Issue 1
On 06/13/2011 01:33 AM, janu mam wrote: > hi DJ Lucas > > rroot[~]#cat /etc/profile > > #set the initial PKG_CONFIG_PATH > export PKG_CONFIG_PATH=/usr/lib/pkgconfig:/usr/share/pkgconfig > Well..there is part of the problem. Please copy and paste from a terminal or output to a file. Conversely, if that is all that is in your /etc/profile, then you have other problems. ;-) > and see this: $PREFIX=/usr No, unfortunately I can not see it. You have to show me. I am not capable of telling the difference between your intentions and what you've actually done without you showing me what you've done. Please execute the following commands and copy and paste both input and output in your response: echo $PKG_CONFIG_PATH bash -c 'echo $PKG_CONFIG_PATH' echo $PREFIX bash -c 'echo $PREFIX' Here are some alternate commands that can show me exactly the same thing vs. tell me what your think you are seeing (and likely provide a hint as to why I suspect that the commands are not working for you as you expect them to): export | grep "PRE" set | grep "PRE" export | grep "PKG" set | grep "PKG" Also, just curious, why would you use an environment variable when the keystrokes are less for the actual prefix? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: The required package libxfce4util-1.0 was not found on your system
On 06/12/2011 01:43 AM, janu mam wrote: > checking GTHREAD_LIBS... -pthread -lgthread-2.0 -lrt -lglib-2.0 > checking for libxfce4util-1.0>= 4.8.0... not found > *** The required package libxfce4util-1.0 was not found on your system. > > here i can find all new versions:i have tried all versions here ,still > same as above: libxfce4util-1.0 > http://archive.xfce.org/src/xfce/libxfce4util/ > > i have google it: but no success: > > where can i get :libxfce4util-1.0 ? > how to solve this problem The error messages says it all. What is the value of $PREFIX? Did you add $PREFIX/lib/pkgconfig and $PREFIX/share/pkgconfig to the PKG_CONFIG_PATH environment variable? What is the output of "pkg-config --modversion libxfce4util-1.0"? You should likely be using http://archive.xfce.org/src/xfce/libxfce4util/4.8/libxfce4util-4.8.1.tar.bz2 -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: The trouble with not following instructions (a problem with X.org)
On 05/23/2011 10:36 AM, William Immendorf wrote: > (Janu, you should either configure the list to not receive messages in > digest, or just change the subject heading when you reply.) > > On Mon, May 23, 2011 at 1:43 AM, janu mam > wrote: >>> janu[~Downloads/xc/lib]echo $XORG_CONFIG >> >> result :nothing > You did not follow the directions, as you can see by reading the > renamed subject header. Go back to "Introduction to Xorg 7.6", and > follow the instructions EXACTLY. AND I REALLY MEAN EXACTLY. > > Also refer to my previous messages about this problem. > William, as per my private email the other day, if a user is unable to follow simple instructions, take for example posting to the correct list after a couple of reminders, that user is unlikely to be capable of completing his goals with regard to LFS. Yes, there must be some consideration for language barriers and such, so a couple of reminders should be acceptable. However, a reasonably capable person would not continue to ignore your suggestion to post to the list. Continuing to provide support may actually be more damaging than simply ignoring the request. The user has now invested more time into the project, and the dissatisfaction will be that much more WRT the impending failure. My suggestion is that you block all messages from that user for a short time...until he/she is able to behave within the established expectations of the community. It has been done many times before, and it happens to work really well. Some of those types of users will sit and observe for a time, and some later even become very valuable members of our little community. ;-) -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: (Sorta) OT: Log level for Iptables?
On 04/23/2011 07:21 PM, Fernando Oliveira wrote: > > If I remember correctly it stops logging to the console by modifying the > initscript /etc/rc.d/init.d/sysklogd: > > boot_mesg "Starting kernel log daemon..." > loadproc klogd -c 3 > Please do not suggest editing distributed scripts as they will be overwritten upon upgrade. There are two supported methods provided by the boot scripts for handling the console log level via configuration file: echo "LOGLEVEL=3" >> /etc/sysconfig/console && echo "kernel.printk=3" >> /etc/sysctl.conf -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: blfs/lfs boot logging
On 05/12/2011 03:53 PM, luxInteg wrote: > I would be most grateful if someone could explain how lfs/blfs logs boot > processes. I cant find a boot.log file in /var/log > > sincerely > luxInteg > -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: > http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above > information page Default boot scripts do not log anything. LSB scripts do, which is currently being worked on...should hopefully be in this weekend or shortly there after. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Configuration files
On 05/12/2011 04:26 PM, Pol Vangheluwe wrote: > Op 12-mei-2011, om 11:34 heeft Simon Geard het volgende geschreven: > >> On Wed, 2011-05-11 at 23:18 +0200, Pol Vangheluwe wrote: >> >>> The bash manual tells me that the structure "command1&& command2" >>> executes command2 >>> if and only if command1 returns the exit status of zero. >>> But here, there is only one command (cat), so I really don't >>> understand why "&&" is needed. >> >> Correct on the syntax, but you're wrong that only one command is >> present. From the sudo example: >> >> cat> /etc/pam.d/sudo<< "EOF"&& >> # Begin /etc/pam.d/sudo >> ...content removed... >> # End /etc/pam.d/sudo >> EOF >> chmod 644 /etc/pam.d/sudo >> >> >> There are two commands - the 'cat', and the subsequent 'chmod'. The >> xinetd examples are similar. >> >> >>> BTW: can you explain why the 'xinetd.conf' script has no "&&" but the >>> 'xinetd.d/' scripts have it on the xinetd-2.3.14 page? >>> The first runs well, the other fail... >> >> Because the xinetd.conf script is a single command, just one 'cat' >> statement. But the xinetd.d case consists of twenty or so commands, all >> separated by&&. >> >> Simon. > > OK, I understand… > > My problem was caused by the fact that I made a (one-by-one) selection of the > 22 cat commands > for the servers I wanted to use. If you copy the content of the grey > rectangle as a whole > then there is no problem. > > However, I think that the script is also working without the "&&" operators. > You only loose the conditional dependency from one cat command to the next > one. > In other words: the script with "&&" stops when a cat command fails but > continues with the next cat command without the"&&"? Correct. The purpose of using the && syntax is that if anything errors, the commands will not continue past the error, which gives you a much better chance to catch the offending command. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Configuration files
On 05/11/2011 04:18 PM, Pol Vangheluwe wrote: > > The bash manual tells me that the structure "command1 && command2" > executes command2 > if and only if command1 returns the exit status of zero. > But here, there is only one command (cat), so I really don't understand > why "&&" is needed. > The problem is that you misunderstand the syntax. When a conditional such as '&&' is used, none of the commands are interpreted by bash until there are no conditionals. Here is a simple example similar to the sudo instructions: cat > file << "EOF" && # begin file put some stuff here # end file EOF chmod 644 file Here is a _logical_ breakdown of how the commands are interpreted. It is bash that interprets '> file << "EOF"', not cat. Bash waits for EOF and puts everything else in a queue, when it receives EOF, it then interprets the && (cat is not executed yet), and it waits for the next command. If you break at this point, bash is interrupted and the cat command is not executed. chmod is the next command that bash interprets and it ends the queue as there is no conditional at the end. In the case of the sudo commands, you must enter both cat and chmod before bash actually executes anything. In the case of xinetd's instructions, there are 22 commands that have to be entered into the queue before any work is done. Do not take the above text as a _technical_ explanation, as the order of events has been greatly simplified for example purposes. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: icedtea on amd64
On 04/19/2011 03:08 PM, luxInteg wrote:
> On Tuesday 19 April 2011 01:14:04 DJ Lucas wrote:
>
>> I honestly don't know. I didn't think it was not possible to *build* it
Ugh...that should have read: "I didn't think it was possible..."
>> without GTK, but the java binary doesn't link to GTK+ libraries
>> directly, and certainly runs without them handy. My guess is that
>> they've removed anything requiring GTK libs and packaged it up. I've
>
> I MAY have got lucky wit jdk-1.7My build now ends like this
>
> -- Build times --
> Target all_product_build
> Start 2011-04-19 15:13:27
> End 2011-04-19 15:39:23
> 00:02:26 corba
> 00:07:53 hotspot
> 00:00:30 jaxp
> 00:00:34 jaxws
> 00:13:40 jdk
> 00:00:53 langtools
> 00:25:56 TOTAL
>
> I then copied the j2sdk image and it seems to be working. I set the copied
> location as ${JAVA_HOME}. I executed ${JAVA_HOME}/bin/java -version and it
> responded. And ant is no longer reporing errors. Incidently I tried
> compiling postgresql-jdbc connector and it failed. Ant was built with
> jdk-1.6 so is it . worth rebuilding ant with openjdk-1.7?
>
>
Is there something wrong with using the pre-compiled 64bit icedtea6
builds I have in the book? Not that I want to deter you from testing,
but I'd be concerned about stability initially as OpenJDK7 is still
experimental and there have been a lot of changes to the platform
between OpenJDK6 and OpenJDK7.
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: GTK texinfo reader?
On 04/18/2011 10:56 AM, Andrew Benton wrote: > Hello World! > > I don't like reading info pages in the terminal. Whenever I've tried > I've spent the whole time trying to figure out how to navigate before I > give up and use Google. For a while I was able to use gnome's yelp to > read info pages but with the transition to Gnome-3 I can't get it to > work. They took that functionality away in 2.26, was supposed to be back in 2.28 and it kinda works in 2.30. Got Gnome3 working already? IIRC, you script, you got any info to share? Dependencies, build order, etc? > It compiles and installs OK but segfaults when I run it. I'd much > rather use something simpler that didn't depend on a bunch of other > gnome things. Does anyone know of an application with a simple gtk > graphical user interface that can read info pages? Not tried it, but web browser of choice and this would seem to fit the bill if a local copy of Apache httpd isn't too much to ask: http://info2html.sourceforge.net/ Otherwise, I see nothing regarding gtk info browsers. There is tkinfo and xinfo available. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: icedtea on amd64
On 04/18/2011 06:05 PM, William Tracy wrote: > On Friday 28 January 2011 11:28:07 luxInteg wrote: >> Greetings >> >> I want to have a go at building openjdk using icedtea. >> http://www.linuxfromscratch.org/blfs/view/cvs/general/icedtea6.html >> I notice that a binary is needed presumeably for so-called >> 'bootstraping' and the one available from blfs is for 32-bit pentium >> cpu's. I am dabbling with 64-bit amdK(8+} cpu's. >> >> Does anyone know of a 'bootstrapping' icetea binary for amd64's OR >> whether such is redundant and if so how to compile icetea >> without the prebuilt binary.? >> >> thanks in advance >> >> luxInteg >> >> is it possible to build and run IcedT without gtk? > Debian and company ship 'headless' Java packages, so I know it's possible: > http://packages.debian.org/squeeze/openjdk-6-jre-headless > I honestly don't know. I didn't think it was not possible to *build* it without GTK, but the java binary doesn't link to GTK+ libraries directly, and certainly runs without them handy. My guess is that they've removed anything requiring GTK libs and packaged it up. I've *heard* that it is possible on OpenJDK-7, but I haven't verified myself. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: LibreOffice-3.3.2.2
On 03/28/2011 12:25 PM, Fernando Oliveira wrote:
> --- Em dom, 27/3/11, DJ Lucas escreveu:
>
>> De: DJ Lucas
>> Assunto: Re: LibreOffice-3.3.2.2
>> Para: blfs-support@linuxfromscratch.org
>> Data: Domingo, 27 de Março de 2011, 15:35
>>
>
> Thank you for your mail.
>
>> LibreOffice suffers the same problem and I
>> do not recommend it for the book unless a rather large
>> group of people can come to a consensus on "required" dependencies.
>
> Please, include me in the group.
>
>> Here are the buildopts for OOo's configure that were used
>> in that patch set. Note that this is not for the top level configure
>> script, but for OOo's internal configure script after LibreOffice
>> does it's thing to the source tree
>> (no mono patches were applied in the set I had defined):
>>
>> PKG_CONFIG=/usr/bin/pkg-config ./configure \
>> --disable-odk --disable-qadevoo --disable-fontooo \
>> --enable-cairo --enable-opengl --enable-dbus \
>> --enable-gio --disable-kdeab --enable-minimizer \
>> --enable-presenter-console --enable-pdfimport \
>> --enable-wiki-publisher --enable-report-builder \
>> --without-fonts --without-ppds --without-afms \
>> --with-system-stdlibs --with-system-cairo \
>> --with-system-zlib --with-system-openssl \
>> --with-system-jpeg --with-system-expat \
>> --with-system-libxml --with-system-python \
>> --with-system-poppler --with-system-db \
>> --with-system-libxslt --with-system-odbc-headers \
>> --with-system-xrender-headers --with-system-curl \
>> --with-system-boost --with-system-neon \
>> --disable-gnome-vfs --with-dict=ALL \
>> --with-vendor="BLFS" --with-build-version=3.2.1-1 \
>> --with-package-format=native --with-use-shell=bash
>
> Next time, I will try using option closer to these.
>
Actually, I think those were the incorrect ones
I forgot, I had deleted my local change sets, however, I had put them in
my home directory for safe keeping:
http://www.linuxfromscratch.org/~dj/ooo-build-3.2.1.4-BLFS_Config-1.patch
>> As you can see, I was pretty heavy handed with the
>> dependencies and am honestly not comfortable making decisions about _your_
>> systems. Unfortunately, this is what was required to get what I
>> deemed "full functionality" out of the build with the least amount of
>> internal products (based on what we have in BLFS now).
>
Same comments apply...however, it might be possible to encourage the
upstream developers to add a couple of generic targets (or we can do
them locally), say --with-vendor={KDE3,KDE4,GNOME2,GNOME3}. That would
make the patch somewhat complex but would certainly make the book more
readable and more malleable for different installations. They could, of
course, be called anything we want if kept locally. ie: BLFS-GNOME3,
BLFS-GNOME2, BLFS-TRINITY, BLFS-KDE4...
> Probably you know already, the page
>
> http://wiki.documentfoundation.org/Development/How_to_build/Configure_options
>
> has some information. I include the address here, in order to be used by me
> and others.
>
Thanks for the link. Looks like a few things have changed with the
predefined patch sets so that patch may be of little use other than a
generic set of directions now.
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: Firefox-4.0 and GCC-4.6
On 04/03/2011 09:49 PM, al...@verizon.net wrote: > > BTW, I'm not sure I have a java compiler on this machine > (I do have jdk-precompiled-1.6.0_12 installed, though) > Yes, you do, however, you really should update it with either IcedTea6 (preferred), or Oracle JDK-1.6.0_24. The version you have installed has many security exploits and is likely very buggy. > > 3. Are there instructions on installing a java compiler in > Linux from Scratch? (separate from jdk? ) There are three java compilers in BLFS. You have the Oracle JDK (previously Sun), IcedTea6 (open source version of the JDK partially provided by, and Open Source project sponsored and paid for by Oracle), and gcc. The compiler for the first two is "javac" and the last is "gcj". You have a working compiler for use by Mozilla installed if you've installed the JDK or IcedTea6 per the book's instructions (though the Sun JDK in the book is really old and likely scheduled for removal if nobody wants to maintain it as IcedTea6 provides a compatible and more up to date implementation that is completely free software). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: boost-1_46_1,icu,python-2.7.1
On 03/27/2011 09:00 PM, luxInteg wrote: > > I came from that route to icu-4.5.2 (the only one that built). In fact I > tried both versions 4_6_1 and 4_7_1 and both failed -see below:- > > > > - > advice would be appeciated AND > do you or any one else on list know what the dependencies of newer > verions of icu are? Doesn't look like anything above a base LFS is required except for the samples (and my own make output). What I have for the relevant section of log output is pretty generic (this is icu4c-4.6, not 4.6.1): generating dependency information for makeconv.c make[2]: Leaving directory `/lfs-sources/icu/source/tools/makeconv' make[2]: Entering directory `/lfs-sources/icu/source/tools/makeconv' gcc -D_REENTRANT -I../../common -I./../toolutil -O3 -Wall -ansi -pedantic -Wshadow -Wpointer-arith -Wmissing-prototypes -Wwrite-strings -Wno-long-long -c -o makeconv.o makeconv.c gcc -D_REENTRANT -I../../common -I./../toolutil -O3 -Wall -ansi -pedantic -Wshadow -Wpointer-arith -Wmissing-prototypes -Wwrite-strings -Wno-long-long -c -o ucnvstat.o ucnvstat.c gcc -D_REENTRANT -I../../common -I./../toolutil -O3 -Wall -ansi -pedantic -Wshadow -Wpointer-arith -Wmissing-prototypes -Wwrite-strings -Wno-long-long -c -o genmbcs.o genmbcs.c gcc -D_REENTRANT -I../../common -I./../toolutil -O3 -Wall -ansi -pedantic -Wshadow -Wpointer-arith -Wmissing-prototypes -Wwrite-strings -Wno-long-long -c -o gencnvex.o gencnvex.c g++ -O3 -W -Wall -ansi -pedantic -Wpointer-arith -Wwrite-strings -Wno-long-long -o ../../bin/makeconv makeconv.o ucnvstat.o genmbcs.o gencnvex.o -L../../lib -licutu -L../../lib -licui18n -L../../lib -licuuc -L../../stubdata -licudata -lpthread -ldl -lm cd ../.. \ && CONFIG_FILES=tools/makeconv/makeconv.1 CONFIG_HEADERS= /bin/sh ./config.status config.status: creating tools/makeconv/makeconv.1 make[2]: Leaving directory `/lfs-sources/icu/source/tools/makeconv' The referenced errors are internal. Are you doing anything non standard WRT the build directory? Maybe not building directly in the source tree or something? Are you trying to build in parallel? Try using make -j1. While it may not seem relevant, you should always include the "make[?]: entering..." lines as well so we can see what make itself is trying to do, and also the actual gcc lines (but these would be included if you include the make lines). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: boost-1_46_1,icu,python-2.7.1
On 03/27/2011 06:56 PM, William Immendorf wrote: > On Sun, Mar 27, 2011 at 6:20 PM, luxInteg wrote: >> ~/boost_1_46_1/tools/build/v2/build/configure.jam:145: in builds-raw >> *** argument error >> * rule UPDATE_NOW ( targets * : log ? : ignore-minus-n ? ) >> * called with: ( >> has_icu >> : >> : ignore-minus-n : ignore-minus-q ) >> * extra argument ignore-minus-q >> (builtin):see definition of rule 'UPDATE_NOW' being called > Looks like an issue with ICU. Looking at Arch Linux, they are using > ICU 4.6.1 (newer than what you were using) and it seems to compile > fine. > > Maybe you should try upgrading ICU to 4.6.1. If that dosen't work, > then use an older version of ICU. > > > Looks like a good suggestion. Build of 1.46.1 worked here. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: boost-1_46_1,icu,python-2.7.1
On 03/27/2011 06:20 PM, luxInteg wrote: > ### I forwarded this message as the original had typos both in heading > and body of email > > Greetings > > I am trying to biuld boost-1.46.1 on a 64bit (amd64) cblfs machine with > gcc-4.5.2/python-2.7.1//icu-4.5.2.. Results are as follows:- > > Ugh...ignore my last message. Rebuilding boost now... -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: boost-1_46_1,icu,python-2.7.1
DIR-LSB ]$ ls /var/log/llog/boost-1.45.0.llog
/var/log/llog/boost-1.45.0.llog
dj [ BOOK-DESTDIR-LSB ]$ icuinfo
International Components for Unicode for C/C++
Copyright (C) 2010, International Business Machines Corporation and
others. All Rights Reserved.
Compiled-Version: 4.6
Runtime-Version: 4.6
Compiled-Unicode-Version: 6.0
Runtime-Unicode-Version: 6.0
Platform: U_LINUX
Build: x86_64-unknown-linux-gnu
C compiler: gcc
C++ compiler: g++
ICUDATA: icudt46l
Data Directory:
ICU Initialization returned: U_ZERO_ERROR
Default locale: en_US
CLDR-Version: 1.9
Default converter: UTF-8
Default TZ: America/Chicago
TZ data version: 2010o
Plugin file is: /usr/lib/icu/icuplugins46.txt
dj [ BOOK-DESTDIR-LSB ]$
Attached are my build scripts.
-- DJ Lucas
#!/bin/sh
# Package Information
export version="2.7.1"
export tarball="Python-${version}.tar.bz2"
export dir="Python-${version}"
export md5sum="aa27bc25725137ba155910bd8e5ddc4f"
export download="http://www.Python.org/ftp/Python/$version}/${tarball}";
export docs_tarball="python-${version}-docs-html.tar.bz2"
export docs_download="http://docs.python.org/archives/${docs_tarball}";
export docs_md5sum="312d4da0c1d8adbfb27f4f0e09f2cffe"
. /etc/lspec.conf
. ${lspec_dir}/.count
let count++
if [ "${count}" -lt "10" ]; then
lcount="000${count}"
elif [ "${count}" -lt "100" ]; then
lcount="00${count}"
elif [ "${count}" -lt "1000" ]; then
lcount="0${count}"
fi
# get_package
cat ${lspec_dir}/Python1 > ${build_dir}/logs/$lcount-Python-${version}-1 &&
{
cd "${build_dir}" &&
llog -p &&
tar -xf "${source_dir}/${tarball}" &&
cd "${dir}" &&
time {
echo "## PREP AND CONFIGURE ##" &&
sed -i "s/ndbm_libs = \[\]/ndbm_libs = ['gdbm', 'gdbm_compat']/" setup.py &&
./configure --prefix=/usr --enable-shared &&
make &&
# No sense testing somthing that will get rebuilt later
# Checking build dir size is negligable
#BUILD_SIZE="$(du -shb ./ | awk '{print $1}')" &&
#echo "## BEGINING TESTSUITE ##" &&
#time {
#make test || true
#} &&
echo "## BEGINING INSTALL ##" &&
make install &&
#chmod -v 755 /usr/lib/libpython${version}.so.1.0 &&
install -v -m755 -d /usr/share/doc/Python-${version}/html &&
tar --strip-components=1 \
--no-same-owner \
--no-same-permissions \
-C /usr/share/doc/Python-${version}/html \
-xvf "${source_dir}/${docs_tarball}" &&
for file in `find /usr/share/doc/Python-${version} -type f`
do touch $file
done
} &&
llog Python-${version} &&
echo "count=${count}" > "${lspec_dir}/.count" &&
INSTALL_SIZE="$( du -shb `awk '{print $1}' /var/log/llog/Python-${version}.llog
| sed /\(M\)/d ` | awk '{print $1}' | paste -sd+ | bc )" &&
FULL_BUILD_SIZE="$(du -shb ./ | awk '{print $1}')" &&
cd .. &&
rm -rf "${build_dir}/${dir}" &&
echo "" &&
#echo -n "Diskspace required with testsuite is: " &&
#echo -e "$FULL_BUILD_SIZE\n$INSTALL_SIZE" | paste -sd+ | bc &&
#echo -n "Diskspace required without testsuite is: " &&
#echo -e "$BUILD_SIZE\n$INSTALL_SIZE" | paste -sd+ | bc &&
echo -n "Diskspace required without testsuite is: " &&
echo -e "$FULL_BUILD_SIZE\n$INSTALL_SIZE" | paste -sd+ | bc &&
echo ""
} 2>&1 | tee -a ${build_dir}/logs/$lcount-Python-${version}-1
#!/bin/sh
# Package Information
export version="1.45.0"
export expver="1_45_0"
export tarball="boost_${expver}.tar.bz2"
export dir="boost_${expver}"
export md5sum="d405c606354789d0426bc07bea617e58"
export
download="http://sourceforge.net/projects/boost/files/boost/${version}/${tarball}/download";
. /etc/lspec.conf
. ${lspec_dir}/.count
let count++
if [ "${count}" -lt "10" ]; then
lcount="000${count}"
elif [ "${count}" -lt "100" ]; then
lcount="00${count}"
elif [ "${count}" -lt "1000" ]; then
lcount="0${count}"
fi
# get_package
cat ${lspec_dir}/boost > ${build_dir}/logs/$lcount-boost-${version} &&
{
cd "${build_dir}" &&
sudo /sbin/llog -p &&
tar -xf "${source_dir}/${tarball}" &&
cd "${dir}" &&
time {
echo "## PREP AND CONFIGURE ##" &&
./bootstrap.sh --prefix=/usr &&
./b
Re: Firefox-4.0
On 03/27/2011 03:40 PM, al...@verizon.net wrote: > > My question: > Is your patch good for any>= 2.22.1 ? > It applies with offsets to 2.24.x IIRC. I don't think I had to rediff, but it should be simple to do so. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: python-2.7.1 gobject-introspection
On 03/26/2011 09:24 AM, Thomas Trepl wrote: > Hiho, > > don't know whether it has do something with it, but did you do the sed-command > which is in the book at gobject-introspection when having Python-2.7 > installed? > Also, what version of gio? There were some syntax changes that need to be accounted for in newer versions. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Firefox-4.0
On 03/22/2011 06:02 AM, Andrew Benton wrote: On Mon, 21 Mar 2011 20:27:23 -0500 (CDT) al...@verizon.net wrote: "ASUS P5E-VM HDMI" with Intel G35/ICH9R. Intel Core2 Duo CPU E8400 @ 3.00GHz, 4GB (B)LFS i686-pc-linux-gnu 2.6.37.3, Xorg-7.6 glib-2.28.3, gtk+-3.0.3, sqlite-3.7.5, etc. To Whom It May Concern: On a dare, I decided to install the freshly released Firefox 4.0. It was also on a "squeeze" - the 3.6.14 and 3.6.15 failed me miserably: I had lost my beloved back and forward arrows of the navigation buttons (see Bug 640850). I was also beyond a GTK+-2/3 version where I may have "fixed" the problem by going back to 3.6.13. You can fix the back and forward arrows by setting a gtk theme You shouldn't have to. The proper fix is in GTK+. The change in GTK+ that broke the back and forward buttons should not have happened in a minor version release. Reverting the compat names has little real effect on GTK+, only that they have been deprecated for a while and were finally removed IIRC. This change should have been put on hold for gtk+-3.0. Patch attached. -- DJ Lucas --- gtk+-2.22.0-orig/gtk/Makefile.in2010-12-31 02:42:51.0 -0600 +++ gtk+-2.22.0/gtk/Makefile.in 2010-12-31 02:46:42.0 -0600 @@ -1598,7 +1598,171 @@ stock-icons/24/folder-remote.png\ stock-icons/24/user-home.png\ stock-icons/24/user-desktop.png \ - stock-icons/24/text-x-generic.png + stock-icons/24/text-x-generic.png \ + stock-icons/16/gtk-quit.png \ + stock-icons/16/gtk-info.png \ + stock-icons/16/gtk-file.png \ + stock-icons/16/gtk-open.png \ + stock-icons/16/gtk-print-preview.png\ + stock-icons/16/gtk-print.png\ + stock-icons/16/gtk-properties.png \ + stock-icons/16/gtk-revert-to-saved-ltr.png \ + stock-icons/16/gtk-revert-to-saved-rtl.png \ + stock-icons/16/gtk-save-as.png \ + stock-icons/16/gtk-new.png \ + stock-icons/16/gtk-harddisk.png \ + stock-icons/16/gtk-clear.png\ + stock-icons/16/gtk-copy.png \ + stock-icons/16/gtk-cut.png \ + stock-icons/16/gtk-delete.png \ + stock-icons/16/gtk-find-and-replace.png \ + stock-icons/16/gtk-find.png \ + stock-icons/16/gtk-paste.png\ + stock-icons/16/gtk-redo-ltr.png \ + stock-icons/16/gtk-redo-rtl.png \ + stock-icons/16/gtk-select-all.png \ + stock-icons/16/gtk-undo-ltr.png \ + stock-icons/16/gtk-undo-rtl.png \ + stock-icons/16/gtk-directory.png\ + stock-icons/16/gtk-unindent-ltr.png \ + stock-icons/16/gtk-unindent-rtl.png \ + stock-icons/16/gtk-indent-ltr.png \ + stock-icons/16/gtk-indent-rtl.png \ + stock-icons/16/gtk-justify-center.png \ + stock-icons/16/gtk-justify-fill.png \ + stock-icons/16/gtk-justify-left.png \ + stock-icons/16/gtk-justify-right.png\ + stock-icons/16/gtk-bold.png \ + stock-icons/16/gtk-italic.png \ + stock-icons/16/gtk-strikethrough.png\ + stock-icons/16/gtk-underline.png\ + stock-icons/16/gtk-goto-bottom.png \ + stock-icons/16/gtk-go-down.png \ + stock-icons/16/gtk-goto-first-ltr.png \ + stock-icons/16/gtk-home.png \ + stock-icons/16/gtk-jump-to-ltr.png \ + stock-icons/16/gtk-jump-to-rtl.png \ + stock-icons/16/gtk-goto-last-ltr.png\ + stock-icons/16/gtk-go-forward-ltr.png \ + stock-icons/16/gtk-go-back-ltr.png \ + stock-icons/16/gtk-goto-top.png \ + stock-icons/16/gtk-go-up.png\ + stock-icons/16/gtk-about.png\ + stock-icons/16/gtk-help.png \ + stock-icons/16/gtk-missing-image.png\ + stock-icons/16/gtk-add.png \ + stock-icons/16/gtk-remove.png \ + stock-icons/16/gtk-floppy.png \ + stock-icons/16/gtk-cdrom.png\ + stock-icons/16/gtk-media-pause.png \ + stock-icons/16/gtk-media-play-ltr.png \ + stock-icons/16/gtk-media-play-rtl.png \ + stock-icons/16/gtk-media-stop.png
Re: LibreOffice-3.3.2.2
On 03/26/2011 12:23 PM, Fernando Oliveira wrote: > Forgot to mention that I happily was using the spell in LibreOffice to write > the previous message. > > []s, > Fernando de Oliveira > Natal, RN, BRAZIL > > > Just an FYI. I had taken a different approach completely. LibreOffice has a concept of target specific patch sets with include directives for smaller, generic sets of patches. I am unable to find my most recent set from Go-oo, but it should apply with a little work when I do manage to find it (it is buried deep in one of my older lfs partitions on my main build rig, I was pretty sure it was on 2 but unable to find it). I was actually planning on Go-oo when I updated the book to upstream OOo-3.2.1. I was just forced to make too many decision on behalf of the end user, so I scrapped it. LibreOffice suffers the same problem and I do not recommend it for the book unless a rather large group of people can come to a consensus on "required" dependencies. More on that: The switches supplied to configure have only part of the desired affect on what is passed to the real OOo build. While probably not required to get a working setup at the moment, they will again be required in the future as we approach more bleeding edge targets, and absolutely are required to use all possible system libraries. These patch sets add to the upstream OOo build and the minimal number of configure directives are not enough to use all system libraries in the final build. You will have to understand both build systems to get it exactly as you want it, and I personally have no idea how to get this into the book without taking the controlling hand and choosing for everyone what are "required" dependencies for a "BLFS" target. When I find my old patch set and instructions, I'll post them for review as it will explain much better than I can in words of what is actually required. Here are the buildopts for OOo's configure that were used in that patch set. Note that this is not for the top level configure script, but for OOo's internal configure script after LibreOffice does it's thing to the source tree (no mono patches were applied in the set I had defined): PKG_CONFIG=/usr/bin/pkg-config ./configure \ --disable-odk --disable-qadevoo --disable-fontooo \ --enable-cairo --enable-opengl --enable-dbus \ --enable-gio --disable-kdeab --enable-minimizer \ --enable-presenter-console --enable-pdfimport \ --enable-wiki-publisher --enable-report-builder \ --without-fonts --without-ppds --without-afms \ --with-system-stdlibs --with-system-cairo \ --with-system-zlib --with-system-openssl \ --with-system-jpeg --with-system-expat \ --with-system-libxml --with-system-python \ --with-system-poppler --with-system-db \ --with-system-libxslt --with-system-odbc-headers \ --with-system-xrender-headers --with-system-curl \ --with-system-boost --with-system-neon \ --disable-gnome-vfs --with-dict=ALL \ --with-vendor="BLFS" --with-build-version=3.2.1-1 \ --with-package-format=native --with-use-shell=bash As you can see, I was pretty heavy handed with the dependencies and am honestly not comfortable making decisions about _your_ systems. Unfortunately, this is what was required to get what I deemed "full functionality" out of the build with the least amount of internal products (based on what we have in BLFS now). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Xorg Applications and MesaLib
On 03/14/2011 05:52 PM, rhubarbpie...@gmail.com wrote: > > Should MesaLib be listed before Xorg Applications in chapter 23 (X > Window System Environment) of the BLFS development documentation? Doesn't have to be, but my intention certainly was to put it in a logical build order. :-) Thanks for the catch. I will fix it in a few moments (won't show up on the website till around 10:30 GMT). > In explanation, I receive a xdriinfo error when compiling Xorg > Applications. The error is apparently due to not compiling Mesa prior > to Xorg Applications. I see MesaLib listed as optionally required for > OpenGL support. However, I see no OpenGL option when compiling Xorg > Applications. > > If MesaLib is compiled prior to Xorg Applications I receive no error. > Does it hurt to compile MesaLib in that order? > -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Xorg Protocol Headers Utilities dependency?
On 03/07/2011 08:51 PM, DJ Lucas wrote: > Hopefully tonight. Fixed, along with a bunch of other xorg related tasks I had in queue. Unfortunately, I haven't completed the new xorg-config page yet, but it's coming soon. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Xorg Protocol Headers Utilities dependency?
On 03/07/2011 04:44 PM, rhubarbpie...@gmail.com wrote: > > The svn version of BLFS documentation lists Xorg Utilities as a required > dependency of Xorg Protocol Headers (proto.html). Should that be? > > I thought dependencies indicate packages necessary to compile and the > Xorg packages are to be compiled in order. However, Xorg Utilities is > listed after Protocol Headers and I can successfully compile X in that > order. > This is a mistake in the book that I overlooked. It was actually brought to my attention, however, I misunderstood the conversation thinking it was to do with splitting the utils packages. The util-macros package is looked for by everything, but does not cause a build error (or any negative side affect beyond a warning message and the README being excluded from the build tree). Fortunately, I do already have it covered (inadvertently) in one of my local copies with the split of macros and makedepend, but I haven't committed just yet. Hopefully tonight. If not, I can jump in and at least fix the misplaced dependency. Thanks for the report. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: icedtea on amd64
On 02/28/2011 06:59 PM, Nathan Coulson wrote: > > It has always been my dream to build java w/o the binary part. The "binary" icedtea6 is just a kit I compile and test myself, and then tar'd up for use in bootstrapping, there are no closed source parts left. There are instructions are in the wiki still to use gcj if you really want to. It needs some minor modifications (symlinks will all have to be moved to valid targets) as RedHat has changed some things in the gcj-compat package, but it is certainly doable (the various distro maintainers build it constantly this way). The external packages (xerces, xalan, rhino, and ecj) are used only for the bootstrap compiler, and with an icedtea-icedtea build, are actually not required, but I haven't taken the time to find a clean way to remove them from the stage1 target yet - need to write test cases for configure to see if the host JDK is capable of building the target without those classes and I'm honestly not sure what exactly is required that isn't in gnu classpath. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: udev in blfs
On 02/19/2011 01:30 PM, DJ Lucas wrote: > Then you do not have gio support in udev. err...gudev -- DJ -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: udev in blfs
0272-libgnomecups-0.2.3 0131-gtk+-2.22.0 0273-libgnomeprint-2.18.8 0131-gtk+-2.22.1 0274-libgnomeprintui-2.18.6 0132-libIDL-0.8.14 0275-telepathy-glib-0.12.7 0133-libnotify-0.4.5 0276-gssdp-0.9.1 0134-startup-notification-0.9 0277-gupnp-0.15.0 0135-nspr-4.8.60278-json-glib-0.12.2 0136-nss-3.12.80279-clutter-1.4.2 0137-curl-7.21.3 0280-clutter-gtk-0.10.8 0138-xulrunner-1.9.2.130281-evolution-2.32.1 0139-firefox-3.6.130282-nautilus-sendto-2.32.0 0140-thunderbird-3.1.7 0283-system-tools-backends-2.10.1 0141-pciutils-3.1.70284-liboobs-2.32.0 0142-libusb-1.0.8 0285-gmime-2.4.21 0143-libusb-compat-0.1.3 0286-totem-pl-parser-2.32.1 0143-usbutils-0.86 0287-alacarte-0.13.2 0144-usbutils-0.86 0288-cdrtools-3.00 0145-udev-165 0289-dvd+rw-tools-7.1 0146-docbook-xsl-1.76.10290-brasero-2.32.1 0147-eggdbus-0.6 0291-libelf-0.8.13 0148-docbook-xml-4.5 0292-bug-buddy-2.32.0 0149-polkit-0.99 0293-cheese-2.32.0 0150-sg3_utils-1.300294-gnome-media-2.32.0 0151-device-mapper-2.2.02.79 0295-lcms-1.19 0152-parted-2.30296-poppler-0.16.0 0153-libatasmart-0.17 0297-evince-2.32.0 0153-udisks-1.0.2 0298-libdvdcss-1.2.10 0154-udisks-1.0.2 0299-flac-1.2.1 0155-gvfs-1.6.60300-vcdimager-0.7.23 0155-vte-0.26.20301-taglib-1.6.3 0156-gvfs-1.6.60302-gst-plugins-good-0.10.25 0157-polkit-gnome-0.99 0303-gst-plugins-bad-0.10.20 0158-desktop-file-utils-0.17 0304-gst-plugins-ugly-0.10.16 0159-shared-mime-info-0.71 0305-totem-2.32.0 0160-icon-naming-utils-0.8.90 0306-gnome-python-desktop-2.32.0 0161-gnome-icon-theme-2.30.3 0307-deskbar-applet-2.32.0 0162-ConsoleKit-0.4.3 0308-libnice-0.0.13 0163-eject-2.1.5 0309-gst-python-0.10.20 0164-apache-ant-1.8.2 0310-farsight2-0.0.22 0165-cpio-2.11 0311-telepathy-farsight-0.0.15 0166-giflib-4.1.6 0312-telepathy-butterfly-0.5.14 0167-git-1.7.3 0313-telepathy-gabble-0.10.4 0168-ORBit2-2.14.190314-telepathy-idle-0.1.7 0169-popt-1.16 0315-gtkspell-2.0.16 0170-libbonobo-2.32.0 0316-xscreensaver-5.12 0171-GConf-2.32.0 0317-pidgin-2.7.9 0172-gnome-mime-data-2.18.00318-telepathy-haze-0.4.0 0173-samba-3.5.6 0319-telepathy-salut-0.4.0 0174-hal-0.5.140320-sofia-sip-1.12.10 0175-openssh-5.6p1 0321-telepathy-sofiasip-0.7.0 0176-gnome-vfs-2.24.4 0322-eog-2.32.1 0177-audiofile-0.2.7 0323-zip-4.0.4 0178-esound-0.2.41 0324-file-roller-2.32.1 0179-libgnome-2.32.0 0325-gcalctool-5.32.1 0179-libogg-1.2.2 0326-gconf-editor-2.32.0 0180-libvorbis-1.3.2 0327-gedit-2.30.4 0181-gstreamer-0.10.30.2 0328-gnome-power-manager-2.32.0 0182-libcanberra-0.26 0329-gnome-screensaver-2.30.2 0183-libgnome-2.32.0 0330-libdvdread-0.9.7 0184-libart_lgpl-2.3.210331-mplayer-20110116 0185-libglade-2.6.40332-gnome-mplayer-1.0.0 0186-libgnomecanvas-2.30.2 0333-boost-1.45.0 0187-libbonoboui-2.24.40334-libebml-1.0.0 0188-libgnome-keyring-2.32.0 0335-libmatroska-1.0.0 0189-libgnomeui-2.24.4 0336-mkvtoolnix-4.2.0 0190-gnome-keyring-2.32.1 0336-mkvtoolnix-4.4.0 0191-aspell-0.60.6 0337-gecko-mediaplayer-1.0.0 0192-enchant-1.6.0 0338-ntfs-3g-2010.10.2 0193-liboil-0.3.17 0339-apr-1.4.2 0194-cdparanoia-III-10.2 0340-apr-util-1.3.10 0195-nasm-2.09.04 0340-apr-util-1.3.9 0196-xvidcore-1.3.0-rc10341-neon-0.29.5 0197-yasm-1.1.00342-subversion-1.6.15 0198-libvpx-v0.9.5 0343-talloc-2.0.1 0199-libtheora-1.1.1 0344-libmng-1.0.10 0200-lame-3.98.4 I hope this list is useful for those plowing ahead. As far as gstreamer, I only install it for gnome's use (whatever that may be). Most of my A/V related tasks are handed to Mplayer/Mencoder. If anybody needs installation instructions, I would be happy to post my tools in my home dir if needs be. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: udev in blfs
On 02/19/2011 12:07 PM, cliffhan...@gardener.com wrote: >> -Original Message- >> From: John Burrell >> To: blfs-support@linuxfromscratch.org >> Sent: Sat, Feb 19, 2011 4:59 pm >> Subject: udev in blfs > >> Current version of udev in lfs-development is 166. > >> In blfs, if gobject-introspection installed, then udev-164 compiles > okay, >> but versions 165 and 166 do not. > >> Error is: > >> /usr/src/core/udev/udev-166/tmp-introspectY1yvnr/GUdev-1.0.o:(.data+0xc) > : undefined reference to `g_udev_enumerator_get_type' >> collect2: ld returned 1 exit status > > >> If use --disable-introspection, then both 165 and 166 compile without > error. >> I have gobject-introspection-0.6.14 installed. >> I haven't tried messing with later versions of gobject-intro. > >> May be the book should say - use 164 if introspection installed or > disable it with later versions - >> but then I don't know what introspection support udev provides, so > disabling it may mean the end >> of the world for some people ;>) > > I haven't seen this error as I normally compile gobject-introspection > after udev, much later, for whatever reason(can't remember). > Then you do not have gio support in udev. gobject-introspection will be held at current version until we update Gnome. If using lfs-svn or RC, you should try to update to latest stable releases available. Gnome in particular (as well as KDE) are difficult to nail down the dependencies, but not impossible, just pay close attention to the output of configure (use something like "./configure 2>&1 | tee myconfigure.log" so that you can look at it with a text editor). I'm thinking that the book should be branched for 6.7 to avoid these issues, but that is simply not possible right now. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: XKB Failed to compile keymap
On 02/09/2011 04:43 PM, Ken Moffat wrote: > If building lots of drivers doesn't help, how about > xkeyboard-config as suggested in > http://linuxfromscratch.org/pipermail/blfs-dev/2011-January/020953.html > ? > > ĸen Ken, did this change fix your XKB layout issue? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: gnome-2.32 packages (extremely long!):wq
On 02/06/2011 07:33 PM, Ken Moffat wrote:
Lots of snippage
gtk+-2.22.1
For everyone else - doesn't build out of the box, at least if you
have the docs tools, and if you build firefox against it you'd best
turn on text as well as icons in the toolbar. I have some adapted
seds from gentoo for the first part, and an unadapted patch from them
for the second (i.e. it needs autofoo). Of course, not everyone
needs the back and forward icons in firefox.
Yes, I had forgotten about the legacy icon names. Sorry, I wish I had
mentioned that earlier to save you some research time. Patch I used is
attached, no need for autofoo.
GConf-2.32.0 : note that 2.32.1 has "fixes" re gtk+-3 and doesn't
build if only gtk+-2 is present. Perhaps there is a switch to tell
it not to be silly, I didn't bothere looking.
gvfs-1.6.6 : I tried 1.7.1, but that needs glib>= 2.27.4.
GConf will soon die! :-) Or at least that was planned, hopefully things
haven't changed.
libproxy-0.3.0 - not part of gnome. I started by trying 0.4.6,
but in best google fashion that needs ./autogen.sh instead of
configure. Nastily, that then tries to invoke cmake, so I fed it to
/dev/null and reverted to an older version. There is apparently a
0.3.1 version, but I see no reason to try that in my own usage.
Can simply rerun cmake with appropriate options. As this is the only
thing on my system that uses it, no biggie. I'll not be happy to see it
used more, but I doubt I can change that trend.
webkit-1.2.6
As dj noted, for introspection-0.10 it needs an altered sed (to
1.2) PLUS a patch - google found a version at lunar-linux (for 1.2.5
- such an old and vulnerable version!) which works for me.
I didn't try the 1.3 series this time because it is still labelled
as a -development version (it will be needed for newer epiphany /
yelp), but the following addition to configure fixes the password
storage and retrieval : CFLAGS="${CFLAGS} \
-DLIBSOUP_I_HAVE_READ_BUG_594377_AND_KNOW_SOUP_PASSWORD_MANAGER_MIGHT_GO_AWAY"
Interesting. I don't see the issue as I use firefox for web browsing and
I don't save my passwords anyway.
boost_1_44_0 and gnash-0.8.8 :
- it looks as if youtube (swf) is no longer working for me. My
gut instinct is that something has been changed there *again*.
Fortunately, the html5 versions (epiphany, also midori) are working
fine. For those, I note that I had to (re) register at
http://www.youtube.com/html5 for epiphany, but midori then worked
without further intervention - last time I checked, I thought I'd
had to register *both* of these for html5.
Share and enjoy!
I actually depend on MPlayer for almost everything media related now,
gnome-mplayer, and then gecko media player for web content. Mplayer
"just works" in my experience, and "just works" is much easier than
dealing with GStreamer when some random plugin screws up the whole stack
on me.
Anyway, thanks for the write-up. Looks like we both stopped at about the
same place, though I think I might have loaded mine down a little
heavier, I'm sitting right around 370 packages installed (including
those in LFS) to get a mostly functional desktop. I am still having some
issues with file associations in the shell, and an issue with
Thunderbird where filters only work on the first IMAP account created.
There are a few packages that I don't build using my PM so I don't have
an exact number (which says 344 currently - java stuff, perl modules (I
use CPAN shell), anything installed singly in an /opt/whatever prefix
like QT, IcedTea, LibreOffice, etc.).
-- DJ Lucas
--- gtk+-2.22.0-orig/gtk/Makefile.in2010-12-31 02:42:51.0 -0600
+++ gtk+-2.22.0/gtk/Makefile.in 2010-12-31 02:46:42.0 -0600
@@ -1598,7 +1598,171 @@
stock-icons/24/folder-remote.png\
stock-icons/24/user-home.png\
stock-icons/24/user-desktop.png \
- stock-icons/24/text-x-generic.png
+ stock-icons/24/text-x-generic.png \
+ stock-icons/16/gtk-quit.png \
+ stock-icons/16/gtk-info.png \
+ stock-icons/16/gtk-file.png \
+ stock-icons/16/gtk-open.png \
+ stock-icons/16/gtk-print-preview.png\
+ stock-icons/16/gtk-print.png\
+ stock-icons/16/gtk-properties.png \
+ stock-icons/16/gtk-revert-to-saved-ltr.png \
+ stock-icons/16/gtk-revert-to-saved-rtl.png \
+ stock-icons/16/gtk-save-as.png \
+ stock-icons/16/gtk-new.png \
+ stock-icons/16/gtk-harddisk.png \
+ stock-icons/16/gtk-clear.png\
+ stock-icons/16/gtk-copy.png \
+ stock-icons/1
Re: xorg-7.6 (long, boring, list of packages)
On 01/27/2011 06:57 PM, Ken Moffat wrote: > > whoops, not used to being Cc'd on lfs mails, I hit reply instead of > reply-to-all. forwarding to the list. > > On Thu, Jan 27, 2011 at 09:53:37PM +, luxInteg wrote: >> On Thursday 13 January 2011 01:24:14 Ken Moffat wrote: >>> (this is what I build after my 'docbook' script which includes >>> Python (last of the 2.6-series), libxml2 (2.7.8 plus patch) and >>> xmlto. In this order: >> >> Can Python-2.7.1 be used with xorg-7.6? >> or must the 2.6 series be used? > > As Randy said. I've used 2.7.1 with xorg-7.5, and I'll be using it > in my own development (gnome-2.32) build (so far, that hasn't gone > very far beyond xorg). > > ĸen Same minor issues. I actually moved to gobject introspection 1.0 instead of 0.9.8, webkit-gtk will require a patch for syntax changes in the gir file for goi-1.0, and the sed is still required 's/1.0/1.2'. Bugged in webkit-gtk tracking and I have a patch locally, let me know if you hit it, but I suspect it'll be fine on 0.9.8. I did run into one other issue using gio-1.0, but it totally slips my mind ATM. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: icedtea on amd64
On 01/28/2011 05:28 AM, luxInteg wrote: > Greetings > > I want to have a go at building openjdk using icedtea. > http://www.linuxfromscratch.org/blfs/view/cvs/general/icedtea6.html > I notice that a binary is needed presumeably for so-called 'bootstraping' > and the one available from blfs is for 32-bit pentium cpu's. I am dabbling > with 64-bit amdK(8+} cpu's. > > Does anyone know of a 'bootstrapping' icetea binary for amd64's OR > whether such is redundant and if so how to compile icetea without the > prebuilt binary.? > > thanks in advance > > luxInteg > > I just built and put up binary packages on anduin a couple of days ago for both archs. It is possible to build icedtea6 without the binary version, but your configure flags will be a lot longer (and you can drop the upstream patch). You'll need to simulate a full JDK Home environment via symlinks. FYI, The binaries that are placed on anduin are debug images (IOW, I have not stripped libjvm.so in client or server VMs). Currently the book does not mention this, but I'll update the book in a few moments anyway, and add a note later. Stipped libjvm.so client goes from about ~70MB to ~4MB on x86 with the 1.8.x (I haven't bothered with 1.9.x). There is no client VM in x86_64 (makes a huge difference in size between x86 and x86_64). Both contain server, and that too can be stripped with similar results. Give me about fifteen minutes, up and rebuild your local copy (else it'll be in the book after tonight's render job). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Authorization example
On 01/20/2011 03:58 PM, Dr.-Ing. Edgar Alwers wrote: > Hi, > I finally got System Settings to accept me. Now, as an example and exercise, > I am trying to set something with the time, e.g. "set date and time > automatically". However, I get now this message: > quote > > Unable to authenticate/execute the action: 7, DBus Backend error: could not > contact the helper. Connection error: Could not get owner of name > 'org.kde.kcontrol.kcmclock': no such name. Message error: The name > org.kde.kcontrol.kcmclock was not provided by any .service files > > unquote > > It is not a problem, if I cannot set the time, but I would like to learn how > to do it. > Could somebody who knows about tell me: > 1.) what is the syntax of "org.kde.kcontrol.kcmclock" or where can I get > infos about ? > 2.) where do I have to place this file ? /usr/share/PolicyKit/policy ? > > Thank you very much for contributions ! > Edgar That is a dbus configuration error, and the solution depends on where you've installed KDE. /opt/KDE, /usr, or elsewhere. If anywhere other than /usr for the prefix, you'll need to add an entry into the /etc/dbus-1/system-local.conf file, and also if the sysconfdir is not /etc (for instance /etc/kde), you'll need to add any config directories like the following example for gnome installed into a prefix of /opt/gnome with a sysconfdir of /etc/gnome: dj [ BOOK ]$ cat /etc/dbus-1/system-local.conf http://www.freedesktop.org/standards/dbus/1.0/busconfig.dtd";> /etc/gnome/dbus-1/system.d /opt/gnome/share/dbus-1/system-services dj [ BOOK ]$ Also, make sure that $KDE_PREFIX/share/dbus-1/system-services/org.kde.kcontrol.kcmclock actually exists, or find it as I may have the path incorrect or it could be session-local or else wise (in fact, it may be more appropriate for session, not system, don't know as I'm typing my response blind as it's just general info). Similar entries are required for /etc/dbus/session-local.conf. Thomas is intending to add KDE-4 to the book, so hopefully all of this will all be spelled out soon. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: xorg-server-1.8.2 and pkg-config
On 01/20/2011 03:47 AM, Göran Boström wrote: > I'm following the instructions in the BLFS-book: > the PKG_CONFIG_PATH environment variable is not set. What is the value of $XORG_PREFIX? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: git install and /usr/libexec directory
On 01/17/2011 01:17 AM, Simon Geard wrote: > On Sun, 2011-01-16 at 07:44 -0600, Randy McMurchy wrote: >> Juan Antonio Moreno Carmona wrote these words on 01/16/11 05:45 CST: >>> git wants to put gitcore in /usr/libexec and, until now, no other package >>> installed files there. That directory does not exits and I think is against >>> FHS. What would be the most appropriate destination directory for that file? >> >> BLFS has established convention that libexec files go to >> /usr/lib/package_name. > > I used to do that, but switched back go /usr/libexec after hitting > packages that assumed that things it depended on could be found in it's > own libexec dir. > > From memory, it was GDM, looking for various other pieces of Gnome... > I'm sure the problem was solvable, but it was easier to simply remove > --libexecdir from my scripts and use the default location. > > Simon. > GDM is actually a corner case, and doing something it really shouldn't do, but it's easy to work around. The maintainers are aware of the problem, but patches have not been accepted yet to gnome-settings-daemon that the GDM patch depends on. One of the GDM maintainers (the?) actually got the Console-Kit change committed for us, so in the interim, rather than patching things out badly, I've elected to use a shared libexec dir for only gnome specific packages. In fact, exactly like Xorg, I use an environment variable for configuring gnome packages (/etc/gnome and /usr/lib/gnome are symlinks to a versioned directory): dj [ ntfs-3g-2010.10.2 ]$ cat /etc/profile.d/gnome.sh GNOME_PREFIX="/usr" GNOME_CONFIG="--prefix=/usr --sysconfdir=/etc/gnome --libexecdir=/usr/lib/gnome --mandir=/usr/share/gnome --infodir=/usr/share/info" pathappend /etc/gnome/xdg XDG_CONFIG_DIRS export GNOME_PREFIX GNOME_CONFIG XDG_CONFIG_DIRS Pretty much all packages use "./configure $GNOME_CONFIG" with a couple using a one off, and only patch I use is for GDM (to find the CK libexecdir for ck-connector). CK must also be patched for now to supply libexecdir in ck-connector.pc. https://bugs.freedesktop.org/show_bug.cgi?id=18427 has links to both bugs and both patches, though both patches are in the repo, and might even be in the book. I haven't looked recently, but at least the CK patch and GDM-CK patch should be in the book as opposed to the symlinks. The GDM-GSD patch becomes unnecessary if using the shared libexecdir (as with a few other minor annoyances). -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: xorg-7.6 (long, boring, list of packages)
On 01/12/2011 07:24 PM, Ken Moffat wrote: > (this is what I build after my 'docbook' script which includes > Python (last of the 2.6-series), This gets even uglier, one of the updated packages I downloaded on my daily driver accidentally was fortunately (or I wouldn't have caught my error) already looking for Python-3 and refused to build against Python-2.7.1. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: SeaMonkey-2.0 compilation failure
On 01/06/2011 06:28 AM, Thanos Baloukas wrote: > # --with-system-nspr is used by configure (not client.mk) > # --with-system-nss is used by configure (not client.mk) > c++ -o nsDocument.o -c -I../../../dist/include/system_wrappers -include > /home/thanos/build/xc/comm-1.9.1/mozilla/config/gcc_hidden.h > -DMOZILLA_INTERNAL_API -DMOZ_SUITE=1 -DOSTYPE=\"Linux2.6.35\" -DOSARCH=Linux > -D_IMPL_NS_LAYOUT > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src -I. > -I../../../dist/include/xpcom -I../../../dist/include/string > -I../../../dist/include/dom -I../../../dist/include/gfx > -I../../../dist/include/thebes -I../../../dist/include/layout > -I../../../dist/include/widget -I../../../dist/include/view > -I../../../dist/include/locale -I../../../dist/include/htmlparser > -I../../../dist/include/js -I../../../dist/include/webshell > -I../../../dist/include/necko -I../../../dist/include/nkcache > -I../../../dist/include/mimetype -I../../../dist/include/exthandler > -I../../../d > ist/include/chardet -I../../../dist/include/caps > -I../../../dist/include/lwbrk -I../../../dist/include/uconv > -I../../../dist/include/docshell -I../../../dist/include/pref > -I../../../dist/inc > lude/xpconnect -I../../../dist/include/unicharutil > -I../../../dist/include/intl -I../../../dist/include/webbrowserpersist > -I../../../dist/include/imglib2 -I../../../dist/include/plugin > -I../../../dist/include/prefetch -I../../../dist/include/xuldoc > -I../../../dist/include/uriloader -I../../../dist/include/rdf > -I../../../dist/include/xultmpl -I../../../dist/include/util > -I../../../dist/include/appshell -I../../../dist/include/shistory > -I../../../dist/include/editor -I../../../dist/include/windowwatcher > -I../../../dist/include -I../../../dist/include/content -I/usr/include/nspr >-I/home/thanos/build/xc/seamonkey-build/mozilla/dist/sdk/include > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../events/src > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../xml > /content/src > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../xul/base/src > > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../xul/content/src > -I/home/thanos/b > uild/xc/comm-1.9.1/mozilla/content/base/src/../../html/content/src > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../base/src > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../xbl/src > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../../layout/generic > > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../../layout/style > > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../../dom/src/base > > -I/home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/../../xml/document/src > -I/home/thanos/build/xc/comm-1.9.1/mozilla/xpcom/io -fPIC -fno-rtti > -fno-exceptions -Wall -Wpointer-arith -Woverloaded-virtual -Wsynth > -Wno-ctor-dtor-privacy -Wno-non-virtual-dtor -Wcast-align > -Wno-invalid-offsetof -Wno-long-long -pedantic -fno-s > trict-aliasing -fshort-wchar -pthread -pipe -DNDEBUG -DTRIMMED -Os > -freorder-blocks -fno-reorder-functions-DMOZILLA_CLIENT -include > ../../../mozilla-config.h -Wp,-MD,.deps/nsDocument.pp > /home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/nsDocument.cpp > /home/thanos/build/xc/comm-1.9.1/mozilla/content/base/src/nsDocument.cpp:167:28: > fatal error: nsIXULDocument.h: No such file or directory > compilation terminated. I would venture a guess that you'll have to use the in-tree nss and nspr as the system versions are probably much newer. Possibly a newer version of Seamonkey is available? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Icedtea6 results
On 01/04/2011 12:20 PM, John Burrell wrote: > I built Icedtea6-1.9 on a 32 bit system with lfs-SVN-20101118 with kernel > 2.6.36.1, so gcc is 4.5.1. > > Xorg is version 7.5. All the components (proto, util, lib, app, font and > driver) > are all 7.5.3. Also installed gtk+-2.20.1, gnome-2.30.2 with > metacity-2.30.1. Cool. Thanks for the quick reply. That narrows things down quite a bit, save for the base LFS version, has nearly same add-ons as the one those instructions were developed on. > > I also have rpm installed and used it to load openoffice-3..2.1, so > there are 2 javas > installed - JDK-update 22 and the one from the openoffice rpm, but I > don't think > that should cause a problem - at least it hasn't up to now. > > I think some of the failures were due to the unprivileged user but I > don't fancy running > the tests as root as it'll probably change my setup in an unpredictable > way. > I use Matthias > Benkmann's package user management system so none of my packages is > installed > as root. > Actually, the tests must be run by the user who started the xserver. If not, then pretty much everything that requires X will fail unless you've given the package user permissions to use the display (which I don't even have the slightest clue how to do now days). Were the tests run as the package user? We could possibly be looking at two separate issues with similar results. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Icedtea6 results
On 01/02/2011 10:35 PM, John Burrell wrote: > Compiling Icedtea6 from source assumes /usr/bin/find, /usr/bin/head and > /bin/touch. > With standard lfs system, need symlinks to /bin/find, /bin/head and > /usr/bin/touch > for it to work. > > Also need a symlink pointing /usr/include/xulrunner-1.9.2.12/stable to > /usr/include/xulrunner-1..9.2.12 in order to find npapi.h > > Running the tests gave me the following results - > > Test results: passed: 3,113; failed: 188; error: 2 > > Nearly all java/awt tests failed, some java/beans tests and most > javax/swing tests. > Also, most sun/java2d tests failed. > > Don't know if this is a major issue but thought I should report it as > the book says all tests > should pass. So must assume that the compiled version is not reliable > and so will use the > binary version. > > Can supply you the test directory or the logs if requested. > > regards > > jb. > Thanks for the report John. What version did you build? 1.9.0 from the book? Also, there is a reason that xulrunner was omitted from the dependencies (there are issues yet and upstream is aware of them...causes an additional ~35 test failures in jtreg). I had forgotten about the hard coded paths (I had symlinks left over from earlier test builds). As far as the tons of awt and swing failures, I just noticed these as well on a recent build of 1.9.3 on a 64bit LFS (Saturday). I take it you are on 32bit? Those instructions were written prior to lfs-6.7. The host had gcc-4.4.3, and gtk+-2.20.1 with gnome-2.30.1. System was completed roughly June 10th of 2010. Working on it now (as well as a lot of other stuff). I was previously prepared to blame the failures on openbox or new xorg as I noticed that the mouse cursor was not moving as expected (not at all in fact) for most of the awt tests. Can you tell me a little about your environment? LFS version, Xorg version, GTK Version, Window Manager in use, etc.? Thanks in advance. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: glibc vulnerability, CVE-2010-3847
On 12/12/2010 06:57 AM, bendeguz wrote: > > I have glibc-2.11.1 and it's not been patched, yet. > It's just a desktop computer, but I'm curious about the issue. I should learn to read to the end. Sorry for my previous message. Please ignore. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: glibc vulnerability, CVE-2010-3847
On 12/12/2010 06:57 AM, bendeguz wrote: > Hi! > > I tried the exploits on the lfs-dev mailing list and the > exploit here http://seclists.org/fulldisclosure/2010/Oct/257, > but something is wrong. It doesn't work as "expected". > > In the latter case: > > $ LD_AUDIT="\$ORIGIN" exec /proc/self/fd/3 > Inconsistency detected by ld.so: dl-open.c: 231: dl_open_worker: Assertion > '(call_map)->l_name[0] == '\0'' failed! > > then, from console, it drops me to login prompt > I guess I don't understand the test case. What was the expectation? 3 == $ppid (in this case exec (a bash bultin)). Bash isn't setuid and is not compiled with -rpath so I don't see the concern. I'm not sure why it killed the parent. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: expat / gimp problem
On 12/05/2010 04:45 PM, Ken Moffat wrote: > A few days ago, I updated my patch to fix the current expat > vulnerabilities, tested that my normal desktop seemed to work, and > committed it (part of the preparations for ghostscript-9.00). Along > the way, I noticed that my previous patch had fallen out of my > script. Since then I've made several other upgrades to my desktops > (particularly freetype, but also xorg-server and the ati driver, to > see if my "xorg sometimes crashes on radeon r100/r200" problem still > exists - so far, it's too early to say). > > Then I took a break from gs9 to catch up with editing my photos. > I'm using the gimp, and my pics need a *lot* of adjustment, partly > because my poor balance and taking most one-handed means my idea of > "which way is up" is only approximate. So, lots of use for the > 'rotate' and 'perspective' tools. Unfortunately, the perspective > tool had become all-but unusable (either the preview didn't show > what changed, or else I couldn't get the controls on the corner of > the image to move). After a lot of experimentation, and getting > ready to revert everything I had changed, I've now got the control > working adequately. The problem is the expat patch. Unfortunately, > that patch ought to be used, because it fixes known vulnerabilities. > > For now, this is just a heads-up that it seems to break the > perspective control in the gimp. It's taken me 3 days to pin the > blame here, I'm still not entirely convinced (for one thing, > rendering of the gimp windows was horribly slow, and now that I've > reverted this patch it still seems slow). Comments welcome. > Unfortunately, while you've given lots of detailed information, there still isn't much to point a finger at. A couple of stabs in the dark is all that I have to offer. It's been a while since I've used Gimp, so I don't know what is in in ~/.gimp, but did you move it out of the way and start fresh? Would seem logical (at least to me) that some items in there would use xml and you messed with expat. First instinct and all...sounds plausible at least. Also, are the slowdowns in any way similar to the slow resize issues you had before with xorg-server-1.9.X? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: ATI-Radeon 4350 display problem
On 12/03/2010 08:43 AM, Mark Mark wrote: > > Any way, when I erased the config file , it works very good, but the > only difference was: > - Previously I was able to write kdm only to enter to kde from console . > - Now: I have to xinit , and kdeinit ,and then startkde > Another case of incorrect xorg.conf getting in the way when autodetect would work. Auto-detect is the way to go from now on. :-) Glad that is fixed. As to kde/kdm, haven't used it in a while, but I'm in the process of building up to Trinity. Best guess is to recreate your ~/.xinitrc (follow the instructions you had used before) and I'd imagine at least startx would just work. I honestly don't know how KDM works. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: ATI-Radeon 4350 display problem
On 12/01/2010 09:18 AM, Mark Mark wrote: > Dear All > > I have installed LFS6.6 , with Xorg- Xserver 1.8.2. > > I had solved the intel VGA problem which I had first , but now when I > installed a new ATI-Radeon 435x card , I compiled a new kernel-2.6.34.2 > , but the screen becomes black, but I still can press Alt+Ctrl+F1 to go > to another session , the PC is still responsive. > > Note: I tried the configuration with and without Radeon_KMS. > > The problem is also that no error exists in the /var/log/Xorg.0.log, > unless I press ctrl+c , I got segmentation fault. > > Kindly find attached the log ,and the xorg.conf.new > Oops...should have read 'xrandr -s 2 -d 0:0' but that won't work anyway. What about the others? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: ATI-Radeon 4350 display problem
On 12/01/2010 09:18 AM, Mark Mark wrote: > Dear All > > I have installed LFS6.6 , with Xorg- Xserver 1.8.2. > > I had solved the intel VGA problem which I had first , but now when I > installed a new ATI-Radeon 435x card , I compiled a new kernel-2.6.34.2 > , but the screen becomes black, but I still can press Alt+Ctrl+F1 to go > to another session , the PC is still responsive. > > Note: I tried the configuration with and without Radeon_KMS. > > The problem is also that no error exists in the /var/log/Xorg.0.log, > unless I press ctrl+c , I got segmentation fault. > > Kindly find attached the log ,and the xorg.conf.new > Nice when there is nothing to go on, right? So, lots of stuff to look at. Here are a couple from the hip. First, get rid of /etc/X11/xorg.conf, and ~/.xinitrc and see what happens if you run startx without any configuration or switches at all. If that doesn't work, while the blank screen is on 7, try from another VT with the same user 'xrander -s 2'. Failing still, what does your /etc/X11/app-defaults/xinitrc contain? Do you have twm, xclock, and xterm installed? -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: OpenOffice 3.2.1 on x86_64
On 11/26/2010 09:53 PM, Andre Keller wrote: > On 27.11.2010 04:08, schrieb DJ Lucas: >> I'd try and use --with-system-db in the OOo build. > > Yeah I thought about this too, there are other tools like libxml2, > libxmlsec, mysql, mozilla, etc.) which could be taken from the system > instead of building them along with the oo build. > > I just wanted to do an initial build first, to see if it would builds at > all. Now I can try and optimize my build. > >> That's why I want the >> BLFS profile upstreamed so that all of those options are set in stone > > That happening would be great! > I've put the latest in my home dir if you'd like to check out what I've already done. Been a few months since I've messed with it. What you have now should work fine, however. I know very well how much work and time it takes to build OOo. :-) http://www.linuxfromscratch.org/~dj/ooo-build-3.2.1.4-BLFS_Config-1.patch http://www.linuxfromscratch.org/~dj/ooo-build-3.2.1.4-db5_configure_check-1.patch Keep in mind that this will not apply directly to current Libre (and I'm not even sure if anything will be applicable to Libre as I haven't checked it out since Go-oo -> Libre). Here are my OOo build-opts (included in the blfs build profile): --enable-cups --enable-cairo --enable-opengl --enable-gnome-vfs --enable-gstreamer --enable-gtk --disable-odk --disable-binfilter --disable-epm --disable-fontooo --disable-kde --disable-kde4 --disable-qadevooo --with-ant-home=/opt/ant --with-jdk-home=$JAVA_HOME --without-fonts --without-openldap --without-system-jars --without-system-mozilla --without-system-poppler --with-system-boost --with-system-cairo --with-system-curl --with-system-expat --with-system-freetype --with-system-icu --with-system-jpeg --with-system-libxml --with-system-libxslt --with-system-neon --with-system-odbc-headers --with-system-openssl --with-system-python --with-system-stdlibs --with-system-xrender-headers --with-system-zlib --enable-minimizer --enable-ogltrans --enable-presenter-console --enable-pdfimport --enable-report-builder --enable-strip --enable-xrender-link --enable-wiki-publisher --with-about-bitmaps=\"$TOOLSDIR/src/openabout_go-oo.bmp\" --with-dict=\"ALL\" --with-extension-integration --with-intro-bitmaps=\"$TOOLSDIR/src/openintro_go-oo.bmp\" --with-lang=\"ALL\" --with-linker-hash-style=both --with-unix-wrapper --with-vendor=\"BLFS\" -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: glibc vulnerability, CVE-2010-3847
On 11/13/2010 06:38 AM, bendeguz wrote: > On Fri, Nov 12, 2010 at 09:50:08PM +0100, Aleksandar Kuktin wrote: >>> Do I have to recompile everything in minor glibc version changes as >>> well? >>> >>> regards >> >> You mean like 2.7->2.8 ? >> I would say 'absolutely, definitely yes'. >> >> For 2.7.1->2.7.2, I would recommend it. >> > Darn, I thought so... > And what if I apply a patch on glibc-2.11.1 and > rebuild it? Maybe it depends on the patch itself... Should work just fine. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: cat /proc/version reports wrong info.
On 10/19/2010 05:56 PM, cliffhan...@gardener.com wrote:
>
>> -Original Message-
>> From: Bruce Dubbs
>> To: BLFS Support List
>> Sent: Mon, Oct 18, 2010 10:59 pm
>> Subject: Re: cat /proc/version reports wrong info.
>
>> Nathan Coulson wrote:
>> On Mon, Oct 18, 2010 at 2:36 PM, wrote:
>>>> -Original Message-
>>>> From: Bruce Dubbs
>>>> To: BLFS Support List
>>>> Sent: Mon, Oct 18, 2010 5:16 pm
>>>> Subject: Re: cat /proc/version reports wrong info.
>>>> cliffhan...@gardener.com wrote:
>>>> Hi
>>>>
>>>> I've recently installed LFS6.7 using jhalfs. No probs. Building
>>> BLFS
>>>> now to own requirements,but I've discovered a problem with
>>>> /proc/version reporting wrong gcc version and compiled kernel.
>>>>
>>>> LFS 6.7 uses kernel 2.6.35.4
>>>>
>>>> 'cat /proc/version' gives:
>>>>
>>>> Linux version 2.6.27.10 (r...@cliffhanger) (gcc version 4.2.3) #1
>>>> PREEMPT Sun Jan 3 19:49:10 GMT 2010
>>>>
>>>> 'uname -r' gives:
>>>>
>>>> 2.6.27.10
>>>>
>>>> AFAIK 'uname' reads its info from /lib/modules/RELEASE, which is
>>>> 2.6.35.4
>>>> /proc/version is compiled into the kernel. It sounds like
>>>> you booted the wrong kernel.
>>> Thanks Bruce, but definitely not. I have two partitions, with the
>>> host(an older LFS)on one and LFS6.7 on the other. LFS6.7
>>> is booting the new kernel(2.6.35.4), but proc/version and uname both
>>> report it as the older kernel, 2.6.27.10.
>>>
>>> Have just reinstalled kernel to be sure, still reporting it wrongly.
>>>
>>>> LFS does not install a /lib/modules/RELEASE file. I've not heard of
>>>> that before.
>>> When I say RELEASE I mean the kernel version, i.e.
>>> /lib/modules/2.6.35.4. This is what 'uname' reads?
>>>
>>> thanks
>>>
>>> MAC
>>
>
>> Right. See 'man 2 uname'
>
>> "This is a system call, and the operating system presumably knows its
>> name, release and version. It also knows what hardware it runs on."
>
>> Actually it uses the uname call to figure out which modules to use,
> not
>> the other way around.
>
>> linux-fan wrote:
>
>> You wil see a line such as:
>> lfs kernel: Inspecting /boot/System.map-2.6.33
>>
>> ... and that 2.6.33 is the kernel version that booted.
>> Assuming that you named System.map-($uname -r)
>
>> nope, it is read right from the kernel's version string. Only way to
>> get that, is by running that version of kernel.
>
> Which I am
>
> I know I'm booting the 2.6.35.4 kernel, but it's not being reported as
> such. This is affecting the module loading on the new system e.g.
>
> modprobe: FATAL: Could not load /lib/modules/2.6.27.10/modules.dep: No
> such file or directory - again the wrong kernel
>
> Surely the big question, is how, on a brand new system with its first
> kernel, did the listing in /proc/version get to be 2.6.27.10? I can go
> to it and see this. It seems to have inherited the kernel version
> from the old lfs host system i used. But how? As I said I used
> jhalfs
> for the install, but was forced to install the new kernel manually.
> How do i change this proc file - I know by booting the correct kernel!!
> This is crazy
>
> thanks to all
um...as the root user:
cd "${SOURCES}" &&
tar -xf linux-2.6.35.4.tar.bz2 &&
cd linux-2.6.35.4 &&
cp /boot/config-2.6.35.4 .config &&
yes | make oldconfig &&
make &&
make modules_install &&
cp arch/i386/boot/bzImage /boot/lfs-kernel-2.6.35.4 &&
cp System.map /boot/System-2.6.35.4
;-)
-- DJ Lucas
--
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Re: parser error : Document is empty
On 10/19/2010 01:24 PM, Jason Klumker wrote: > Trying to build xorg via blfs-tool (is this the right list?) > Error follows: > > ret: 1 luit Using luit Xinc to solve xorg7../:1: parser error : > Document is empty > > ^ > ../:1: parser error : Start tag expected, '<' not found > > ^ > unable to parse ../ > make: *** [menuconfig] Error 6 > I'm not sure what the status is for blfs-tool. Don't know if anybody uses it here, I know I've never tried it, and rarely hear anything about it on alfs-discuss either. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: Chapter 23 - Xorg Utilities
On 10/13/2010 07:13 PM, Mike Hollis wrote: > I compiled Xorg-7.5 last night and got it working with a couple > adjustments. It was working Ok but deleted it today and started > over to see if these adjustments were due to some failure to follow > directions on my part. > > Chapter 23. X Window System Environment --Xorg Utilities > Contents > Installed Programs:ccmakedep, cleanlinks, gccmakedep, imake, lndir, > makedepend, makeg, >mergelib, mkdirhier,mkhtmlindex, revpath, xmkmf > > The only one of these I have in $XORG_PREFIX/bin is makedepend. Is this > a typo or should I have these now ? > This is a typo. The others come from the other utility packages that were dropped in more recent releases. see ftp://ftp.x.org/pub/individual/util/ for all of the old monolithic build tools from which those programs come. -- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/blfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
