Bug#171125: ITP: cdw -- A console tool for burning CD's CDW is an ncurses-base frontend for cdrecord and mkisofs.

[Robert Nagy]

Package: wnpp
Version: unavailable; reported 2002-11-29
Severity: wishlist

* Package name: cdw
  Version : 0.1.4
  Upstream Author : Balazs Varkonyi <[EMAIL PROTECTED]>
* URL : http://cdw.sourceforge.net
* License : GPL
  Description : A console tool for burning CD's.

 CDW is an ncurses-base frontend for cdrecord and mkisofs.
 This is the best choise for console users.
 it can handle audio and data CD burning, through a CD image or
 directly from the files.
 
 My apt lines are:
 deb http://ers.linuxforum.hu/debian/ ./
 deb-src http://ers.linuxforum.hu/debian/ ./

-- System Information:
Debian Release: testing/unstable
Architecture: i386
Kernel: Linux thuglife.homelinux.org 2.4.19 #19 Wed Nov 27 23:18:19 CET 2002 
i686
Locale: LANG=en_US, LC_CTYPE=en_US

Re: New maintainer behaviour with NMU and LogJam's hijacking

[martine]

On Fri, Nov 29, 2002 at 03:47:29AM +0100, Christian Surchi wrote:
> Maybe I forgot something... maybe this episode could be seen as not so
> important but it hit me strongly, in particular from the point of view
> of my idea about Our Project and Our Work.
> 
> I don't know if I'll take care further of logjam debian packages.

I'm the upstream author of LogJam.

When I finally released this most recent version (the one Ari packaged),
I sent the release announcement to everyone who has packaged LogJam in
the past, including Christian.  

However, due to Christian's backlog, LogJam in Debian has often been
behind a few versions.  The version currently in testing (3.0.3) is even
back a patch version; according to the datestamps on my files I released
3.0.4 on May first of this year, which makes that almost half a year
old.  (I am sorry to say I don't know the proper protocol in Debian for
this-- do I file a bug report on the package saying there's a new
version?)

I have a friend at Redhat who builds packages for me.  I am a devoted
Debian user and I was a little embarrassed that the RPMs I provided were
more recent than the debs.  I appreciate the problem of backlog.  I
myself am often backlogged in my mail.  So when I decided 4.0 (a port to
GTK2 with a lot of new code) was ready I did not try to circumvent
Christian-- for example, I didn't contacting another Debian maintainer--
and instead just built my own DEBs and provided them myself.  When Ari
saw I had done this, he told me he'd package it himself and so I removed
mine.

Though it is true the version he packaged was from CVS, he did so
because it includes bugfixes.  Ari watches my CVS commit log so I think
he knows what building a CVS version meant.


After everything's said and done, though:
If you cannot do the packaging (as you yourself have written), what harm
does it do for someone else to do it?  If it's too much work for you,
Ari has already demonstrated he has the free time...
Otherwise, once you're caught up on your mail, I'm sure you can continue
packaging LogJam.

Sorry to be a bother, everyone!  :)

-- 
  Evan Martin
[EMAIL PROTECTED]
  http://neugierig.org

Re: Pick a name, any name...

[Jérôme Marant]

Raphael Hertzog wrote:
Le Thu, Nov 28, 2002 at 01:57:16PM +0100, Jérôme Marant écrivait:
 

I hope it is for packaging activities and debian-related stuff only.
   

Not really. It is for any free software related project launched by
a Debian developer (ie the project request should come from a Debian
developer).
 

I hope it will be as reliable as SourceForge and Savannah in terms of
MTBF and backups. Losing a packaging work is less critical than
loosing an upstream one.
Of course, any project hosted there should really have Debian
packages from the very beginning.
 

Sure.
Cheers,

Re: Pick a name, any name...

[Russell Coker]

On Fri, 29 Nov 2002 04:39, John H. Robinson, IV wrote:
> Ben Armstrong wrote:
> > Bah, that's what CNAME is for.
>
> that is _NOT_ what a CNAME is for. a CNAME is for when the hostname is
> in a domain that is OUTSIDE of your control.
>
> ie: evil.debian.org -> www.msn.com = CNAME (we don't control the msn.com
> domain)
> forge.debian.org -> quantz.debian.org = A (we control the debian.org
> domain, so we can save the internet by REDUCING THE NUMBER OF
> UNNECESSARY DNS LOOKUPS AND REDUCE THE END USERS DELAY WITH DNS LOOKUP
> REQUIREMENTS)

How does that save the Internet?

DNS entries are cached and don't cause that much traffic.

Having CNAME entries pointing to your own A records is OK, just as having 
symbolic links pointing to your own files on the same file system is OK.

-- 
http://www.coker.com.au/selinux/   My NSA Security Enhanced Linux packages
http://www.coker.com.au/bonnie++/  Bonnie++ hard drive benchmark
http://www.coker.com.au/postal/Postal SMTP/POP benchmark
http://www.coker.com.au/~russell/  My home page

KDEE3 question

[Michael Meskes]

Hi,

I was just being told that we won't get any KDE3 packages until the
whole Debian project is moved to gcc 3.2. Is this correct? If so, why?
And is there a timetable? Is it a problem to compile the packages with
gcc 2.95?

The matter of the fact is that almost all Debian/KDE users are already using
ftp.kde.org to get the KDE packages. I would surely prefer to have these
packages as part of our archive instead of redirecting everyone away
from debian.org.

Thanks in advance for the explanations.

Michael
-- 
Michael Meskes
Michael@Fam-Meskes.De
Go SF 49ers! Go Rhein Fire!
Use Debian GNU/Linux! Use PostgreSQL!

Re: Are we losing users to Gentoo?

[Gerfried Fuchs]

* Joey Hess <[EMAIL PROTECTED]> [2002-11-28 17:13]:
> So maybe you click on the "Debian on CD" link, right? And from there on
> the 4th bulletted link ("Download CD images using HTTP or FTP"), after
> wading past unofficial minimal CD images, and learning what jigdo is.

 Because those options are prefered.  We still like to help the sensible
users instead of doing everything for the DAUs[1] who simply don't care,
don't we?

> And then on scroll way down the list to your country. And then into the
> current directory on the mirror, oops, that was jigdo only?!  back out
> and to the 3.0r0 directory.

 Uhm, just a second.  When I click a on the links in that list I get to
the debian-cd directory on the mirrors where I can choose between jigdo
and the 3.0r0 directory.  Why do you find yourself in the jigdo
directory?

> What, that was jigdo again?! Hmm, try another mirror.

 Now it would be _more_ than helpful which link you are refering to.
Without knowing it that seemingly broken link can't be fixed, thank you.

> Maybe the one in Austria, because it's the top of that list of
> mirrors. Hmm, no, it only has a jigdo directory too.

 Uhm, you seem to be blind, sorry.  Both ftp and http of the austrian
mirror has the 3.0r0 directories.  Can you please check before accusing?
I guess you seem to be puzzled by the layout of the HTML-Page there --
but you can't accuse debian for third party layouts, or do you try to do
so?

> Finally, by picking the FTP site (not the HTTP site) in Austria, and
> digging two more directories deep, you find an iso.

 "Digging" one directory, there are just two and it's the one that is
not named jigdo.

> But maybe instead, back at debian.org's front page, you picked the
> "Getting Debian" link instead. Only to end up on a page that links to cd
> vendors and "downloading over the Internet". Ok, the latter. But it
> points to a page that only lets one download unnofficial netinst iso
> images, which are of varying quality, and well, unnoficial.

 You are right, there should be added a link to the $(HOME)/CD/http-ftp/
on that page, too.  Thanks for the (quite hidden) suggestion.  Joy, do
you think that would help, too?

> (feel free to use me as one data-point; I have never used the debian
> website to try to download a debian CD before; indeed I have never
> downloaded a debian CD).

 But your first approach was correct, only that you seem to have been
confused by third party html layouts on which we don't have any
influence.

 So long,
Alfie

[1] Dumbest Assumable User
-- 
16:46  sorry fuer die andauernden rejoins
16:46 -!- Molle [EMAIL PROTECTED] has quit [leaving]


pgprm91hmbW2S.pgp
Description: PGP signature

Re: Pick a name, any name...

[Andrew Suffield]

On Wed, Nov 27, 2002 at 12:58:12PM +, Colin Watson wrote:
> On Wed, Nov 27, 2002 at 07:07:45AM -0500, H. S. Teoh wrote:
> > On Wed, Nov 27, 2002 at 11:03:58AM +0100, Roland Mas wrote:
> > > - Another idea I had was something along the lines of Debsmith or
> > >   Iansmith, to keep both the idea of Debian and the idea of the forge.
> > >   Unfortunately, plenty of people are called that way.  Any idea to
> > >   improve that line is welcome.  Maybe just smith.debian.org would be
> > >   enough (and I'll bet we can find a music composer named Smith if we
> > >   look hard enough).
> > 
> > Speaking of composers... how about Sibelius? Finnish composer, in honor of
> > Linus. :-) Or Debussy, if you're going for French names.
> 
> We've got a debussy; it's an arm machine.

And there's already a Sibelius as a commercial software project; stay
away from things that could raise trademark issues that way.

Will people *please* at least feed names to google and see if any
conflicts come up in the first few hits? A repeat of the sawfish
debacle would not be appreciated by anybody.

-- 
  .''`.  ** Debian GNU/Linux ** | Andrew Suffield
 : :' :  http://www.debian.org/ | Dept. of Computing,
 `. `'  | Imperial College,
   `- -><-  | London, UK

Re: KDEE3 question

[Tom Badran]

On Friday 29 Nov 2002 9:12 am, Michael Meskes wrote:
> Hi,
>
> I was just being told that we won't get any KDE3 packages until the
> whole Debian project is moved to gcc 3.2. Is this correct? If so, why?
> And is there a timetable? Is it a problem to compile the packages with
> gcc 2.95?
>
> The matter of the fact is that almost all Debian/KDE users are already
> using ftp.kde.org to get the KDE packages. I would surely prefer to have
> these packages as part of our archive instead of redirecting everyone away
> from debian.org.

The plan as i understand it is to supply gcc 3.2 built packages once kde 3.1 
is released, and never supply gcc 2.95 ones in debian

Tom

Re: Pick a name, any name...

[Hamish Moffatt]

On Fri, Nov 29, 2002 at 09:44:17AM +, Andrew Suffield wrote:
> And there's already a Sibelius as a commercial software project; stay
> away from things that could raise trademark issues that way.

I don't think the existence of a commercial product called Sibelius
prevents us from naming our machines the same way - especially when the
services those machines provide are generally accessed by other names
anyway.


Hamish
-- 
Hamish Moffatt VK3SB <[EMAIL PROTECTED]> <[EMAIL PROTECTED]>

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Andrew Suffield]

On Fri, Nov 29, 2002 at 03:47:29AM +0100, Christian Surchi wrote:
> On 17 Nov 2002 Ari made a NMU for logjam 4.0.0+cvs.2002.11.17 and
> another one a few days after that date, IIRC, without a note to me.
> I was handling bugs for logjam, as you can see in BTS (#165281). Build
> failure reported by Junichi Uekawa in that bug was actually a
> libcurl-dev bug (#169654). I reported and maintainer closed with an
> upload.
> 
> So Ari made the NMU only to close a not so old wishlist bug filed by
> himself, faking to close #165281 with his upload. No bug for it against
> logjam and instead he closed it with an unuseful "New upstream version"
> entry in changelog. I was not MIA and he didn't write any note to me
> about his proposal for an NMU. Then he changed to a cvs version, while I
> have always packaged stable released version. No notes about it too.
> Then he did not follow our guidelines for NMU, because he uploaded
> directly to incoming and not to the 7-DAYS delayed queue, so I couldn't
> stop his NMU.

I had more here originally, but it's all pretty obvious. I can
summarise with the aphorism:

Where NMUs are concerned, get it right. All *else* can be forgiven.

-- 
  .''`.  ** Debian GNU/Linux ** | Andrew Suffield
 : :' :  http://www.debian.org/ | Dept. of Computing,
 `. `'  | Imperial College,
   `- -><-  | London, UK

Re: KDEE3 question

[Jérôme Marant]

Tom Badran wrote:
On Friday 29 Nov 2002 9:12 am, Michael Meskes wrote:
 

Hi,
I was just being told that we won't get any KDE3 packages until the
whole Debian project is moved to gcc 3.2. Is this correct? If so, why?
And is there a timetable? Is it a problem to compile the packages with
gcc 2.95?
The matter of the fact is that almost all Debian/KDE users are already
using ftp.kde.org to get the KDE packages. I would surely prefer to have
these packages as part of our archive instead of redirecting everyone away
from debian.org.
   

The plan as i understand it is to supply gcc 3.2 built packages once kde 3.1 
is released, and never supply gcc 2.95 ones in debian
 

AFAIK, both gcc-3.2 and binutils have to be checked on every 
architecture and
the transition will start once all architectures are ready for the 
transition.

Cheers,

apt-src - some ideas

[Jon Kent]

Hi,

apt-src, good start, really like the fact that it
figures out dependancies and creates .deb file once
complete and is fairly easy to use.

However, there are some things I would like to see
added at some point if possible:

1) Option to compile dependances as opposed to
automatically downloading the dependant packages.

2) Option to add additional compile flags to gcc.

3) Option not to create a .deb file.  I wouldn't want
this but some people might.

4) With depandancies it would be nice to be told what
must be installed and what depandancies are optional.

5) Not sure it this is possible, but maybe downloading
source directly from author's site??

Comments?

Jon


__
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

Re: Bug#171116: ITP: tsclient -- GNOME2 frontend for rdesktop

[Thorsten Sauter]

On Fri, Nov 29, 2002 at 01:09:51PM +1100, Andrew Lau wrote:
> * Package name: tsclient
>   Version : 0.56
>   Upstream Author : Erick Woods <[EMAIL PROTECTED]>
> * URL : http://www.gnomepro.com/tsclient/
> * License : GPL


Hi,

there are now two open ITP's for the tsclient program. Are they
different? 

* #170400: ITP: tsclient -- Terminal Server Client - GNOME 2 rdesktop frontend
* #171116: ITP: tsclient -- GNOME2 frontend for rdesktop


Regards
Thorsten
 
-- 
Thorsten Sauter
<[EMAIL PROTECTED]>

(Is there life after /sbin/halt -p?)



pgpg7EQOCWTfN.pgp
Description: PGP signature

Re: location of UnicodeData.txt

[Bernhard R. Link]

* Jim Penny <[EMAIL PROTECTED]> [021128 03:35]:
> So, according to Branden, international standards are supposed to allow 
> debian the right to modify them and to distribute the modified versions.  
> Absent said permission, which is hardly ever going to be given,  they 
> must be considered non-free.  (This is, of course, logically forthright.) 
> Moreover, according to the non-free removal proponents, we should not 
> even distribute the un-modified copies of these files.
> 
> Yet, unicode is supposed to be the canonical character encoding scheme
> for debian.  
> 
> Does this mean every unicode text editor belongs in contrib (depends on
> something non-free)?

I think you are missing the points here.

First of all, DFSG applied to the standard does not want to change the 
standard, 
but wants all to be able to change the text of the standard.

This is a good thing, the text of standards should be modifiable. How else 
shall someone write the following standard without having written the first 
or having to write all from scratch?

Secondly: What has a unicode editor have to do with the unicode
standard? It should only implement it. If it would contain parts
of the standard-text (tables or whatever) that were protected by
copyright law and the standard would allow no modifications, then noone 
would be allowed to copy the editor. (No special problem with debian)


Hochachtungsvoll,
  Bernhard R. Link

-- 
 sagen wir mal...ich hab alle sourcen in /lost+found/waimea
 gEistiO: [...] Warum lost+found?
 wo haette ich es denn sonst hingeben solln?

Did the Satie fire break WNPP updates?

[Andrew Lau]

On Fri, Nov 29, 2002 at 11:22:07AM +0100, Thorsten Sauter wrote:
> there are now two open ITP's for the tsclient program. Are they
> different?

> * #170400: ITP: tsclient -- Terminal Server Client - GNOME 2
rdesktop frontend
> * #171116: ITP: tsclient -- GNOME2 frontend for rdesktop

Dear Thorsten,
No wonder I had a sense of deja vu when I filed #171116 today!
They're the same thing and in fact, I filed them both. Thanks for
spotting the duplication.

No wonder I had filed it twice. I must have panicked today when I
didn't see tsclient on WNPP.


Last Modified: Sat, Nov 9 03:28:23 UTC 2002

Hmm, WNPP seems be lagging heavily behind the BTS. Did the update
mechanism for WNPP get affected by the Satie fire? If so, someone's
forgotten to repair it.

Yours sincerely,
Andrew "Netsnipe" Lau

-- 
---
* Andrew "Netsnipe" LauComputer Science & Student Representaive, UNSW *
*   # apt-get into it Debian GNU/Linux Package Maintainer *
**
* GnuPG 1024D/2E8B68BD 0B77 73D0 4F3B F286 63F1  9F4A 9B24 C07D 2E8B 68BD *
---


pgpBK1UbepEL1.pgp
Description: PGP signature

Re: Bug#170400: ITP: tsclient -- Terminal Server Client - GNOME 2 rdesktop frontend

[Andrew Lau]

On Sun, Nov 24, 2002 at 12:21:00PM +0100, Thorsten Sauter wrote:
> > * Package name: tsclient
> thats the same like grdesktop, which was (before the fire) in
> the non-us incoming directory.
> Please see ITP: #164025.

Ya, different GNOME 2 RDP client. May I suggest that both our packages
include "Provides: rdpviewer"?

Yours sincerely,
Andrew "Netsnipe" Lau

-- 
---
* Andrew "Netsnipe" LauComputer Science & Student Representaive, UNSW *
*   # apt-get into it Debian GNU/Linux Package Maintainer *
**
* GnuPG 1024D/2E8B68BD 0B77 73D0 4F3B F286 63F1  9F4A 9B24 C07D 2E8B 68BD *
---


pgpstWoxFLwo5.pgp
Description: PGP signature

Help with kernel source installation

[Charles Garcia-Tobin]

Hello all

I'm trying to compile the linux-wlan drivers. To do so I need to
have the kernel source installed. I am not new to linux, but I am new to
debian and therefore I am a bit lost. I have installed it kernel 2.4.18bf.
When building the wlan drivers I was asked for the kernel source, so I
installed the kernel-source-2.4.18-5 package. However during the
configuration of the wlan drivers' source I am told that the kernel source
is incomplete or broken in some way. I guess that is because there is no
.config file. Looking in the distribution files I found a kernel-config file
packaged with each kernel flavour. So the question is would the following
steps be correct for setting up the kernel source tree:


1.  Install kernel-source-2.4.18-5 package
2. Uncompress and untar /usr/src/kernel-source-2.4.18-5.tar.bz2
3. Symbolic link /usr/src/linux --> kernel-source-2.4.18-5
4. cp kernel-config to /usr/src/linux/.config


Would that be right??? 

Any help appretiated

Cheers

Charles

/
// Charles García-Tobin ""-.
// Toy Designer   /[] _ _\ 
// Software and  _|_o_LII|_
// Firmware engineer/ |  | \  
//  |_|  |_|
//   ||" ||  ||
//   ||LI  o ||
//   ||''||
// Creature Labs/__||__\
/

Re: automatic selection which kernel image to install

[Jens Ruehmkorf]

On Thu, 28 Nov 2002, sean finney wrote:
> On Wed, Nov 27, 2002 at 06:45:21PM -0500, Michael Stone wrote:
> > Can you detect smp from non-smp kernel?
>
> i would guess not...

Just count the processors in /proc/cpuinfo, here is an example:

$ grep "^proc" /proc/cpuinfo
processor   : 0
processor   : 1
processor   : 2
processor   : 3

--
Gruss Jens

Re: Help with kernel source installation

[Andrew Lau]

On Fri, Nov 29, 2002 at 11:57:53AM -, Charles Garcia-Tobin wrote:
>   I'm trying to compile the linux-wlan drivers. To do so I need
> to have the kernel source installed. I am not new to linux, but I am
> new to debian and therefore I am a bit lost. I have installed it
> kernel 2.4.18bf.  When building the wlan drivers I was asked for the
> kernel source, so I installed the kernel-source-2.4.18-5
> package.

Dear Charles,
Could you please describe the linux-wlan drivers in a bit more
detail please. Can you give us a URL, are they proprietary modules, or
do they come included with the kernel source? What brand, model?

> However during the configuration of the wlan drivers' source I am
> told that the kernel source is incomplete or broken in some way. I
> guess that is because there is no .config file.

Paste the exact error message please. /usr/src/linux/.config isn't
really relevant. .config is the file where make {config, menuconfig,
xconfig} saves all the choices that one makes when deciding what to
they want included or not as part of that kernel. The kernel-config
file off the install media (or from /boot/config-$VERSION) is just a
renamed copy of the .config that was used to compile the current
running kernel for future reference purposes. So .config is unlikely
to the cause of your compile difficulties.

Yours sincerely,
Andrew "Netsnipe" Lau

-- 
---
* Andrew "Netsnipe" LauComputer Science & Student Representaive, UNSW *
*   # apt-get into it Debian GNU/Linux Package Maintainer *
**
* GnuPG 1024D/2E8B68BD 0B77 73D0 4F3B F286 63F1  9F4A 9B24 C07D 2E8B 68BD *
---


pgpQ1TMGM7Vep.pgp
Description: PGP signature

Re: How to validate Debian woody CDs?

[Vera Friederichs]

On Thu, 28 Nov 2002 18:58:46 +0100
Richard Atterer <[EMAIL PROTECTED]> wrote:

> First, mount the CD. Just working on the unmounted device doesn't
> always seem to work. Then, try one of two things:
> 
>   cat /dev/cdrom | md5sum
> 
> Apparently, this is preferable to "md5sum /dev/cdrom" because cat does
> a better job of really reading all the available data - YMMV. I'm not
> sure whether the above will work if more padding was added e.g. by
> cdrecord, so better try the alternative:
> 
>   dd if=/dev/cdrom bs=1k count=# | md5sum
> 
> where the # is the "1k-blocks" value from the output of
> "df /cdrom".
> 

I have tried several procedures (only with CD2):

md5sum /dev/cdrom

dd if=/dev/cdrom of=/tmp/image
md5sum /tmp/image

dd if=/dev/cdrom bs=1k count=8656864 | md5sum

The results were all the same, but not  equal to 
http://cdimage.debian.org/jigdo-area/current/jigdo/i386/MD5SUMS

Yes, I checked the README-files. They say it is woody (unofficial, but that 
should not be the problem?) and it is for i386. The "youngest" directories are 
from 18th July. Does anybody know what else could be the reason for the 
different checksums?

Thanks,
Vera

pgp6G5OzryLgJ.pgp
Description: PGP signature

Re: [Agnubis] Packaging Agnubis for Debian GNU/Linux

[Andrew Lau]

On Fri, Nov 29, 2002 at 08:02:07AM +0100, Rapha?l Bordet wrote:
> I've started an ITP for diacanvas a few time ago than is a
> dependencies for agnubis. I want to do the same with agnubis when
> the development is restarted. If you're interested by my project, we
> could be work both on supporting agnubis to debian.

Dear Raphael,
I'm curious as to why you did not file an ITP for Agnubis [1]
at the same time you filed an ITP on DiaCanvas2 [2] back in
October. You had originally hoped to package both Agnubis and
DiaCanvas for Debian GNU/Linux, right? I've just checked Quality
Assurance for your name and it seems like you have not yet packaged
anything before for Debian [3]. It would also seem that you have not
yet applied to join Debian either [4]. Are you new to the whole
debian-devel community?
If you are seriously considering joining Debian and need a
helping hand and a package or two to start off with and call your own,
then I'm willing to let you have Agnubis to yourself since I'm a
generous person.  Anyway there's not much I can do with Agnubis CVS in
its current state until development resumes.
If you would like me to hand Agnubis over to you (since it's
so heavily dependant on DiaCanvas) or co-maintain it with you until
you're ready to maintain it on your own, then you better join the new
maintainer queue and package DiaCanvas as soon as possible. That way,
I know that you're serious about Debian.
Does this offer sound fair to you?

Yours sincerely,
Andrew "Netsnipe" Lau

[1] http://bugs.debian.org/163544
[2] http://bugs.debian.org/151347
[3] http://qa.debian.org/
[4] http://nm.debian.org/

-- 
---
* Andrew "Netsnipe" LauComputer Science & Student Representaive, UNSW *
*   # apt-get into it Debian GNU/Linux Package Maintainer *
**
* GnuPG 1024D/2E8B68BD 0B77 73D0 4F3B F286 63F1  9F4A 9B24 C07D 2E8B 68BD *
---


pgpk3AJD3Rgux.pgp
Description: PGP signature

RE: Help with kernel source installation

[Charles Garcia-Tobin]

Hi Andrew



Thanks for your e-mail and offering to help, it is very kind of you.


>Dear Charles,
>   Could you please describe the linux-wlan drivers in a bit more
>detail please. Can you give us a URL, are they proprietary modules, or
>do they come included with the kernel source? What brand, model?


I cannot give you the exact error message at the moment as my computer is at
home and I'm at the office :-( I can give you the URL which is
www.linux-wlan.org The device I am trying for is an Actiontech USB 80211b
card. I believe that the drivers can be included in the kernel source and
compiled with it, but generally they are done post-kernel compilation, to
create modules. 

The machine is Toshiba 440CDT laptop which runs a P133MMX, with 32Mb
of RAM. I know its a very poor machine but my plan is to create a baby home
network gateway with it. I actually changed to Debian because people on the
linux-wlan list reported success when using debian. 

Thanks again 

Chalres

Re: [Agnubis] Packaging Agnubis for Debian GNU/Linux

[Amaya]

Andrew Lau dijo:
>   Does this offer sound fair to you?

He doesn't need to be a Developer (or even want to become one) in order
to package Agnubis. 

Maybe you could be his mentor and/or sponsor, or help him find one,
instead of making him go through NM?

-- 
 .''`.I'd like to fly, but my wings have been so denied...
: :' :Sand rains down and here I sit - Alice in Chains
`. `'Proudly running Debian GNU/Linux (Sid + 2.4.19 + Ext3)
  `-www.amayita.com  www.malapecora.com  www.chicasduras.com

Re: Pick a name, any name...

[Josip Rodin]

On Fri, Nov 29, 2002 at 09:44:17AM +, Andrew Suffield wrote:
> > > > - Another idea I had was something along the lines of Debsmith or
> > > >   Iansmith, to keep both the idea of Debian and the idea of the forge.
> > > >   Unfortunately, plenty of people are called that way.  Any idea to
> > > >   improve that line is welcome.  Maybe just smith.debian.org would be
> > > >   enough (and I'll bet we can find a music composer named Smith if we
> > > >   look hard enough).
> > > 
> > > Speaking of composers... how about Sibelius? Finnish composer, in honor of
> > > Linus. :-) Or Debussy, if you're going for French names.
> > 
> > We've got a debussy; it's an arm machine.
> 
> And there's already a Sibelius as a commercial software project; stay
> away from things that could raise trademark issues that way.

This isn't about our sourceforge fork package name, it's about the machine
name.

-- 
 2. That which causes joy or happiness.

reliable streams over UDP

[Russell Coker]

Do we have a library in Debian that provides reliable stream based 
communication over UDP?

I want to be able to deal with asymetric links and end-points that change IP 
address so TCP won't work.  Surely someone must have written something 
similar to TCP but implemented on top of UDP.  Is such a thing in Debian?  If 
not is there a DFSG compliant library that I can package?

-- 
http://www.coker.com.au/selinux/   My NSA Security Enhanced Linux packages
http://www.coker.com.au/bonnie++/  Bonnie++ hard drive benchmark
http://www.coker.com.au/postal/Postal SMTP/POP benchmark
http://www.coker.com.au/~russell/  My home page

Re: Pick a name, any name...

[Peter Mathiasson]

On Fri, Nov 29, 2002 at 02:15:05PM +0100, Josip Rodin wrote:
> > And there's already a Sibelius as a commercial software project; stay
> > away from things that could raise trademark issues that way.
> 
> This isn't about our sourceforge fork package name, it's about the machine
> name.

I thought it was about the service name, the machine already got one.

Quoting Martin Michlmayr <[EMAIL PROTECTED]>
"The machine itself has a name already (quantz)."

-- 
Peter Mathiasson, peter at mathiasson dot nu, http://www.mathiasson.nu
GPG Fingerprint: A9A7 F8F6 9821 F415 B066 77F1 7FF5 C2E6 7BF2 F228


pgpDC3igqPJuN.pgp
Description: PGP signature

Re: [Agnubis] Packaging Agnubis for Debian GNU/Linux

[Andrew Lau]

On Fri, Nov 29, 2002 at 02:08:50PM +0100, Amaya wrote:
> Andrew Lau dijo:
> > Does this offer sound fair to you?
> 
> He doesn't need to be a Developer (or even want to become one) in order
> to package Agnubis. 
> 
> Maybe you could be his mentor and/or sponsor, or help him find one,
> instead of making him go through NM?

Dear Amaya,
I'm in no position to sponsor his packages, since I'm still
awaiting DAM approval (11 months and counting). I was hoping that by
joining the NM process, he would be able to find a better candidate
(French speaking perhaps?) to help him learn the ropes of Debian (and
advocate/sponsor, if he takes the plunge). It's pretty much the best I
can do in my situation. I don't want to abandon my ITP on Agnubis
altogether if he's not committed, and the only measure I have to go so
far is a two month old unfinished ITP that happens to be an Agnubis
dependency. I'm more than happy to give an aspiring Debian developer a
break, since I know what's it's like starting out, not to mention
opportunities to package promising and prominent (GNOME-office in this
case) projects are rare these days.

Yours sincerely,
Andrew "Netsnipe" Lau

-- 
---
* Andrew "Netsnipe" LauComputer Science & Student Representaive, UNSW *
*   # apt-get into it Debian GNU/Linux Package Maintainer *
**
* GnuPG 1024D/2E8B68BD 0B77 73D0 4F3B F286 63F1  9F4A 9B24 C07D 2E8B 68BD *
---


pgpmnt6swpIlq.pgp
Description: PGP signature

Re: Bug#171116: ITP: tsclient -- GNOME2 frontend for rdesktop

[Colin Watson]

On Fri, Nov 29, 2002 at 03:15:25PM +1100, Andrew Lau wrote:
> On Thu, Nov 28, 2002 at 10:22:01PM -0500, Colin Walters wrote:
> > How about: Windows Terminal Services (RDP) client for GNOME 2
> 
> Thanks for the comments Kamion.

Kamion is me; Colin Walters is known on IRC as "walters".

Cheers,

-- 
Colin Watson  [EMAIL PROTECTED]

Adopting base-passwd

[Colin Watson]

Since it's an essential package I figured I should mention this before
acting: Wichert and I have agreed that I will be taking over maintenance
of base-passwd. An upload will be in incoming by this weekend.

Cheers,

-- 
Colin Watson  [EMAIL PROTECTED]


pgpejqdpExuhA.pgp
Description: PGP signature

Re: Bug#166311: roxen2: Roxen2 don't work with pike7.2

[Turbo Fredriksson]

Quoting Andrew Shirrayev <[EMAIL PROTECTED]>:

> 15:42:35  : --
>  0m 0.0s  : Pike v7.2 release 340, Roxen WebServer 2.1.265_Debian-6
>   : Loading pike modules ... Done [54.5ms]
>   : Loading roxen ... etc/modules/RXML.pmod/module.pmod:3049:Illegal
> to redefine variable "flags" as constant.

I _MIGHT_ have found a solution to this, and if I do, the rest should be ok
to... Try applying the enclosed patch on the RXML.pmod/module.pmod file.

- s n i p -
cd /usr/lib/roxen2/etc/modules/RXML.pmod
cat /tmp/RXML-Module.patch | patch
- s n i p -

Save the patch you receive in /tmp...

Unfortunatly I don't have a development system where I can try
out this myself...

--- module.pmod~	Thu May 17 04:34:25 2001
+++ module.pmod	Fri Nov 29 14:46:12 2002
@@ -3046,7 +3046,7 @@
 //! according to the type (which typically has a parser set).
 {
   inherit Frame;
-  constant flags = FLAG_UNPARSED;
+  Type flags = FLAG_UNPARSED;
   mapping(string:mixed) args = ([]);
 
   void create (Type type, string to_parse)

Re: Bug#171116: ITP: tsclient -- GNOME2 frontend for rdesktop

[Andrew Lau]

On Fri, Nov 29, 2002 at 01:43:27PM +, Colin Watson wrote:
> On Fri, Nov 29, 2002 at 03:15:25PM +1100, Andrew Lau wrote:
> > On Thu, Nov 28, 2002 at 10:22:01PM -0500, Colin Walters wrote:
> > > How about: Windows Terminal Services (RDP) client for GNOME 2
> > 
> > Thanks for the comments Kamion.
> 
> Kamion is me; Colin Walters is known on IRC as "walters".

Apologies to you both for the slip up.

Yours sincerely,
Andrew "Netsnipe" Lau

-- 
---
* Andrew "Netsnipe" LauComputer Science & Student Representaive, UNSW *
*   # apt-get into it Debian GNU/Linux Package Maintainer *
**
* GnuPG 1024D/2E8B68BD 0B77 73D0 4F3B F286 63F1  9F4A 9B24 C07D 2E8B 68BD *
---


pgpPx4T92jIvH.pgp
Description: PGP signature

Re: Pick a name, any name...

[Josip Rodin]

On Fri, Nov 29, 2002 at 02:25:28PM +0100, Peter Mathiasson wrote:
> > > And there's already a Sibelius as a commercial software project; stay
> > > away from things that could raise trademark issues that way.
> > 
> > This isn't about our sourceforge fork package name, it's about the machine
> > name.
> 
> I thought it was about the service name, the machine already got one.
> 
> Quoting Martin Michlmayr <[EMAIL PROTECTED]>
> "The machine itself has a name already (quantz)."

Now that is a bit more tricky, since it's more exposed... dunno. Anyhow,
nobody proposed sibelius, did they? :)

-- 
 2. That which causes joy or happiness.

Re: KDEE3 question

[Michael Meskes]

On Fri, Nov 29, 2002 at 09:52:52AM +, Tom Badran wrote:
> The plan as i understand it is to supply gcc 3.2 built packages once kde 3.1 
> is released, and never supply gcc 2.95 ones in debian

but since gcc 2.95 is still standard in Debian I guess the switch to 3.2
will take much longer than the release of KDE 3.1 which is due next
week.

And the explanation "we have no KDE 3 since we are still using gcc 2.95"
doesn't sound too good because there is no technical reason to couple
these two.

Michael
-- 
Michael Meskes
Michael@Fam-Meskes.De
Go SF 49ers! Go Rhein Fire!
Use Debian GNU/Linux! Use PostgreSQL!

Re: KDEE3 question

[Michael Meskes]

On Fri, Nov 29, 2002 at 11:07:15AM +0100, JÃrÃme Marant wrote:
> AFAIK, both gcc-3.2 and binutils have to be checked on every 
> architecture and
> the transition will start once all architectures are ready for the 
> transition.

And of course we have to create a plan for this transition before we can
start, don't we? Or do we have such a plan?

Frankly, I can see the transition taking some more time, but still fail
to see a good reason for not compiling KDE3 on gcc 2.95 as this
definitely works or else there wouldn't be a single KDE3 for woody
package on kde.org.

If it's just bandwidth tell me and I do the upload.

Michael
-- 
Michael Meskes
Michael@Fam-Meskes.De
Go SF 49ers! Go Rhein Fire!
Use Debian GNU/Linux! Use PostgreSQL!

Re: apt-src - some ideas

[Jose Carlos Garcia Sogo]

El día 29 nov 2002, Jon Kent escribía:
> Hi,
> 
> apt-src, good start, really like the fact that it
> figures out dependancies and creates .deb file once
> complete and is fairly easy to use.
> 
> However, there are some things I would like to see
> added at some point if possible:
> 
> 1) Option to compile dependances as opposed to
> automatically downloading the dependant packages.
> 
> 2) Option to add additional compile flags to gcc.
> 
> 3) Option not to create a .deb file.  I wouldn't want
> this but some people might.

  Do you mean calling directly dpkg?
  

> 4) With depandancies it would be nice to be told what
> must be installed and what depandancies are optional.

  Build-Dependecies are always mandatory.

> 
> 5) Not sure it this is possible, but maybe downloading
> source directly from author's site??

  As a package mantainer you're encouraged to void makig changes to the
  upstream sources, but sometimes that has to be done. Some of them
  could be avoided with dpkg-source v2, but others are needed (for
  example when files are removed because it doesn't follow DFSG).

  The other problem I see with this is that you lose this way the
  benefit of having a bunch of mirrors near you. And (if some day)
  packages are signed and checked in download time, you couldn't do that
  either.


-- 
  Jose Carlos Garcia Sogo
 [EMAIL PROTECTED]


pgpsAXEte9zEE.pgp
Description: PGP signature

Re: Bug#166311: roxen2: Roxen2 don't work with pike7.2

[Turbo Fredriksson]

Sorry, to tired. Just did a reply, didn't see where it was going...
-- 
radar Nazi plutonium strategic pits toluene Delta Force
counter-intelligence Serbian Uzi AK-47 Noriega colonel Waco, Texas
Iran
[See http://www.aclu.org/echelonwatch/index.html for more about this]

Re: Help with kernel source installation

[Matthew Garrett]

In chiark.mail.debian.devel, you wrote:

>Dear Charles,
>   Could you please describe the linux-wlan drivers in a bit more
>detail please. Can you give us a URL, are they proprietary modules, or
>do they come included with the kernel source? What brand, model?

They're an alterative set of drivers for Prism 2/2.5/3 chipset based
wireless cards. They're DFSG free and packaged in unstable.

>Paste the exact error message please. /usr/src/linux/.config isn't
>really relevant. .config is the file where make {config, menuconfig,
>xconfig} saves all the choices that one makes when deciding what to
>they want included or not as part of that kernel. The kernel-config
>file off the install media (or from /boot/config-$VERSION) is just a
>renamed copy of the .config that was used to compile the current
>running kernel for future reference purposes. So .config is unlikely
>to the cause of your compile difficulties.

Kernel modules need to be built against a kernel source that's
configured the same way as the current kernel. The original poster wants
to cp /boot/config-`uname -r` /usr/src/linux/.config and make oldconfig;
make dep before trying to build the linux-wlan modules. That'll result
in a kernel source that matches the running kernel.

-- 
Matthew Garrett | [EMAIL PROTECTED]

Re: KDEE3 question

[Colin Watson]

On Fri, Nov 29, 2002 at 03:05:46PM +0100, Michael Meskes wrote:
> Frankly, I can see the transition taking some more time, but still fail
> to see a good reason for not compiling KDE3 on gcc 2.95 as this
> definitely works or else there wouldn't be a single KDE3 for woody
> package on kde.org.

The good reason is that all the KDE library sonames have to be changed
when switching to gcc 3.2. The KDE developers rightly wish to avoid this
disruption.

-- 
Colin Watson  [EMAIL PROTECTED]

Re: Are we losing users to Gentoo?

[Joey Hess]

Josip Rodin wrote:
> Yes, there is. The debian-cd mirrors on our list are very diverse: some have
> 2.2r*, some have 3.0r*, some don't have full ISOs at all. Expecting
> debian-www team to start making grossly hackish scripts to compensate for
> whatever the hell people put in debian-cd/ directories on their sites, or
> maintain separate lists of sites that have debian-cd organized properly,
> that is just unreal.

So you define what a debian cd mirror should have, and mark all
incomplete mirrors as such (with a scanning program), and don't use them
for this purpose. An inconsistent mirror network can be worse than no
mirror network at all.

> (Especially, I should note, when it's basically going to help not only some
> honest people who just got confused, but also give a false sense of easyness
> to a bunch of people who really shouldn't be installing Debian in the first
> place since they have a total aversion to reading documentation.)

Well, no, I have read plenty of documentation in the past, and that was
an honest example of me trying to find an iso and how it just doesn't
work. Sure, I was skimming the web pages quite fast. Everyone does.

To tie back to the thread, maybe one reason we're possibly losing
(new) users to gentoo is:
- go to gentoo.org
- "Gentoo Linux/x86" on left hand side
- click on link directly to directory on ibiblio.org with an iso on it

Another reason might be evidenced by the finger pointing,
insults/elitism, and probably lack of any action at all I just got from
a developer in response to a valid complaint.

-- 
see shy jo


pgpR0vhXJirzA.pgp
Description: PGP signature

Re: KDEE3 question

[Paul Cupis]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On Friday 29 November 2002 09:12, Michael Meskes wrote:

> The matter of the fact is that almost all Debian/KDE users are already
> using ftp.kde.org to get the KDE packages.

Um, how do you know this?

Paul Cupis
- -- 
[EMAIL PROTECTED]

-BEGIN PGP SIGNATURE-
Version: GnuPG v1.2.0 (GNU/Linux)

iD8DBQE954B8IzuKV+SHX/kRAhIcAJ9/CkGTZigibsgfhjf2p1tR39TL3gCffRoN
0HZadkXoB4qrfP3aaL70cl0=
=kUF/
-END PGP SIGNATURE-

Re: Are we losing users to Gentoo?

[Jon Kent]

OK, howzabout some useful links that show that
although Debian may be losing some users, which is
still a shame, it perhaps not as bad as some would
think.

http://forums.gentoo.org/viewtopic.php?t=3614

http://forums.gentoo.org/viewtopic.php?t=24417

The first link shows a poll done a while ago on which
OS was used before Debian, and the second one ask
specifically if someone has moved from Debian to
Gentoo, which is a bit new to be mega useful, but
still interesting.

BTW, God I wish Debian had forums like this, far
easier that email lists (and no I can't set this up
before someone suggests it).

Jon

__
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

Re: KDEE3 question

[Michael Meskes]

On Fri, Nov 29, 2002 at 02:57:32PM +, Paul Cupis wrote:
> > The matter of the fact is that almost all Debian/KDE users are already
> > using ftp.kde.org to get the KDE packages.
> 
> Um, how do you know this?

Sorry, I of course don't know all Debian/KDE users. I was just
extrapolating. I talked to quite some users and they all did. No single
KDE user remained with 2.2.2. Also we have some customers using
Debian/KDE with the complete networks using KDE3.

Michael
-- 
Michael Meskes
Michael@Fam-Meskes.De
Go SF 49ers! Go Rhein Fire!
Use Debian GNU/Linux! Use PostgreSQL!

Re: Are we losing users to Gentoo?

[Joey Hess]

Gerfried Fuchs wrote:
> > And then on scroll way down the list to your country. And then into the
> > current directory on the mirror, oops, that was jigdo only?!  back out
> > and to the 3.0r0 directory.
> 
>  Uhm, just a second.  When I click a on the links in that list I get to
> the debian-cd directory on the mirrors where I can choose between jigdo
> and the 3.0r0 directory.  Why do you find yourself in the jigdo
> directory?

I found myself in a directory that had nothing but a jigdo subdirectory.
I didn't bother looking in the subdirectory, not wanting jigdo.

> > What, that was jigdo again?! Hmm, try another mirror.
> 
>  Now it would be _more_ than helpful which link you are refering to.
> Without knowing it that seemingly broken link can't be fixed, thank you.

http://aurolinux.mit.edu/debian-cd/ -- no debian 3.0!
ftp://carroll.aset.psu.edu/pub/linux/distributions/debian-cd/ --
  overloaded, cannot check
ftp://debian-cd.rutgers.edu/pub/ -- no debian 3.0
ftp://debian.fifi.org/pub/debian-cd/ -- jigdo only
ftp://debian.orst.edu/debian-cdimage/ -- jigdo only
ftp://debian.tod.net/debian-cd/ -- cannot connect (routing?)
ftp://debian.uchicago.edu/debian-cd/ -- stable points to debian 2.2r6
ftp://ftp-linux.cc.gatech.edu/pub/debian-cd/ -- no debian 3.0
ftp://ftp-mirror.internap.com/pub/debian-cd/ -- jigdo only
ftp://ftp.cs.stevens-tech.edu/pub/Linux/distributions/debian-cd/ -- no
  debian 3.0
ftp://ftp.keystealth.org/debian-cd/debian-cd/ -- no debian 3.0
ftp://ftp.egr.msu.edu/debian-cd/ -- 404
ftp://ftp.linux.tucows.com/pub/ISO/Debian/ -- connection refused
ftp://ftp.lug.udel.edu/pub/iso-images/Debian/ -- no debian 3.0
ftp://ftp.rutgers.edu/pub/debian-cd/ -- no debian 3.0
ftp://linux.csua.berkeley.edu/debian-cd/ -- no 3.0
ftp://mirror.cs.wisc.edu/pub/mirrors/linux/debian-cd/ -- jigdo only
ftp://mirror.csit.fsu.edu/debian-cd/ -- has isos, but directory is not readable
ftp://mirrors.kernel.org/debian-cd/ -- no 3.0
ftp://mirrors.xmission.com/debian-cd/ -- no 3.0
http://telia.dl.sourceforge.net/mirrors/debian-cd/ -- jigdo only
http://umn.dl.sourceforge.net/mirrors/debian-cd/ -- jigdo only
http://unc.dl.sourceforge.net/mirrors/debian-cd/ -- jigdo only

I don't know which one I clicked on, but it was one of these. The above
are every one of our listed cd mirrors in the US, none of which have
isos, and a full half of which think that 2.2 was our last release.

> > Maybe the one in Austria, because it's the top of that list of
> > mirrors. Hmm, no, it only has a jigdo directory too.
> 
>  Uhm, you seem to be blind, sorry.  Both ftp and http of the austrian
> mirror has the 3.0r0 directories.  Can you please check before accusing?
> I guess you seem to be puzzled by the layout of the HTML-Page there --
> but you can't accuse debian for third party layouts, or do you try to do
> so?

Why should we have a bunch of diverse third party web pages on something
that's supposed to be a mirror network?

-- 
see shy jo


pgpkaj8JZLCF3.pgp
Description: PGP signature

Re: [Agnubis] Packaging Agnubis for Debian GNU/Linux

[Amaya]

Hi all...

Andrew Lau dijo:
>   I'm in no position to sponsor his packages, since I'm still
> awaiting DAM approval (11 months and counting). I was hoping that by
> joining the NM process, he would be able to find a better candidate
> (French speaking perhaps?) to help him learn the ropes of Debian (and
> advocate/sponsor, if he takes the plunge). 

No, really, it won't work like that. He needs a DD to be willing to
sponsor him first. 

> It's pretty much the best I can do in my situation. I don't want to
> abandon my ITP on Agnubis altogether if he's not committed, and the
> only measure I have to go so far is a two month old unfinished ITP
> that happens to be an Agnubis dependency.

Ok, he didn't ITP Agnubis, did he? Just go ahead, package it well,
upload it. Everyone's happy :-) 

> I'm more than happy to give an aspiring Debian developer a break,
> since I know what's it's like starting out, not to mention
> opportunities to package promising and prominent (GNOME-office in this
> case) projects are rare these days.

I know, been there, done that. It was fun for me, fullfilling. Probably
because it was something I wanted to do. Don't push him into NM.

Condider that his ITP has been there for two months. He's either not
that motivated or simply busy. And NM is much more time consuming than
closing an ITP. Offer your help, whatever help you are willing or able
to give. 

If he is too busy to package whatever was that depended so heavily on
Agnubis, offer to retitle the bug to a RFP. 

But all this discussion is kind of useless until he tells us what's on
his mind first ;-)

-- 
 .''`.I'd like to fly, but my wings have been so denied...
: :' :Sand rains down and here I sit - Alice in Chains
`. `'Proudly running Debian GNU/Linux (Sid + 2.4.19 + Ext3)
  `-www.amayita.com  www.malapecora.com  www.chicasduras.com

Re: Are we losing users to Gentoo?

[Joey Hess]

Gerfried Fuchs wrote:
> > But maybe instead, back at debian.org's front page, you picked the
> > "Getting Debian" link instead. Only to end up on a page that links to cd
> > vendors and "downloading over the Internet". Ok, the latter. But it
> > points to a page that only lets one download unnofficial netinst iso
> > images, which are of varying quality, and well, unnoficial.
> 
>  You are right, there should be added a link to the $(HOME)/CD/http-ftp/
> on that page, too.  Thanks for the (quite hidden) suggestion.  Joy, do
> you think that would help, too?

Perhaps I was too obscure, because you still missed my point: In user
interface design, having two suboptimal ways of doing something is
*worse* than have one, even suboptimal, way of doing something. There
are currently two paths from the front page to two different pages for
downloading debian on iso.

The current layout is:

front page > getting debian
|  . |
| .  | 
|.   |
|   .|
v   vv
debian on cd downloading over the internet
|   \|
|\   |
| \  |
|  \ |
v   \v  / a few inconsistent
mirror list  `-> minimal cd - but useful
/| | | | |\ \ and unofficial images
   / | | | | | \
  v  v v v v v  v
 oodles of badly maintained
 and inconsistant mirrors

Adding the proposed link shown by the dotted line does not make this one
whit easier to use. For a linear task like deciding how to download
debian and finding a place to do so, it's a mess. A sane structure would
look something like this:

front page > gimme an i386 iso now! --> .iso
|
|
|
v
downoad debian
|
|
|
v
debian on cd
|  |
|  |
|  |
|  v
| unofficial mini isos
|
v
mirror list
/|   |\
   / |   | \
  v  v   v  v
 just a few well
 maintained and
 consistent mirrors
  
>  But your first approach was correct, only that you seem to have been
> confused by third party html layouts on which we don't have any
> influence.

If my first approach was correct, and my second approach therefore,
presumably, incorrect, they why do we have a prominent link to the
second approach on the debian web site, under the attractive label of
"Getting Debian"? Think about it.

-- 
see shy jo


pgpcan5zFGIsF.pgp
Description: PGP signature

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Christian Surchi]

On Thu, Nov 28, 2002 at 11:14:14PM -0500, Ari Pollak wrote:
> Firstly, I don't think this is a matter for debian-devel as it doesn't 
> involve the project as a whole and would best be dealt with privately. 

I've already written to you. I wanted to share my ideas because I don't
like this situation and I don't care the technical question... I don't
consider Debian only from a technical point of view and I didn't see any
collaborative step towards me. I'm sure you are a better maintainer for
logjam and I'm happy fro upstream development too.

> Secondly, I'd like to apologize to Christian and the Debian project for 
> not following exact policy. However, in my defense, I did contact 
> Christian several times before making the NMU, 

I listed all your contacts... I could be happy if you ask me to work on
logjam. You asked only for new packages. I passed it to you if you
offered help. You sent only a wishlist bug the day 4.0.0 was released
and a mail with a one line question. :)

> and did explain to him 
> numerous times privately after the NMU why I did certain things (such as 
> using bug fixes from CVS, since certain things in the actual release 
> needed to be corrected). 

No information to me about your intention.

> Also, the actual logjam package had not 
> received any attention since May even though a new 3.0.x release had 
> been out, so having not received any response from any of my e-mails to 
> Christian before the NMU, I felt things were going unnoticed. Also in my 
> defense, the new 4.x version did fix a serious-level bug filed against 
> the package, and the 3.x series of logjam was basically very unstable as a 
> whole. I have also been in close contact with the upstream author of 
> logjam, and maintain one of his other software packages, gtkspell, for
> Debian.

Good. As I've said I didn't like your way to act and I don't think it's
good for Project.

> In any case, if you have any further discussions about this, 
> please contact Christian and me privately, unless this is appropriate 
> to post in debian-devel for some reason.

I explained my reasons.

-- 
Christian Surchi, [EMAIL PROTECTED], [EMAIL PROTECTED] |   ICQ 
www.debian.org - www.softwarelibero.it - www.firenze.linux.it| 38374818
Disk crisis, please clean up!

Re: automatic selection which kernel image to install

[sean finney]

On Fri, Nov 29, 2002 at 01:25:09PM +0100, Jens Ruehmkorf wrote:
> On Thu, 28 Nov 2002, sean finney wrote:
> > On Wed, Nov 27, 2002 at 06:45:21PM -0500, Michael Stone wrote:
> > > Can you detect smp from non-smp kernel?
> >
> > i would guess not...
> 
> Just count the processors in /proc/cpuinfo, here is an example:

well the trouble is, if you're not running an smp kernel, the kernel
doesn't know bout multiple cpus to begin with, right?  afaik there'd
only be one cpu entry in /proc/cpuinfo if you're not running an
smp kernel 


sean


pgpeq2VGBArrX.pgp
Description: PGP signature

Re: Help with kernel source installation

[sean finney]

On Fri, Nov 29, 2002 at 11:57:53AM -, Charles Garcia-Tobin wrote:
> 1.  Install kernel-source-2.4.18-5 package
> 2. Uncompress and untar /usr/src/kernel-source-2.4.18-5.tar.bz2
> 3. Symbolic link /usr/src/linux --> kernel-source-2.4.18-5
> 4. cp kernel-config to /usr/src/linux/.config

yeah, followed by a make dep and you should be set _if_ the kernel
source is the same version as the kernel you're running.   

actually, the reason why building those drivers requires the kernel
source is because it needs the kernel-header files.  so, alternatively, if
you're running one of the debian pre-configured kernel-image packages, you
can just install the kernel-headers package of the same version kernel,
and then you don't need the kernel source at all.  or, if you're made
your own kernel with make-kpkg, make-kpkg binary will make not only a
deb for your kernel image but also make a kernel-headers package as well.


anyway, good luck
sean


pgpWVdM9GNzdQ.pgp
Description: PGP signature

New thread about people switching to other distributions

[David B Harris]

Hey there :)

I'm starting a new thread about people switching to other distributions.
Why? Because I'd rather we start first with an information-gathering
thread.

I'm shortly going to relate my first-hand knowledge of why people have
switched from Debian, to Gentoo. Everybody else who has any first-hand
knowledge about it, please relate your experiences too. Also, if you
know of somebody who was thinking "Hmmm, Debian or Gentoo?", and they
chose Gentoo, do tell why. (I don't know anybody who's been in that
situation, unfortunately. I think they've probably got great insights.
Joey's comments on the web site would probably also come from somebody
in this group of Gentoo users.)

If you don't have any first-hand knowledge, please don't respond to this
thread. We'll start a "what do we do with this knowledge now?" thread
later.

Okay, so, my experiences.

I've watched about three dozen people from #debian "switch" to Gentoo.
About two dozen of them switched simply because it was cool. I mean,
these people showed up in #debian because they had tried Slackware and
failed miserably (yes. This happens like every other day). Now they were
trying Debian, because it was the next step up in the coolness meter.

No, I'm not just assuming these people are idiots :) I actually talk to
them privately, help them out, and whatnot.

The other third of people I've watched switched just wanted something
new to play with. Would have been FreeBSD, probably, if Gentoo hadn't
been around. Most of them have already tried SuSE, Mandrake, Red Hat,
and their ilk.

One person had a very good reason - they had a client whose company
policy was to use specific gcc flags (can't remember which, but IIRC
they had something to do with hardware stability and debugging) - Gentoo
let them do that more easily.

Anybody else?


pgpuDqpwBovp8.pgp
Description: PGP signature

Re: gpg-agent?

[Joey Hess]

Brian May wrote:
> I would hope that gpg-agent follows similar principles...

Oh, so it is a running gpg instance with key, and running gpg just
passes it data. That makes sense. Someone package it, please! :-)

-- 
see shy jo


pgpbjjlMuKoy1.pgp
Description: PGP signature

Re: reliable streams over UDP

[Radovan Garabik]

On Fri, Nov 29, 2002 at 02:16:20PM +0100, Russell Coker wrote:
> Do we have a library in Debian that provides reliable stream based 
> communication over UDP?
> 
> I want to be able to deal with asymetric links and end-points that change IP 
> address so TCP won't work.  Surely someone must have written something 
> similar to TCP but implemented on top of UDP.  Is such a thing in Debian?  If 
> not is there a DFSG compliant library that I can package?

probably not what you are looking for, but utalk contains an
implementation of semi-reliable protocol over UDP, used for, well,
talk-like behaviour.

-- 
 ---
| Radovan Garabík http://melkor.dnp.fmph.uniba.sk/~garabik/ |
| __..--^^^--..__garabik @ melkor.dnp.fmph.uniba.sk |
 ---
Antivirus alert: file .signature infected by signature virus.
Hi! I'm a signature virus! Copy me into your signature file to help me spread!

Re: reliable streams over UDP

[Tommi Virtanen]

On Fri, Nov 29, 2002 at 02:16:20PM +0100, Russell Coker wrote:
> Do we have a library in Debian that provides reliable stream based 
> communication over UDP?
> 
> I want to be able to deal with asymetric links and end-points that change IP 
> address so TCP won't work.

Why would TCP have trouble with asymmetric links? And if you
mean links that only allow traffic to pass in one direction
(there is no alternative backchannel), you really can't have
reliability.

For end-points that change IP, either look at SCTP, or build
a protocol that allows you to establish new TCP connections and
switch the logical connection to another TCP connection.

There are _really_ good reasons for using TCP, and really quite
few for not using it. In most real world scenarios, you end up
reimplementing TCP bit-by-bit, in your own protocol, over UDP.

Also, apt-cache show rocks.

-- 
:(){ :|:&};:

Re: reliable streams over UDP

[Russell Coker]

On Fri, 29 Nov 2002 17:48, Tommi Virtanen wrote:
> On Fri, Nov 29, 2002 at 02:16:20PM +0100, Russell Coker wrote:
> > Do we have a library in Debian that provides reliable stream based
> > communication over UDP?
> >
> > I want to be able to deal with asymetric links and end-points that change
> > IP address so TCP won't work.
>
>   Why would TCP have trouble with asymmetric links? And if you

Asymmetric in that the send and receive IP addresses for one end are not the 
same...

>   mean links that only allow traffic to pass in one direction
>   (there is no alternative backchannel), you really can't have
>   reliability.

True.  But if the backchannel is only out part of the time then it can be 
reliable when there is a backchannel.

>   For end-points that change IP, either look at SCTP, or build
>   a protocol that allows you to establish new TCP connections and
>   switch the logical connection to another TCP connection.
>
>   There are _really_ good reasons for using TCP, and really quite
>   few for not using it. In most real world scenarios, you end up
>   reimplementing TCP bit-by-bit, in your own protocol, over UDP.
>
>   Also, apt-cache show rocks.

Thanks, I'll look into it.

-- 
http://www.coker.com.au/selinux/   My NSA Security Enhanced Linux packages
http://www.coker.com.au/bonnie++/  Bonnie++ hard drive benchmark
http://www.coker.com.au/postal/Postal SMTP/POP benchmark
http://www.coker.com.au/~russell/  My home page

Re: Are we losing users to Gentoo?

[Thomas Bushnell, BSG]

Joey Hess <[EMAIL PROTECTED]> writes:

> Thomas Bushnell, BSG wrote:
> > Indeed, the Debian home page is so well organized and so easy to find
> > and get around in, that people don't *need* so many secondary sources
> > of information.  Our success at doing our job well has meant that the
> > distrowatch counter is especially inaccurate in our case.
> 
> Wooh, that's a rich one. Above speaks a man who does not read the
> debian-www mail I suppose.

But I use the website.  Here's a questions.  Go to eh redhat site and
see if you can figure out where to get a complete RedHat CD downloaded
from the net?

Of course our site has lots of room for improvement.  I still submit
it's way better than the competitors.

Re: Pick a name, any name...

[Thomas Bushnell, BSG]

"John H. Robinson, IV" <[EMAIL PROTECTED]> writes:

> Francesco Paolo Lovergine wrote:
> > I like maintaining the idea of forge, so my proposal is VULCAN
> > (from Roman mithology).
> 
> i prefer greek: hephaistos

I prefer Greek too, but that means spelling it in Greek.  :)

Re: location of UnicodeData.txt

[Thomas Bushnell, BSG]

Richard Braakman <[EMAIL PROTECTED]> writes:

> On Thu, Nov 28, 2002 at 02:58:38PM +0100, Tim Dijkstra wrote:
> > > UnicodeData is different, because we need the data in our program,
> > > not only the ideas. And it this case we see that as software!
> >  
> > Maybe you're right that we don't really need the rfc's in main. They
> > actually are now and it would be a shame if we dropped them. But we need
> > files like this unicode file in main, which is part of a specification
> > (I think), so can't  be altered.
> 
> But do you think it's _okay_ for such a file not to be free?  (Whether
> it actually is or not is a topic for debian-legal).

Whether it's OK is not for debian-devel.

Whether it is or is not is for debian-legal, and I'll comment on the
thread if and when it shows up there.

Re: gpg-agent?

[martin f krafft]

also sprach Joey Hess <[EMAIL PROTECTED]> [2002.11.29.1722 +0100]:
> Oh, so it is a running gpg instance with key, and running gpg just
> passes it data. That makes sense. Someone package it, please! :-)

from the current packager i had to hear that it's largely unstable.
many bugs and such. we'll have to wait...

-- 
 .''`. martin f. krafft <[EMAIL PROTECTED]>
: :'  :proud Debian developer, admin, and user
`. `'`
  `-  Debian - when you have better things to do than fixing a system
 
NOTE: The public PGP keyservers are broken!
Get my key here: http://people.debian.org/~madduck/gpg/330c4a75.asc


pgp9e1YvAgjok.pgp
Description: PGP signature

Re: automatic selection which kernel image to install

[Jens Ruehmkorf]

On Fri, 29 Nov 2002, sean finney wrote:

> well the trouble is, if you're not running an smp kernel, the kernel
> doesn't know bout multiple cpus to begin with, right?

That's true.

> afaik there'd only be one cpu entry in /proc/cpuinfo if you're not
> running an smp kernel

But it's a good default. For non-interactive, automatic installations boot
an i386-smp kernel, maybe even as default kernel for the d-i.

--
Gruss Jens

Re: Are we losing users to Gentoo?

[Joey Hess]

Thomas Bushnell, BSG wrote:
> But I use the website.  Here's a questions.  Go to eh redhat site and
> see if you can figure out where to get a complete RedHat CD downloaded
> from the net?

The comparison is only fair with organizations that *want* you do do so
(so not redhat, probably not openbsd, or mandrake, or others whose
principal developers try to sell cds).

> Of course our site has lots of room for improvement.  I still submit
> it's way better than the competitors.

There's a straight 2 click path to a directory with an ISO image on it
on the gentoo site. For freebsd, it's 3 obvious clicks to a ftp site
directory, then click on arch and version. For netbsd, 5 clicks to a
mirror (one hidden far down a page). For debian, it's 4 clicks, _if_ you
avoid the unofficial images, and the false path that leads only to them.
And _if_ you happen to pick one of the small fraction of listed mirrors
that really work.

So no, in this case our web site is behind all but netbsd in structure,
and behind netbsd in the sory state of our cdrom mirror network.

The debian web site has some nice stuff, but mostly for developers. And
it suffers from accreting for years, with no overall vision, and little
refactoring. If it were code I'd call it quite crufty and overfeatured
and badly designed.

-- 
see shy jo


pgpeotMEWMNRk.pgp
Description: PGP signature

Re: Are we losing users to Gentoo?

[Joey Hess]

Jon Kent wrote:
> BTW, God I wish Debian had forums like this, far
> easier that email lists (and no I can't set this up
> before someone suggests it).

debianplanet.org has stuff like this (incidentially and only because I'm
stuck on the subject -- it's marginally easier to find a a debian cd
image from debianplanet's web site then from debian.org :-P).

Interesting polls BTW, they do bear out that a lot of gentoo users are
ex-debian users.

-- 
see shy jo


pgpzcdLsnsNwW.pgp
Description: PGP signature

Re: Pick a name, any name...

[John H. Robinson, IV]

Russell Coker wrote:
> >
> > that is _NOT_ what a CNAME is for. a CNAME is for when the hostname is
> > in a domain that is OUTSIDE of your control.
> >
> > ie: evil.debian.org -> www.msn.com = CNAME (we don't control the msn.com
> > domain)
> > forge.debian.org -> quantz.debian.org = A (we control the debian.org
> > domain, so we can save the internet by REDUCING THE NUMBER OF
> > UNNECESSARY DNS LOOKUPS AND REDUCE THE END USERS DELAY WITH DNS LOOKUP
> > REQUIREMENTS)
> 
> How does that save the Internet?

because with a A record one lookup -> IP address(es)
with a CNAME one lookup -> another lookup -> IP address(es)

CNAME cause you to at least DOUBLE the DNS lookups, DOUBLING the
potential timeouts, DOUBLING dns load

can you do it? certainly. should you be aware of the effects WHEN you do
it? yes. you may break the rules only after you understand the rules.

> DNS entries are cached and don't cause that much traffic.

the second and subsequent times, yes. not the first time.

> Having CNAME entries pointing to your own A records is OK, just as having 
> symbolic links pointing to your own files on the same file system is OK.

except that a symlink does not impact networks (except in the case of
NFS)

-john

Re: Are we losing users to Gentoo?

[Jon Kent]

--- Joey Hess <[EMAIL PROTECTED]> wrote:

> debianplanet.org has stuff like this (incidentially
> and only because I'm

yeh I know but not as easy to use as Gentoo's are
IMHO.  BTW I agree with you regarding CD images.  Gave
up in the end trying to download and order CDs from
Linux Emporium instead.  Its was just too bloody
painful :-(

Jon

__
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

Re: apt-src - some ideas

[Joey Hess]

Jon Kent wrote:
> apt-src, good start, really like the fact that it
> figures out dependancies and creates .deb file once
> complete and is fairly easy to use.
> 
> However, there are some things I would like to see
> added at some point if possible:
> 
> 1) Option to compile dependances as opposed to
> automatically downloading the dependant packages.

I believe that apt-build can do this. Unfortunatly, we have some
divergent efforts with apt-build and apt-src, and I have not allocated
enough time to trying to merge the two. Not too worried about this yet 
since cometition can be good in the early stages.

> 2) Option to add additional compile flags to gcc.

pentium-builder lets you bolt this onto the side, but is a hack. Doing
it right involves putting something into policy, and checking/modifying
every build process of every package.

> 3) Option not to create a .deb file.  I wouldn't want
> this but some people might.

apt-src only builds debian packages if you pass it the --build option
when doing something else. For example:

apt-src upgrade # upgrades all installed source trees
apt-src upgrade -n # bpgrades and builds all installed source trees

You can also force --build on by default with the config file, and I
have been considering making that the default in which case you'd use
--no-build or the config file to turn it off. So I think I have this one
covered.

> 4) With depandancies it would be nice to be told what
> must be installed and what depandancies are optional.

We don't have a Build-Recommends or Build-Suggests, and it's not clear
they're worthwhile at all (from the perspective of autobuilders).

I could add an option to apt-src to make it not resolve any build
dependencies, and just warn about missing ones or something.

> 5) Not sure it this is possible, but maybe downloading
> source directly from author's site??

Not something I want to touch. Maybe if everyone added watch files to
their packages, then use uscan.

-- 
see shy jo


pgpRAxphqVdqU.pgp
Description: PGP signature

Re: apt-src - some ideas

[Jon Kent]

--- Joey Hess <[EMAIL PROTECTED]> wrote:
> > 1) Option to compile dependances as opposed to
> > automatically downloading the dependant packages.
> 
> I believe that apt-build can do this. Unfortunatly,

Ok, I haven't looked at apt-build yet.

> 
> > 2) Option to add additional compile flags to gcc.
> 
> pentium-builder lets you bolt this onto the side,
> but is a hack. Doing

That great, got that installed so I'll have a play :-)

> > 3) Option not to create a .deb file.  I wouldn't
> want
> > this but some people might.
> 
> apt-src only builds debian packages if you pass it
> the --build option

Whoops, I should rtfm a bit more closely.

> > 4) With depandancies it would be nice to be told
> what
> > must be installed and what depandancies are
> optional.
> 
> We don't have a Build-Recommends or Build-Suggests,
> and it's not clear
> they're worthwhile at all (from the perspective of
> autobuilders).

I more thinking along the lines of xyz apps need X and
can optionally have GTK or QT as front end.  So
apt-src would inform you that you need X for it to
work and the rest to get the fancy bits.
 
> I could add an option to apt-src to make it not
> resolve any build
> dependencies, and just warn about missing ones or
> something.

That would be OK, although maybe a bit painful if
there are a lot.  Again the required vs. optional
would need to be made clear.
 
> > 5) Not sure it this is possible, but maybe
> downloading
> > source directly from author's site??
> 
> Not something I want to touch. Maybe if everyone
> added watch files to
> their packages, then use uscan.

Fair enough.


Jon

__
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

Re: Are we losing users to Gentoo?

[Tollef Fog Heen]

*  (Thomas Bushnell, BSG)

| But I use the website.  Here's a questions.  Go to eh redhat site and
| see if you can figure out where to get a complete RedHat CD downloaded
| from the net?

http://www.redhat.com -> download -> click the download link besides
«Red Hat Linux 8.0», and if it weren't for the fact that
ftp.redhat.com is a bit busy I'd have half the image on my hard drive
already.

| Of course our site has lots of room for improvement.  I still submit
| it's way better than the competitors.

It's not, sorry.

-- 
Tollef Fog Heen,''`.
UNIX is user friendly, it's just picky about who its friends are  : :' :
  `. `' 
`-  

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Sam Hartman]

> "Christian" == Christian Surchi <[EMAIL PROTECTED]> writes:


Christian> Ari wrote to me in the end of october to ask me about
Christian> my intention about logjam packaging. I had an enormous
Christian> backlog and I could not be able to reply. Then he filed
Christian> a wishlist bug report (#166993) for the new upstream
Christian> release for logjam (4.0.0). Upstream web site
Christian> (http://logjam.danga.com) reports that 4.0.0 is
Christian> released on 29 Oct 2002.  Bug report was sent on 29 Oct
Christian> 2002 (!).

Christian> On 17 Nov 2002 Ari made a NMU for logjam
Christian> 4.0.0+cvs.2002.11.17 and another one a few days after
Christian> that date, IIRC, without a note to me.  I was handling
Christian> bugs for logjam, as you can see in BTS (#165281). Build
Christian> failure reported by Junichi Uekawa in that bug was
Christian> actually a libcurl-dev bug (#169654). I reported and
Christian> maintainer closed with an upload.

So, honestly after reading your message, I'm not quite sure what
you're complaining about.

If you're complaining that the NMU was handled improperly and that the
communication/policy should have been better, then it seems you're
right.  The person performing the NMU has already indicated that they
are sorry they didn't follow policy and so unless you can give
evidence that you think they will continue to fail to follow policy in
future then it seems like an honest mistake.

If you're arguing that the NMU shouldn't have been done then I think I
disagree.  Based on the evidence that you presented, our users and the
free software community were better served by having that package
updated.  And frankly no response from October 29 to mid November
seems like enough time that an NMU is reasonable.  Yes, the NMU should
have been to delayed; yes you should have been contacted.  But other
than your feelings getting hurt, what was the actual harm done to
Debian?

And yes, your feelings getting hurt is a real concern; it sucks to be
a volunteer and to have someone disrespect your work.  But
communication problems do happen and it seems reasonable to treat them
as communications problems and move on with life.

Re: reliable streams over UDP

[Sam Hartman]

> "Russell" == Russell Coker <[EMAIL PROTECTED]> writes:

Russell> Do we have a library in Debian that provides reliable
Russell> stream based communication over UDP?


librx from openafs provides this functionality; it may be somewhat
more complexity than you are looking for.

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Colin Watson]

On Fri, Nov 29, 2002 at 04:41:38PM -0500, Sam Hartman wrote:
> If you're complaining that the NMU was handled improperly and that the
> communication/policy should have been better, then it seems you're
> right.  The person performing the NMU has already indicated that they
> are sorry they didn't follow policy and so unless you can give
> evidence that you think they will continue to fail to follow policy in
> future then it seems like an honest mistake.

IIRC Ari has caused upset with NMUs before; xscreensaver, I believe. (I
express no opinion about whether that upload was a good idea or not.)

-- 
Colin Watson  [EMAIL PROTECTED]

Re: Are we losing users to Gentoo?

[David B Harris]

On Fri, 29 Nov 2002 15:58:51 -0500
Joey Hess <[EMAIL PROTECTED]> wrote:
> The comparison is only fair with organizations that *want* you do do
> so(so not redhat, probably not openbsd, or mandrake, or others whose
> principal developers try to sell cds).

Strictly speaking, given the ISO mirroring situation, we've never wanted
people to use full 640M ISOs when we knew that 99% of people downloading
would use a couple hundreds megs, at most.


pgpYSbDdC8lGr.pgp
Description: PGP signature

Re: reliable streams over UDP

[Peter Palfrader]

On Fri, 29 Nov 2002, Sam Hartman wrote:

> > "Russell" == Russell Coker <[EMAIL PROTECTED]> writes:
> 
> Russell> Do we have a library in Debian that provides reliable
> Russell> stream based communication over UDP?
> 
> 
> librx from openafs provides this functionality; it may be somewhat
> more complexity than you are looking for.

Or you could try to build an encrypted tunnel with cipe. cipe works over
UDP and is surprisingly good at dealing with IP changes as long as not
both peers change them at the same time.

yours,
peter

-- 
 PGP signed and encrypted  |  .''`.  ** Debian GNU/Linux **
messages preferred.| : :' :  The  universal
   | `. `'  Operating System
 http://www.palfrader.org/ |   `-http://www.debian.org/


pgpMBPGzTZSNq.pgp
Description: PGP signature

Re: Pick a name, any name...

[Joel Baker]

On Thu, Nov 28, 2002 at 07:39:02PM -0800, John H. Robinson, IV wrote:
> Ben Armstrong wrote:
>  
> > Bah, that's what CNAME is for.
> 
> that is _NOT_ what a CNAME is for. a CNAME is for when the hostname is
> in a domain that is OUTSIDE of your control.
> 
> ie: evil.debian.org -> www.msn.com = CNAME (we don't control the msn.com
> domain)
> forge.debian.org -> quantz.debian.org = A (we control the debian.org
> domain, so we can save the internet by REDUCING THE NUMBER OF
> UNNECESSARY DNS LOOKUPS AND REDUCE THE END USERS DELAY WITH DNS LOOKUP
> REQUIREMENTS)
> 
> isn't this a FAQ somewhere?
> 
> -john

Except that every major implementation of the DNS protocol in the past ten
years or so has generated a reply that had the A record the CNAME points
to, if it was also authoritative for that record (which, generally, is the
case if it's an in-domain indirection).

And, by the way, that *is* one of the uses of a CNAME. To allow things such
as service names (www, ftp, etc) to point to a single IP, which might have
one of those names, or something else, as it's formal name.
-- 
***
Joel Baker   System Administrator - lightbearer.com
[EMAIL PROTECTED]  http://users.lightbearer.com/lucifer/


pgpzMoBwuSBiG.pgp
Description: PGP signature

Bug#171208: ITP: tlpr -- a Trivial LPR client

[Wouter Verhelst]

Package: wnpp
Version: unavailable; reported 2002-11-29
Severity: wishlist

* Package name: tlpr
  Version : 0.1
  Upstream Author : me
* URL : none (yet)
* License : GPL
  Description : a Trivial LPR client

tlpr is a trivial line printer client that will send a file to a
remote printer. It does not know the concept of a configuration file,
an input filter, or any other form of complexity; instead, it is the
LPR-equivalent of 'cat foo >/dev/lp0'.

-- System Information:
Debian Release: testing/unstable
Architecture: i386
Kernel: Linux rock 2.4.19 #2 di okt 1 01:06:59 CEST 2002 i586
Locale: LANG=nl_BE, LC_CTYPE=nl_BE

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Ari Pollak]

> IIRC Ari has caused upset with NMUs before; xscreensaver, I believe. 
> (I express no opinion about whether that upload was a good idea or 
> not.)

Didn't you sponsor the upload?

Re: Pick a name, any name...

[Thomas Bushnell, BSG]

"John H. Robinson, IV" <[EMAIL PROTECTED]> writes:

> because with a A record one lookup -> IP address(es)
> with a CNAME one lookup -> another lookup -> IP address(es)
> 
> CNAME cause you to at least DOUBLE the DNS lookups, DOUBLING the
> potential timeouts, DOUBLING dns load
> 
> can you do it? certainly. should you be aware of the effects WHEN you do
> it? yes. you may break the rules only after you understand the rules.

There is no rule that CNAME records are only for non-local pointers.
Yes, a CNAME does additional lookups.  DNS traffic to leaf servers
is an insignificant piece of total traffic.

> > DNS entries are cached and don't cause that much traffic.
> 
> the second and subsequent times, yes. not the first time.

A local CNAME doesn't do double traffic.  Suppose I have a CNAME for
www.bar.baz -> fletch.bar.baz.

Looking up an A record in this domain will require:

query for baz.
query for bar.baz.
query for fletch.bar.baz.

Looking up the CNAME is only 33% more traffic, not "at least double":

query for baz.
query for bar.baz.
query for www.bar.baz.
query for fletch.bar.baz.

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Rene Engelhard]

Hi,

Wouter Verhelst wrote:
> Package: wnpp
> Version: unavailable; reported 2002-11-29
> Severity: wishlist
> 
> * Package name: tlpr
>   Version : 0.1
>   Upstream Author : me
> * URL : none (yet)
> * License : GPL
>   Description : a Trivial LPR client
> 
> tlpr is a trivial line printer client that will send a file to a
> remote printer. It does not know the concept of a configuration file,
> an input filter, or any other form of complexity; instead, it is the
> LPR-equivalent of 'cat foo >/dev/lp0'.

Sory, but why not simply use "cat foo | lpr" ?

I do not see a need for this package.

Maybe you can enlighten me?

Regards,

Rene

-- 
 .''`.  Rene Engelhard -- Debian GNU/Linux Developer
 : :' : http://www.debian.org | http://people.debian.org/~rene/
 `. `'  [EMAIL PROTECTED] | GnuPG-Key ID: 248AEB73
   `-   Fingerprint: 41FA F208 28D4 7CA5 19BB  7AD9 F859 90B0 248A EB73


pgpFgqKTfIjxd.pgp
Description: PGP signature

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Rene Engelhard]

Rene Engelhard wrote:
> Sory, but why not simply use "cat foo | lpr" ?
> 
> I do not see a need for this package.
> 
> Maybe you can enlighten me?

Hmm. After thinking about it more, I see:

cat foo | lpr does not work for raw data (e.g. PS files for PS
printers).

echo foo > /dev/lp0 does.

Am I right?

Regards,

Rene

-- 
 .''`.  Rene Engelhard -- Debian GNU/Linux Developer
 : :' : http://www.debian.org | http://people.debian.org/~rene/
 `. `'  [EMAIL PROTECTED] | GnuPG-Key ID: 248AEB73
   `-   Fingerprint: 41FA F208 28D4 7CA5 19BB  7AD9 F859 90B0 248A EB73


pgpylQ2z1NWqd.pgp
Description: PGP signature

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Colin Watson]

On Fri, Nov 29, 2002 at 05:50:36PM -0500, Ari Pollak wrote:
> Colin Watson wrote:
> > IIRC Ari has caused upset with NMUs before; xscreensaver, I believe. 
> > (I express no opinion about whether that upload was a good idea or 
> > not.)
> 
> Didn't you sponsor the upload?

No. Perhaps that was Colin Walters?

-- 
Colin Watson  [EMAIL PROTECTED]

Re: New maintainer behaviour with NMU and LogJam's hijacking

[Colin Walters]

On Fri, 2002-11-29 at 17:50, Ari Pollak wrote:
> > IIRC Ari has caused upset with NMUs before; xscreensaver, I believe. 
> > (I express no opinion about whether that upload was a good idea or 
> > not.)
> 
> Didn't you sponsor the upload?

No, that was me...

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Amaya]

Rene Engelhard dijo:
> echo foo > /dev/lp0 does.

Can you easily do that on a remote printer?


-- 
 .''`.I'd like to fly, but my wings have been so denied...
: :' :Sand rains down and here I sit - Alice in Chains
`. `'Proudly running Debian GNU/Linux (Sid + 2.4.19 + Ext3)
  `-www.amayita.com  www.malapecora.com  www.chicasduras.com

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Wouter Verhelst]

On Sat, Nov 30, 2002 at 12:11:21AM +0100, Rene Engelhard wrote:
> > tlpr is a trivial line printer client that will send a file to a
> > remote printer. It does not know the concept of a configuration file,
> > an input filter, or any other form of complexity; instead, it is the
> > LPR-equivalent of 'cat foo >/dev/lp0'.
> 
> Sory, but why not simply use "cat foo | lpr" ?
> 
> I do not see a need for this package.
> 
> Maybe you can enlighten me?

* It's small. The whole of C code in this package is less than 6k.
  That's three pages.
* it does not require a /etc/printcap (which, IMHO, is written in the
  most horrible format ever invented)
* it does not use spool files, so still works when the file system is
  mounted read-only.
* it's safer, as it doesn't require a daemon on the local machine to
  run. It just opens a connection, sends it away, and exits. Of course,
  that also implies that you can't use it to print to local printers.
* It does not try to be 'smart', fucking up your already-filtered input
  file (which happened to me once too many).
* ... maybe other things.

That said, the version number may suggest you that I don't see this as
finished code. It works, I'm sure it has some uses already, and it will
have more in the future.

Convinced?

-- 
wouter at grep dot be

"Human knowledge belongs to the world"
  -- From the movie "Antitrust"

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Alan Shutko]

Rene Engelhard <[EMAIL PROTECTED]> writes:

> Sory, but why not simply use "cat foo | lpr" ?

Or "rlpr", if you don't want to set up a printcap.

-- 
Alan Shutko <[EMAIL PROTECTED]> - In a variety of flavors!
Something came out of my.BUTT!

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Alan Shutko]

Rene Engelhard <[EMAIL PROTECTED]> writes:

> cat foo | lpr does not work for raw data (e.g. PS files for PS
> printers).

Huh?  It works fine, depending on how you set up your filters.

-- 
Alan Shutko <[EMAIL PROTECTED]> - In a variety of flavors!
Parents have eyes in the backs of their heads.

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Wouter Verhelst]

On Sat, Nov 30, 2002 at 12:18:07AM +0100, Rene Engelhard wrote:
> Rene Engelhard wrote:
> > Sory, but why not simply use "cat foo | lpr" ?
> > 
> > I do not see a need for this package.
> > 
> > Maybe you can enlighten me?
> 
> Hmm. After thinking about it more, I see:
> 
> cat foo | lpr does not work for raw data (e.g. PS files for PS
> printers).
> 
> echo foo > /dev/lp0 does.
> 
> Am I right?

Ah, yes, you're right. I didn't even think about that ;-)

-- 
wouter at grep dot be

"Human knowledge belongs to the world"
  -- From the movie "Antitrust"

Re: Bug#171208: ITP: tlpr -- a Trivial LPR client

[Rene Engelhard]

Hi Amaya,

Amaya wrote:
> Rene Engelhard dijo:
> > echo foo > /dev/lp0 does.
> 
> Can you easily do that on a remote printer?

Of yourse you can't. I was trying to regognize why this package makes
sense...

At least for local printers (PS raw to ps printers) this may make
sense.

Regards,

Rene
-- 
 .''`.  Rene Engelhard -- Debian GNU/Linux Developer
 : :' : http://www.debian.org | http://people.debian.org/~rene/
 `. `'  [EMAIL PROTECTED] | GnuPG-Key ID: 248AEB73
   `-   Fingerprint: 41FA F208 28D4 7CA5 19BB  7AD9 F859 90B0 248A EB73


pgpFv43WNf903.pgp
Description: PGP signature

Re: reliable streams over UDP

[Brian May]

On Fri, Nov 29, 2002 at 02:16:20PM +0100, Russell Coker wrote:
> Do we have a library in Debian that provides reliable stream based 
> communication over UDP?
> 
> I want to be able to deal with asymetric links and end-points that change IP 
> address so TCP won't work.  Surely someone must have written something 
> similar to TCP but implemented on top of UDP.  Is such a thing in Debian?  If 
> not is there a DFSG compliant library that I can package?

I assume you have a way of notifying the remote peer when the
local receive IP address changes? Otherwise, it won't know where
to send packets.

Also, I assume that it is not possible to set the "from" IP address in
transmitted packets so it is the same as the receive IP address?

(this is what happens on one way satellite links I have configured).
--
Brian May <[EMAIL PROTECTED]>

Re: reliable streams over UDP

[Glenn McGrath]

On Fri, 29 Nov 2002 14:16:20 +0100
Russell Coker <[EMAIL PROTECTED]> wrote:

> Do we have a library in Debian that provides reliable stream based 
> communication over UDP?
> 
> I want to be able to deal with asymetric links and end-points that
> change IP address so TCP won't work.  Surely someone must have written
> something similar to TCP but implemented on top of UDP.  Is such a thing
> in Debian?  If not is there a DFSG compliant library that I can package?
> 

Have you looked at l2tp, its like ppp but you can move the end point
(something like that)



Glenn

Re: Pick a name, any name...

[Martijn van Oosterhout]

On Fri, Nov 29, 2002 at 01:11:34PM -0800, John H. Robinson, IV wrote:
> because with a A record one lookup -> IP address(es)
> with a CNAME one lookup -> another lookup -> IP address(es)
> 
> CNAME cause you to at least DOUBLE the DNS lookups, DOUBLING the
> potential timeouts, DOUBLING dns load

Not necessarily. It depends a bit on the intelligence of the DNS server and
the client. A DNS response has an "additional section" for stuff like this.
If you so a "dig net. ns" you will see that not only are you provided with a
list of nameservers in the answer section, but also all their IP addresses
in the additional section. Thus avoiding the second lookup.

The same trick can be applied to CNAMEs. It also applies to MX records
(which may only specify the name of a server, not IP address).

Try:

dig freshmeat.net. mx
dig www.freshmeat.net

for more examples.

> can you do it? certainly. should you be aware of the effects WHEN you do
> it? yes. you may break the rules only after you understand the rules.

There is no rule. The designers of DNS thought of this and solved it.

-- 
Martijn van Oosterhout  http://svana.org/kleptog/
> Support bacteria! They're the only culture some people have.


pgp0Kn4pZrzUN.pgp
Description: PGP signature

Re: reliable streams over UDP

[Shyamal Prasad]

"Russell" == Russell Coker <[EMAIL PROTECTED]> writes:

Russell> I want to be able to deal with asymetric links and
Russell> end-points that change IP address so TCP won't work.

TCP works well with asymmetric data flows. I think you should consider
this carefully before dropping TCP for this reason.

For IP address changes I presume your application will require some
signaling to determine what the new IP addresses are? If so, be sure
to consider how much more expensive this is than setting up a new TCP
connection each time (three packets + slow start to get to max flow
capacity). 

In most applications I have found TCP is good enough. The only thing
TCP is very poor at dealing with is "head of line blocking." However
this is not a problem that is easily solved with UDP hacks either if
you actually do worry about congestion control and fairness.

I *am* second guessing you, but do think about it.

Russell> Surely someone must have written something similar to TCP
Russell> but implemented on top of UDP. 

Too many people have tried this ;-) Try SCTP, a recent attempt to deal
with the "reliable UDP" solution:

ftp://ftp.rfc-editor.org/in-notes/rfc2960.txt
ftp://ftp.rfc-editor.org/in-notes/rfc3309.txt

and for some implementations look around at

http://www.sctp.org/

Good luck!

Shyamal

Re: Pick a name, any name...

[Georg Lehner]

Hello!

El vie, 29-11-2002 a las 16:38, Joel Baker escribió:
...
 
> And, by the way, that *is* one of the uses of a CNAME. To allow things such
> as service names (www, ftp, etc) to point to a single IP, which might have
> one of those names, or something else, as it's formal name.
...
What would be wrong with multiple A records in this case?

Regars,

Jorge-León

P.S.: feel free to reply off-list, that is not about debian-devel.