Re: Rescue disk hurts my ethernet card
On Fri, 13 Jun 1997, Jason Gunthorpe wrote: > On Thu, 12 Jun 1997, Bruce Perens wrote: > > > fit in the kernel. If you can tell me about the I/O ports of your network > > card we can give you magic words to put on the boot command line for that > > device that reserve its ports and prevent other drivers from touching them. > > Fraid the reserve boot command did not help this problem, it's EEProm IO > ports may not have been in the region I had the card set to or one of the > drivers may be ill behaved. I've struggled with some software configurable isa ne2k's. They appeared to function quite nicely with linux, but win95's autoprobing had severe difficulties with the cards. The dos configuration utility would let me change io and irq settings just fine, I could also write them to the "register" but not to the card's eeprom, giving me an error in the card's own setup utility. The problem appeared to be that the cards used a different combination of irq and io for configuration and actual operation. It listened only half the time to the operation addresses and the configuration addresses were conflicting with other hardware. This kept me stumped for quite a while and until I figured it out I too had long bought pci cards (which also gave some trouble .) This was the solution in my case: -on the motherboard bios setup, disable com2:, freeing irq 3. -with the dos utility, set the io and irq to what the card uses to receive programming data. -then set it to what would suit the computer's available interrupts and io addresses. Don't forget to write the settings to the card's eeprom. In your case, the conflicts may ofcourse be a little bit harder to resolve before you can access the card again. Maybe it doesn't work altogether on your card. Maybe it's really burnt. You can always try. >I've long since gotten rid of it, now have a RealTek PCI NE2k. Hmmm, I bought winbonds. They don't work without a patch. It would be great if this patch would make it to the next bootfloppy, or even better - 2.0.31. Until then, I have to make my own rescue-floppy kernels. > I heard of two fried network cards, one was mine and one was a friends, > both D-Link DE-250's and both fried by booting the rescue disk. My other > De-250 just got disabled during linux's boot, had a bit of a time figuring > out why Linux didn't want to use my card till I figured it out. Please check my solution if that may help you, if the reprogrammed io and irq aren't too outrageous there may still be hope. I think that it is however possible to fry hardware with linux: while trying 1.3 I inserted a wrong module for the cdrom interface and it fried the cdrom drive. Good luck, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: Shadow passwords under X...
On Tue, 10 Jun 1997, Curt Howland wrote: > I have shadow passwords turned on, under Debian 1.3. However, > the X login started automatically at boot does not recognize > any of the valid userID/password combinations. You're probably referring to the "xdm" login. > My assumption is that X does not know about shadow passwords, > and from my investigations it seems I may be correct. This is indeed correct. The problem is, that there is a separate "xdm-shadow" to handle shadow passwords. The bug is with xbase, which does not properly install xdm-shadow when you've enabled shadow passwords. If you first install xbase and then turn on shadow support, this problem does not arise. > 1) Is there a setting to make it work? The solutions are: - change to the virtual console with Ctrl-Alt-F1, log in as root and type shadowconfig off but then you lose shadow support - change to the virtual console with Ctrl-Alt-F1, log in as root and type /etc/init.d/xdm stop # because you can't update it when it runs shadowconfig on # makes xdm xdm-shadow /etc/init.d/xdm start# if this were windows, you'd now reboot :-) - wait for the release of 1.3.1, which will have a fixed xbase. > 2) If !1, then how do I turn off the automatic X boot-up >so I can startx after logging on? "man xdm". No, seriously, look in /etc/X11/config, you'll find the lines start-xdm xdm-start-server These lines were placed there by the xbase configure script when you answered "yes" to the question to run xdm. Anyway, comment those lines out with a "#" and xdm is gone. If you later decide that you want xdm back, you can always just uncomment them. When you have x running, be sure to install tkman. Makes reading manpages a lot more comfortable. Good luck, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: lpr: unable to get official name for local machine (fwd) fixed.
Name resolving works again on my machines! Still don't know what exactly happened though. It seems that the reason for bind not working out of the 1.3-box for me, is that I answered '' where I should have answered 'none' to bindconfig (at least that is the only difference that I can think of.) Looking at the files, the big difference was a missing '@' in front of the line for localhost. When I placed the '@' back in the configuration file on the machine that had a working named before the 1.3 upgrade, it worked again as well. Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: rlogin breaks terminal console
On Mon, 9 Jun 1997, Alex Yukhimets wrote: > > > I'v got a strange phenomenon: sometimes when I close the rlogin > > > connection started from console (not xterm), the console appeared to be > > > broken - all the output is confined in the last line of the screen. > > > > I've encountered the same problem. It often seems to be related to the > > use of "pine" on a remote (rlogged-into) machine. I've had the same kind of problem lately when: -in an xterm dialing in with minicom to my isp -from the isp's shell telnetting to another machine -running pine.. It (output confined to bottom line) only happens after I've tried to resize the screen parameters with 'stty rows ' on the isp's and other machine. Ususally, pine will work more or less (screen gets garbled al too easily though.) I'm quite baffled by this behaviour. Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: Newbie
On Fri, 6 Jun 1997, Rick Morrison wrote: > 2. Any tips as to how to start NFS services for mounting from my PC /Win95 > client? If you want to interface linux with MouseDriver95 then forget NFS and install the samba suite; it does SMB aka MS LanManager aka "Microsoft Networking" over TCP/IP. Good luck, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: lpr: unable to get official name for local machine (fwd)
On Thu, 5 Jun 1997, Nathan E Norman wrote: > For what it's worth, I upgraded from 1.2 to 1.3 about a week ago on a > server that is authoritative for a few zones ... no problems at all. I forgot to mention the following: I let the install script change my existing setup, did you do this too? > I realise this doesn't assist you in solving your immediate problem, but > I'm not so sure that the Debian package is to blame.. Well, I'm not really authoritative on bind matters, but I thought that on previous ( < 1.3 ) installations, you could issue "nslookup localhost" and it would work then. Now I get: calypso# nslookup localhost *** Can't find server name for address 127.0.0.1: Non-existent host/domain *** Default servers are not available -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: lpr: unable to get official name for local machine (fwd)
To me it looks like nameserving doesn't quite work out-of-the-box in upcoming debian 1.3. I've upgraded a box to frozen and that broke a working bind configuration, I don't know how badly because I haven't really had the time to look into it. On fresh installations, nslookup can't find even localhost. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: new kernel
On Wed, 4 Jun 1997, Fredrik Ax wrote:
> On Tue, 3 Jun 1997, Syd Alsobrook wrote:
>
> > Hello,
> > I have a machine that i just upgraded to v1.3 (that went perfectly). My
> > question is this I want to upgrade the kernel (now v2.0.27) the easiest way
> > possible can I copy the kernel from the install floppies for v1.3 (has
> > kernel 2.0.30) or is it easier just to recompile the kernel?
> >
> It would probably be easier to copy the kernal, BUT I still recommend you
> to recompile the kernal. Then you will get a kernel that is optimized
> for your needs, which doesn't include alot of non-used drivers.
I rolled a kernel on an existing debian machine using the kernel-package:
cd /usr/src/linux
make {,menu,x}config # configure ramdisk and initrd
make-kpkg clean
make-kpkg -revision custom.install.boot.1 kernel-image
after that I got a /usr/src/kernel-image-2.0.27_custom.install.boot.1.deb
(or something like that) that can be installed with dpkg -i.
I took this file to the new machine with base already installed and tried
dpkg -i kernel-image-etc.. This did not work as I expected because some
links were missing in / (/vmlinuz -> /boot/vmlinuz and /System.map ->
/boot/System.map) After I had made these links it installed fine.
The next installation attempt I just put the custom kernel on the
resc1440.bin diskette as linux (hence the compiled in ramdisk and initrd
support.) This way, your own kernel gets installed during the base install.
Of course, you have to have an existing debian box to prepare a custom
kernel, so if you don't, you'll have to install the kernel-source and gcc
packages (and quite a few others) to brew one yourself on the new
machine. Make sure you also get the kernel-package package, the special
debian tool for this purpose (you don't have to but it is really nice.)
Have fun,
Joost
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
[EMAIL PROTECTED] . Trouble?
e-mail to [EMAIL PROTECTED] .
Re: XDM not allowing logins
On Tue, 3 Jun 1997, Jason wrote: > I just install Debian 1.3 on a fresh system and set XDM to load on startup > like I have done before. But XDM isn't allowing me to logon using root or > my normal account. It just tells me that the password is incorrect. Does > this have something to do with XDM not knowing about shadow passwords? Exactly (sort of.) Until this is fixed in xdm, turn off shadow passwords: Ctrl-Alt-F1 log in as root and type: shadowconfig off return to x (Ctrl-F7) you can log in again Good luck, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: Help me to build a fileserver!!!
On Tue, 3 Jun 1997, Eliezer Figueroa Puello wrote: > > I want to build a small network using Linux as the fileserver and > windows 95 computers as the clients. What I have in mind is something > similar to what novell network do, but using Linux in the place of the > novell server. I want to be able to map a linux partition in the server > to work as a directory in the windows 95 computers. I tried to do that > using the tcp/ip protocol but then I got realized that windows 95 do not > have a client for UNIX networks using the tcp/ip protocol so, I have > been unable to map. Are you _very_ sure that there exists no nfs client for win95? Anyway, what you want can be done in other ways, novell/ipx being one of them. As I understand you can even have your linux box act as a novell server using the mars suite. I have no experience with it though. I suggest you have a look at the samba suite, which does the SMB protocol - the "microsoft networking" protocol. You can mount windows (3.11, 95 and NT) diskshares on the network on your linux box with smbmount //doze-box/sharename /mountpoint -n and reversely you can export linux filesystems (requires a little setup.) Also you can let linux be a "windows-nameserver" and - if I'm well-informed - maybe soon there will be support for NT-style authentification services. Printers can be shared both ways as well. Good luck, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: Help: Problem in setting up network
On Sun, 1 Jun 1997 [EMAIL PROTECTED] wrote: > I am trying to connect my compute to a LAN in the lab. > The network card (NE2000 IRQ 03 0x300) works fine and is detecked by the > kernel. After setting up all the stuff (/etc/hosts, /etc/resolv.conf, > /etc/sysconf/network, ...), it still cann't recognize other > computer when I use 'ping'. However, I noticed that when > setting up, there is a message saying there is something wrong with > 'SIOCADDRT'. And in file /proc/net/arp, all the hosts (gateways, > nameserver) all have hw-address to be 00 00 00 00 . So, what is the > problem? If you can't ping any hosts, your network interface probably doesn't work properly. If your kernel detects the presence of a network card, that doesn't necessarily mean that the kernel can make the card work. The error message with "SIOCADDRT" is a clear sign of such a problem. > Also, by setting the ether card, the mouse doesn't work anymore. Is > this because of the ether card (IRQ 3)? is there any utility to > detech this problem? You have a interrupt conflict between your ne2000 and serial port. They both try to own irq 3. Since it is generally very hard to reassign the serial port irq, you'll have to try to reassign the ne2000 irq. This might be as easy as changing a jumper or running a "setup" program on the driver diskette that is included by the manufacturer (unfortunately, these often require you to be able to boot into dos.) Sometimes you need to remove the interrupt conflict before you can reprogram the ne2000. In that case, you have to temporarily disable the serial port in your bios setup (or move the ne2000 to a machine that will allow you to do so if this one doesn't.) Maybe the card is plug-'n-pray. In that case check out the isaplugnplay package. Good luck, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
*-* How to make dselect use local mirror for testing frozen?
Hi, On my home lan I have a couple of machines that I am currently using to test the new frozen distribution on. Because I don't want to have to download all the packages that I want to test each time I try a new machine or method, I have made a little mirror of part of the distribution. It has some frozen/binary, contrib, non-free and non-US packages. I made a tree structure like on ftp.debian.org and put in some symlinks and the Packages files. Dselect does most things I expected; when I point it at /debian, it finds the all packages files and I can choose from frozen, contrib, non-free and non-US (which I linked local.) But at the point where it should install the selected packages, dselect only iterates through stable/binary and never installs anything from contrib or non-free. I know it should do that, because it does so on my 1.1.4 cd (never tried it "live" over the internet.) Can anyone give me some clue please? Thanks. Cheers, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: Installing Debian Linux on a new system w/out MSDOG
On Mon, 19 May 1997, Dale Scheetz wrote: > On Mon, 19 May 1997, Jean Pierre LeJacq wrote: > > > I'd recommend creating a partition for MS-DOS right away. > > No need to actually install it till required. > > Agreed. Dos/Win/NT want the first partition of the first drive to be > "their" boot file system. So decide what you want for DOS and build the > partition (from a DOS rescue disk that has fdisk) with fdisk (msdos > version) before you do any linux work. Well, my /dev/hda1 has an ext2 root partition on it _and_ is marked bootable. Windows95 is on /dev/hda2 and has no problems with that, contrary to popular belief and many manuals. If you have a floppy to boot linux from in any case, the worst that can happen is that you can't boot your freshly installed windows95. That's bad, because it takes a lot longer to install and setup than debian linux. If you're a bit paranoid about your existing setup you can always make a couple of backups of the first 512 bytes of all primary, extended and logical partitions and the master boot record before you let the microsoftware touch your system. That's what I actually did, but I don't think it is necessary at all. Adding an entry for windows in lilo.conf and rerunning lilo is enough. Cheers, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: D-Link 220 card detection problems??
On Sat, 3 May 1997, Brian Freeze wrote: > I am trying to install the latest debian release onto a system that had > slackware running on it. It has a dlink 220 ethernet card in it and was > running fine with the slackware system and also with win95. I have tried > every module that was installed with the base system and nothing will work. > > Is there another module I could download for this card? In the Ethenet > How-To it said to use the ne2000 driver but the module doesn't work. > > Any help would really be apreciated. Hi Brian, A more descriptive report of your situation would help helping a little. For instance, what does the kernel print on bootup? Does it see your card at all or as something else than a ne2000 or even a network card? What do /etc/modules and /etc/conf.modules look like? Does the latter one include lines alias eth0 ne options ne irq=10 io=0xe400 like mine? I also commmented out all the lines in /etc/modules and added auto to enable the kernel daemon, which loads any necessary module when needed. You have to fill in the irq and io values for _your_ card, of course. If the card works in win95, you can peek in the system panel for the apropriate values. You can configure these settings in debian during the modules configuration stage of installation (if I remember correctly.) Just add the irq= and io= statements as parameters when you select the ne2000 module. Alternatively, you can edit the /etc/conf.modules file by hand after the base installation. For the kernel daemon I think you have to edit /etc/modules in any case - if I am correct. You can even pass some options to the kernel at boot time (at the lilo prompt, that is.) Don't know the details on that, never had to use it. If all of this doesn't help you, there's one more option: visit the Donald Becker site, address is in the Ethernet-HOWTO. You can find some links to patches for unusual cards here. I had to use a patch to get my sort-of-broken pci ne2000 to get recognised as such by the kernel. After I rebuilt the kernel, it recognised my card and the card worked fine with linux. Cards that need patches to be applied to the kernel source are exceptions, but you might just have such an exception. Hope this helps you to get going, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .
Re: Audio, Printer queue and Mouse button
On Wed, 30 Apr 1997, Daniel Karlsson wrote:
> 3) How can I make my middle mouse button work. I've tried all options in the
> configuration, but none of them works.
If you have read the appropriate MINI-HOWTO and tried all possible
combinations of mouse type * {gpm,XF86Config} options, killed gpm,
restarted X and unplugged the mouse and plugged it back in again with the
button depressed, then you know what I too have been through.
My three button mouse claims to be "mouse systems compatible" tho (ahum)
I found that I could get the middle button working only if I either:
-rebooted the pc holding the (left?) mouse button depressed;
-unplugged the mouse and plugged it back in with the button depressed;
-took up the soldering iron and connected two pins of the controller ic
in the mouse (this is described in the MINI-HOWTO.)
The first and second options are pretty annoying. The third works great for
linux, however the mouse could not be used in windos anymore.
In the end I decided that I had had a good afternoon's fun and had picked
up some fair skills in rodent epidemiology and -heart surgery, but bought a
Happy Mouse from Genius anyway.
They're quite reasonably priced and work out of the box on both linux and
windos. If you can find a mouse with a 2/3-button switch, those work fine
too, I guess.
Good luck,
Joost
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
[EMAIL PROTECTED] .
Trouble? e-mail to [EMAIL PROTECTED] .
Re: Problems working with bash.
On Mon, 31 Mar 1997, Karl M. Hegbloom wrote: > > "Gertjan" == Gertjan Klein <[EMAIL PROTECTED]> writes: > > Gertjan> "Karl M. Hegbloom" <[EMAIL PROTECTED]> wrote: > >> Open up your info reader; > > Gertjan> Don't get me started on info! > > Why not? Elucidate. I think info sucks. It is obsoleted by html. Clear enough? Just my 2 cents, Joost Karl M. Hegbloom <[EMAIL PROTECTED]> http://www.inetarena.com/~karlheg Portland, OR USA Debian GNU 1.2 Linux 2.0.29t You tell me and we'll both know.
Secret debian lists? (was: Debian Book list)
On Sun, 30 Mar 1997, Jaldhar H. Vyas wrote: > I am terribly sorry, the subscribe address I gave was debian-doc-request > (no s) @lists.debian.org. The address to post to was correct. So once > again it is: > > [EMAIL PROTECTED] -> subscribe/unsubscribe address > debian-doc@lists.debian.org -> address to send mail to. Yes, it works for me now. Did I really stumble on that s? Hmm, well that's what you get when you do things with the mouse. Still, while the list does exist, it doesn't appear on the list of lists that comes with the subscription failure notification. Which raises the question: what happened to debian-admintool (for rantings and whinings about dselect)? It was mentioned some time ago, but it doesn't appear on the list of debian-lists either? Cheers, Joost
Re: Debian Book list
On Sun, 30 Mar 1997, Jaldhar H. Vyas wrote: > The developers have realized more and better documentation is needed. Did > you know there is now a mailing list for discussing this type of thing? > ([EMAIL PROTECTED] the subscription address is > [EMAIL PROTECTED]) This might be a better place to > discuss a Debian book. (which is something we desperately need IMO.) This certainly doesn't work for me. I just get a list of all valid debian related lists and debian-doc isn't one of them. And it is still one day before april fools (just checked :-) Cheers, Joost
Re: Installing debian 1.2 from december 1996 InfoMagic 6 CD set
On Tue, 18 Mar 1997, Laurent PICOULEAU wrote: > I bought InfoMagic Linux Developper's Ressource 6 CD set od december 1996. .. > So what should I download to fix this apparently bad organised CD. This is unfortunately a FAQ. Someone was kind enough to write a little document about it. Point your browser to: http://ece.wpi.edu/~rulnick/GlinuX/debian-1.2-faq.html It is a list of problems and fixes for the 1.2 distribution on the Infomagic december 96 6 CD set. Hope this will get it right for you.. Have fun, Cheers, Joost
Re: Ideal partition sizes.
There's a somewhat short mention of cluefull disk partitioning in the debian-faq, but I found much better discussion of it in the "Multiple-disks-HOWTO" > I have two machines. I dislike extended/logical partitions. I like > performance tuning. My machines have at least two drives in them. All > that aside, here's the scoop: [ examples snipped because anyone interested probably kept the message ] > For the most part, this scheme works exceptionally well. My spare > partition holds a complete (and completely separate) linux install (single > partition, just like you had). If I should ever trash my main install, I > can boot into the spareland and run a tape backup of whatever's left - > very handy to have around. This is actually a great idea if you can spare the space. If these partitioning matters are an item for you, then you probably can. I don't know if it is with a single 1.2M disk though. > Both of my swap areas are directly between critical system areas (/ and > /usr, /tmp and /var), making swap an easy trip from any typical program, > logfile access, or tempfile. The biggest performance hit comes when > sendmail is delivering to my mailbox (which is in /local) and is logging > to /var at the other end of the disk, but some recent sendmail tunings > have really helped that. You might have been better off with /home in the place of /. But see the "Multiple-disks-HOWTO" for a thorough elaboration on that. I have my /usr/libs on a partition with /lib and made /usr/lib a symlink. This is a performance issue (well actually more of an experiment :) Other valuable contemplations on partitioning can be found in the FileSystem Standard documentation. Anyone wondering if and how he should partition his disk is IMHO best to read both the FSS and the HOWTO. Again, if you've only got one disk, there is probably only the argument of having a complete backup root filesystem that can make it interesting. Joost
Re: kernel-config
On Mon, 10 Mar 1997, Mikael Hallendal wrote: > Hi! > > I try to get the sound working i linux and when I do make config in th > kernel-source I'm aksed > to enter the I/O-base but the only thing that happens is that it says > 'no help is available. > > Can anyone help me, please! I find your question not very clear: Do you mean to say that the config script hangs at the point where you should be able to enter the address of your soundcard's IO-port? What version of the source do you use? Or do you mean that you need help finding the right address number to enter for your card? What soundcard do you have? Joost
Re: Installing a new kernel....
On Sat, 8 Mar 1997, Bjoern Starke wrote: > Hello, > > today i tried to install a new kernel. After making xconfig, make > zImage and running lilo Did you issue "make mrproper" before "make dep" and "make zImage"? Your output indicates a problem with modules. After makeing a new kernel image, you have to also make new modules with "make modules" and "make modules_install". It is not clear from your message if you did this as well. When that has all been done, do a "rm /lib/modules/modules.dep" before "depmod -a". See if that gives better results. Even better, get the kernel-package. It is a really cool util. Not only can you use it to register the kernel-source and kernel-image to the debian system (very convenient if you patch or customize it), but it also greatly simplifies building kernels. With the kernel-package you won't have to bother anymore about all the targets to make. Just make your favorite config and run "make-kpkg" (read the man page of course for the proper parameters.) Cheers, Joost
Re: Idea for the burner
(sorry for the bad cc: people, debian-admintool@lists.debian.org doesn't exist, I just found out) On Thu, 6 Mar 1997, William Chow wrote: > > On Fri, 7 Mar 1997, J.P.D. Kooij wrote: > > > > Like, dselect is a great and easy tool, but there is no manpage for it... > > Dselect is pretty intuitive once you get the hang of it, but it needs to That's like saying: The water's pretty warm when you're swimming in it for a while." Anyway, I believe that for first use dselect is in fact intuitive enough, that is not my point. > LOT of packages in the Debian system. While it may be a headache to run > dselect and go through all those packages, I guess it's better than having > very few packages... No complaints from me about that aspect of dselect either here. It's not really that essential. Besides, my pc is fast enough (could be better thoug - both) Here _is_ my point: dselect is "all or nothing" and has no documentation. > > That is a BUG, people. I have spelled the dpkg manual to get a hunch > I'm not quite sure that dselect doesn't have documentation. The package is Well, I finally found some: "dpkg programmers' manual - chapter 14: dselect's interface to its installation methods". Not a really place I'd look into for some basics on dselect if I were a newbie and not yet fully at ease with all the difficult switches and options in the dpkg manpage. > However, I agree, it should have a complete > man page if it doesn't already. This is just "professional." Manpages (and maybe, ok, info) are for people who have at least some sense of orientation in unix. Dselect should be even more conservative in its estimate of the user. This is a sidetrack however. > I tend to do this too. The thing is that dselect is too unwieldy to use > for small upgrades/installations of specific programs. It's too much of a > pain to configure it to install one package... The problem isn't with a > lack of dselect documentation, IMHO, but due to the way dselect is setup. Right. You have to do some wizzo tricks with dpkg that aren't even told in its manpage before you can use the user friendly front end to dpkg. That's kind of silly. _If_ it has to be like this, why not abolish dselect and give people a fully-scripted standard installation and let them figure it out with dpkg for all the rest. I'm not trying to say that this is what should be, it's just pretty much what I had to do. > > .. Yet, from what I've read it [ menu ] is a wonderful package, > > integrating other packages the way the debian system can do so fine. The > > way a structured documentation should be (and I believe most building blocks > > for something like it are already there.) > Documenting every available package would seem unwieldy. I'm afraid that > this will be more the rule than the exception, when useful packages get > lost in the sea of packages available. One could just say all that as well about the packaging and dependency system. Yet it works. It works very well and quite painless once the framework is up. It is even creating an added value with the all the scripted configuration options that come with the packaging scripts. Or the menuing system. IMHO, if that sort of thing could be done for the documentation, then that would be much more valuable than the nice menu thing. > Tips and Hints FAQ/HOWTOW/book/whatever that collects little bits of > useful info that don't necessarily correlate with one another. And keep it up to date and relevant, _just_like_the_packages_ they should come along with. Like there is a dpkg for the packaging there should be something similar for the documentation (maybe as an integral part of dpkg.) However, the current info program and documentation style clearly doesn't provide or even possibly support this. > essentially the same document. Not everyone will have lynx, and not > everyone will use texinfo, etc. I know what I would put my money on though... (O'Reilly Associates Press shares probably anyway. Who says this linux thing is free unix? These books do cost, man. :-) Cheers, Joost
Re: Idea for the burner
On Thu, 6 Mar 1997, Chad Zimmerman wrote: > .. You go to any book store you see 7 or 8 > books that deal with slackware and redhat, why not Debian? Maybe because debian isn't known widely enough. Maybe that is because too many books don't mention it at all. The Infomagic booklet in the cd-set I installed from doesn't for instance. I heard about debian and what it is because I asked an apparent guru on irc about the slackware bugs (glad I didn't ask this sort of question when 1.2 was just out..) I chose debian because of all the volunteers working on it, as volunteers, which is one of the reasons I wanted to play with linux anyway, I like that state of mind. Now, if I hadn't already had net access and if I hadn't asked "which" to a guru I would probably not have found my way to debian. That is really bad because IMHO debian is _naturally_ the best linux, besides being GNU as well. Debian needs good publicity. The RC5 thing should be on behalf debian.org. Better still, lets have packages for it and show that off to the other linuxers. Why would they want to complain about that? Let them join debian if linux is their concern. > If/when I do start on this, it won't be untill the fall probably... have > another book project I am finishing up. One of the great things in debian is that it moves really fast. That is however not so great for book publishers. If you had written the book now, you'd probably have to rewrite it by fall anyway. For debian it would be a good thing if the general documentation procedures were better organised. Most packages seem to have quite good documentation, its just hard to know where to get started. It would be tops if the coherency of the packaging system could be imposed on the documentation. Like, dselect is a great and easy tool, but there is no manpage for it... That is a BUG, people. I have spelled the dpkg manual to get a hunch of what is going on in dselect and I got used to dpkg. So, apart from the first time at installation, I never use dselect anymore. I guess that was not the intention of dselect's developers. Why isn't there a dselect-HOWTO? (Maybe I should try writing that then instead of this :-) Like, a couple of days ago I just read about the menu package on this list. I bet I am not the only one who didn't have a clue about even the existance of it. Yet, from what I've read it is a wonderful package, integrating other packages the way the debian system can do so fine. The way a structured documentation should be (and I believe most building blocks for something like it are already there.) Also, I think it would be better to drop info (sorry, you die-hards GNUdes out there.) Lets have all the info stuff as lynx-enabled HTML, it provides all the info functionality and any dummy can view it with his browser, even before having linux really working. A much better bet in the long run I estimate. It could also be merged along with configuration scripts (java?) - another fine point of debian by the way - into the dselect front-end. Oh well, it's becoming another "dselect musing" again. Not the right place for it on debian-user. Anyway, for what it's worth: I heard rumors (but just that) of "Running Linux" being rewritten to include info on debian. All I know for sure is that others mentioned this as rumor. Cheers, Joost
Re: csh
On Thu, 6 Mar 1997, Pete Poff wrote: > Hi, > on what ftp site can I get the csh shell and what file name is it > called. At ftp.debian.org of course, you silly ;-) Better still, use a mirror of ftp.debian.org that's nearby to you. If you have a cd with debian, that's even faster. Try the dselect program on your system. (login as root first if you want to install the cshell.) Dselect shows all available packages in the select screen. cshell is in the section shells. Maybe you should read the debian FAQ as well, because your question is really very basic. You'll probably have more questions that are answered quite clearly in the FAQ. You'll find it linked on http://www.debian.org Good luck
Re: turning off computer
I've seen people write some nice scripts to do the job for mortal users. As already noted though: The setuid shell script can be run by any user logged in on your system, either on the console or over the network. As not already stated: Shell scripts that are setuid to root are generally considered specifically risky. It's probably not a real item in your situation anyway. The solution with a file in /tmp is an elegant way around the the setuid-root scripts problems. On the other hand, because /tmp is world-writable and many system processes write to it, it might create it's own backdoors (again, I'm not an expert, but I wouldn't bet that you can be absolutely sure that there won't be a cracker that finds a way to remotely sendmail your machine into reboot without even having to login to it at all.) I think the canonical solution would be to: Create a group rebooters, set the shutdown binary group to rebooters and add any user that is allowed to perform shutdowns to the rebooters group. Then you can wrap the command in any customizable script with all the options and switches you want to give your users as default. I think I like the suggestion about /etc/shutdown.allow the best. Shutdowns should be done from the console anyway. > Of course, if you can trust people to manage to switch off a re-booting > computer during the "safe period" (i.e. after re-boot has started but > before Linux really gets under way) then simply Ctrl-Alt-Del should do it. Also, the options to shutdown in /etc/inittab are IMHO better with -h than -r. Of course, sysadmins don't trust users to know what's safe :-) My pc has a big reset button that I can use when I want to reboot after shutting the system down. Having the machine to just halt on Ctrl-Alt-Del is much more practical in most cases where you just want to shut down without having to log in as root. Joost
Re: Xfree86 Question
On Wed, 5 Mar 1997, William Chow wrote: > > On Wed, 5 Mar 1997, Terry Martin wrote: > > > Could not find config file! > > - Tried: > > /etc/XF86Config > > /usr/X11R6/lib/X11/XF86Config.neverland > > /usr/X11R6/lib/X11/XF86Config > > > > Fatal server error: > > No config file found! > > Note, the X server no longer looks for XF86Config in $HOME > > > Huh? I thought the X server usually looks for this file in /etc/ or the > X11 lib directory or even the bin directory depending on how you set it > up. No, that's not quite right. I just reran xf86config on my 1.1 box with 3.1.2 XFree tonight. When it's nearly finished, the script asks if it's o.k. to save the configuration as /etc/X11/XF86Config and says that is the standard in Debian. If I'm right you also have a /usr/X11R6/lib/X11/XF86Config.eg as an example config file installed with XFree. Don't know if that is of any use with your hardware. > Have you tried rerunning xf86config, or the better XF86Setup program Find the booklets with your monitor and videocard specs (sorry, RTFM) and run xf86config. That should get you started, I hope. Joost
Re: People Demanding Credit in the Press Release Silliness
On Thu, 20 Feb 1997 Bruce Perens CC'd a press release to this list; Between Thu, 20 and Fri, 21 Feb 1997, some people responded: > < arguing snipped by me > > > > < more arguing snipped > > > > > > < arguing snipped > > > > > > > > < sigh.. > > > > > Thank you, Bruce. You made a very good press release. I think it is very readable, maybe even to my neighbour who doesn't even know what an operating system is, but might get an impression of what debian (and linux) is about. You see, most people don't think in C-like structures and don't parse press-statements through a preprocessor to verify if all the includes are allright. If one did include all of it, their stack would probably overflow and in this case it wouldn't give you an access to their system. Moreover, focussing on the honours that certain people would proposedly be entitled to at all circumstance, completely neglects the fact that Bruce doesn't get paid to write press statements, yet does a wonderful job as he did. So please, people, if you don't like it, start contributing better material yourself, because there's no money back here. Joost (well, I'm probably going to be flamed now for continuing to waste bandwith. Serves me right, probably) -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
RE: Some thoughts for Debian
Regarding the debian system Willie Daniel thought out loud: > I've been watching the debates about Cds, dividing up non-free, and etc.. > And I had a few thoughts I wanted to throw out at everyone. > > I think if debian is going to succeed, it's going to have to be reduced > to a standard set of "core" applications that will make up the "offical" > distribution of debian. Right now, the distribution is huge and it's > getting bigger. IMHO that creates not much of an issue. It just creates the need for a second silver dish. This appears inevitable anyway, considering the architectures linux is currently being ported to. Who knows, maybe debian linux 1.5 developers resource will be a 6 cd set and the slackware97 fans will complain that there's no mention of their favorite distribution in the inside leaflet ;-) > In the process, it seems less is getting done in the way of > providing comprehensive testing, marketing, etc.. of each new release. > Many people are handling multiple packages and getting either burnt out > or don't have enough time to adequatly test the packages they do maintain. That is an orthogonal issue, but I must agree that there's a sense of reality in your observation of a problem. > While I understand and applaud the basic concepts behind Debian, it's > success will depend on what the commercial community does with it. I guess that that's not an issue at all. > Therefor, I think the following needs to be done to help Debian get > going in the right direction. > > 1) I realize this process will start a religious debate over what > program is the best, but, someone ( everyone? ) needs to decide > on the core packages that will make up the offcial distribution. > ie smail, sendmail, qmail, etc. which one will debian choose > as the default for the distribution? There have been some things said about this with respect to dselect. IMHO if there should be anything like a "core distribution", it would have to be a system that is minimally confusing to new users. All the other packages are just packages and are equal part of the debian system. I think debian should anticipate more and more packages will appear, each providing more or less the same functionality. > in the case of user apps, one application could be choosen that > works on terminals and one that works in X. ( where possible ) > ie users will need a mail reader. > for terminals: elm > for X: exmh Now, I see we agree :-) Just take that idea all the way. > this will create a standardized distribution that commercial vendors > can easily support. it will reduce greatly the size of the main > distribution and will allow devlopers to focus on just the applications > that make up the offical distribution plus allow easier beta testing > of the distribution as a whole. I dear to disagree again, but think that the point you raise at 4) would make sense here > some things, like games, should definatly not be part of the > offcial distribution. while i dearly love xtetris, you can't consider > it a necessity. Disagree again!! My dad once upgraded to a 486 to run windoze to play minesweeper.. > 2) Everything else could be moved over to contrib. [snipped a bit here, your point covered at 1) and 5)] > 3) revamp the web pages. its the first place a user might check for info > on debian and they look really bad right now. ( i know this has > already been discussed ) Some point here. But IMHO bad is a big word. At least there's no ad's or animation. > 4) open up the release dates a bit. last i heard, the push was for 3 month > cycles. open it to 6 months. while those who want to stay on the > cutting edge can do so, some people and most companies want stabilty. > this is one area i don't have a clear understanding of. > alot of people were excited about the release of "slackware96" but > the release of debian 1.2 came and went. weird. Here's a good point. However, ensuring that the "minimal" system is stable at every release, while alowing some packages to have bugs in some installations is IMO quite tolerable. New users should be be able to get running to the point that they can start reading man pages and documentation and learn to configure some things on their own. If you really want to run ip-masquerading, you'd have to have some experience and wouldn't be so easily blown away by a small bug in a package. > 5) the quest for a debian logo has produced some really good work from > users of debian. i think a similiar push should be made for > debian documentation. i know there are some debian users out there > who are good at writing and debian could use some really well written > FAQs, installation procedures, HOWTOs, etc... Now here you've got my support! IMHO this should be a area of major concern for everybody using debian.
Re: Making kernel using "make install"
Regarding compiling and installing new kernels, On 17 Jan 1997, Guy Maor wrote: > Victor Torrico <[EMAIL PROTECTED]> writes: > > > What exactly does "make install" do? > > See installkernel(8) and mkboot(8). Hey, this is not a very elaborate answer. I would like to know more about details of installing new (and older) kernels and have an overview of the process as well. IMHO this is something that is not quite exhausively covered in the documentation. And yes, I did read the lot in the source tree, which is great literature when your kernel won't boot because your XYZ scsi-tape won't bargain with the interface on your DEF souncard because it has the kind of obsolete 456PQ123 chip, so you'll have to hack the source a bit. I also read "Running Linux" and "Raven" and I think they're great for everyone who wants to get an overview of linux, especially newbies. But when compiling kernels is addressed, they only tell you to do make this, make that. There's hardly any documentation of what the makefile does, is supposed to do and can do for you. Of course, there's the kernel-HOWTO and it is very good where it makes configuring a new kernel very easy, explains a lot about what the kernel does, how it handles devices, what modules are, where to get the source, how to patch it, etc.. But when it comes down to the final part: installing the kernel, there's not much more than a reference to the lilo manual. I would really like to see some additions made about how the kernel is (or kernels are) embedded in the filesystem. IMHO installkernel(8) and mkboot(8) and are not good enough as the only reference to the install option of the kernel make. The process of installing a new kernal is much to fundamental to linux to be documented only in the huge lilo documentation or the kernel hacking guide. Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
Another general clueless upgrade question (fwd)
On Mon, 23 Dec 1996, Don Prezioso wrote: > On Mon, 23 Dec 1996, Ed Down wrote: > > > OK, I instaled Debian 1.1 a while ago and it all went fine. Now I want to > > add a couple more packages, so I ftp'd to my local mirror. I went to the > I have Debian 1.1.4 and it came with kernel 2.0.6. Now I want to upgrade to kernel 2.0.27 without upgrading entirely to Debian 1.2 yet. Neither do I want to download the gzipped 2.0.27 source, I just want to patch my old source tree. Is there a Debian way to do this? Thanks in advance, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
Re: X & Serial Ports
Uhh.., I am a relative newcomer myself, but I already learned something from this list: don't use /dev/cua*, it's sort of obsolete. I had some minor problems as well but they disappeared when I converted all cua* occurrances in configuration files to ttyS*'s. > >Upon booting up X with startx, the computer can no longer communicate with > >the /dev/cua3, which my modem is connected to. > Hmm, it's look like X is trying use /dev/cua[13] or a symbol link to one of > them. For instance, it might be that X is using /dev/mouse and the latter is > a symbol link to /dev/cua[13]. Check out your '/etc/X11/XF86Config' file to > see which devices are used So do go check XF86Config en your modem/network config files as well. Hope this helps you out. Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
Re: Problems with Sendmail and X..
On Sat, 21 Dec 1996, H. Manas wrote: > A work-around would be to: > > - install gpm. I have done that > - Make sure that In your '/etc/gpm.conf' > device=/dev/xxx (xxx is the device name of your mouse I believe psmouse) > responsiveness= > type=ps2 > append='-R' Hey, I don't have that file, I can't find any mention either in the documentation. I have Debian-1.1.4 and gpm-Linux 1.06. I do have gpm-root and /etc/gpm-root.conf. Is that what you meant? Because the man for gpm-root does mention the gpm-root.conf file. > - check out that gpm is running and you can see the mouse cursor on your Linux > console (if not you have to work on your gpm config). you may start gpm by > typing as root '/etc/init.d/gpm start' I found that doing /etc/init.d/gpm stop in a xterm just hangs until I ctrl-c or kill the gpm -k from another xterm (but then again, the gpm docs do hint at strange interference with X.) > PS: this trick is drawn from the documentations of gpm package. Many thanks > goes to it's author Alessandro Rubini What documentations did you see that I didn't? Hmm.. I don't have any source for gpm installed. Is that what I am missing? How do I get it in Debian? Thanks, Joost -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
