Re: /usr/src/linux symlink
Andy Firman [EMAIL PROTECTED] writes: Okay. I have a question about this from the kernel README: Do NOT use the /usr/src/linux area! This area has a (usually incomplete) set of kernel headers that are used by the library header files. They should match the library, and not get messed up by whatever the kernel-du-jour happens to be. This seams vague to me. It says area. Shouldn't it say something like this???: I think I've seen at least one system with a /usr/src/linux/include directory, but nothing else under /usr/src/linux. In that case, unpacking the tarball into /usr/src/linux directly could be...confusing. So if one compiled a kernel from source, where are the complete kernel-headers anyway? In the include directory of your configured kernel source tree. If you're building extra modules using kernel-package, this information will automatically be passed on (your source tree really can be anywhere). For building things that aren't available as Debian packages, you might need to go through an extra step to tell it where your kernel source is. (Alternatively, kernel-package can produce kernel-source and kernel-headers packages for you...but then you already have the source so this isn't the most useful thing. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Kerberos problem!
Beck Zoltan Gyula [EMAIL PROTECTED] writes: I'm trying to configure a kerberos server, I read the documentation and followed the instructions, but something is wrong I think. Make sure you've checked the usual things, in particular that the clocks on all of your machines agree to within five minutes. Which Kerberos are you trying to set up? I have two debian sarge linux nodes on intranet (10.0.0.0/24) with hostnames ha1.aitia and ha2.aitia. Teh kdc and the krb-admin server is the ha1.aitia. The krb5.conf looks like: [libdefaults] default_realm = INTRA.NET Do you have a more proper name? You can't rename Kerberos principals, even to change the realm, so if you have a better name that you're eventually going to use, you might want to start over and use that. I have made the ktabbs for the two host: ktadd -k /etc/ha1.keytab host/[EMAIL PROTECTED] ktadd -k /etc/ha2.keytab host/[EMAIL PROTECTED] then I moved the ha1.keytab to ha1.aitia mashine /etc/krb5.keytab and the ha2.keytab to ha2.aitia mashine /etc/krb5.keytab. It might be worthwhile to check that your keytabs agree with the Kerberos server, since generating a new keytab invalidates any old ones that exist. Run 'ktutil' on ha2 as root, then do 'rkt /etc/krb5.keytab' and 'l', and note the kvno listed for host/[EMAIL PROTECTED]; then, exit ktutil, and run 'kvno host/[EMAIL PROTECTED]' and make sure those agree. I installed the ssh-krb5, krb5-user krb5-config libpam-krb5 packages on each mashine and modified the ssh pam.d configuration to authenticate with kerberos. You shouldn't need to do any PAM configuration; # cat /etc/pam.d/ssh #%PAM-1.0 auth required pam_nologin.so auth required pam_krb5.so that will just cause you to get a TGT on login if you use ssh password auth. (I think; I doubt it will cause Kerberos tickets to be a valid substitute for password auth.) So I login to ha2.aitia and use the kinit: [EMAIL PROTECTED]:~$ kinit Password for [EMAIL PROTECTED]: [EMAIL PROTECTED]:~$ Then I try to ssh to ha1.aitia from ha2.aitia: [EMAIL PROTECTED]:~$ ssh ha1 [EMAIL PROTECTED]'s password: why prompt the password? And this is not the kerberos prompt :( What does 'ssh -v ha1' say? It should mention which authentication it's trying. Also, check to see if 'klist' shows you with a host/ha2.aitia service ticket. You also might check to see if your ssh/sshd options disable ticket-based authentication; the default at least disables ticket forwarding, since malicious hosts could use that to steal your TGT. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: aptitude marking everything packages held back?
Monique Y. Herman [EMAIL PROTECTED] writes: Just wondering if anyone else has seen this. On one of my machines, aptitude works like a champ. On the other, as of a week or so ago, every package that would normally be aqua (available for upgrade) in the g screen is instead white and listed under Packages being held back. I can hit + on a package and it will then install. Very odd. Which Debian? I noticed this happening on my unstable machine a week or so ago; I hit '+' on the Upgradeable Packages line and everything went off and installed. I never entirely figured out why, though, and it hasn't done it since. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Compiling .deb Packages
Raquel Rice [EMAIL PROTECTED] writes: I'm looking for a step-by-step HOWTO for creating a .deb package. Have you looked at http://www.debian.org/devel/? There's links to lots of official documentation there, including Debian Policy and the Developers' Reference. The debian-mentors list might be more appropriate for this sort of question and other packaging questions, too. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: fsck needed with ext3?
Joel Konkle-Parker [EMAIL PROTECTED] writes: For some reason I had thought that fsck wasn't needed if I was running ext3. I installed Woody using ext3 for my main partition, but still, every 32nd boot, the system runs fsck. Is there some setting I'm missing that tells it that I'm running ext3 and don't need fsck? You still need fsck. Broadly, there are two particular cases where fsck is important (in the abstract): (1) The system shuts down unexpectedly. Most filesystems, including ext2, need an fsck to get things in order. ext3's advantage is that it doesn't need an fsck here, it just needs to replay the journal and it's done. (2) Your hard drive is flaky, or there's otherwise hardware-derived corruption. You always need an fsck if you suspect this is going on; running one every month or two (or every 25 reboots or so) is good preventative medicine, so you can catch if something bad is happening. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Can't find ntpdate (noob alert)
Francois Lachance [EMAIL PROTECTED] writes: I want to keep my PC clock set to the right time based on a time server. I think that this should work, but I get the following: phoney:~# apt-get install ntpdate Reading Package Lists... Done Building Dependency Tree... Done E: Couldn't find package ntpdate As you've probably noticed, 'apt-get install' is only useful if you happen to know the exact name of the package you're looking for. Maybe you want to get a listing of all of the NTP-related packages in the Debian archive; I'd do this in aptitude, by typing 'l' to limit the list of packages to '~dntp' (whose description contains 'ntp'). Unstable does happen to have an ntpdate package, but it's probably the wrong thing. ntpdate will synchronize your time once, at boot time; if your system is up for a while and your clock is off, your time will gradually drift out of sync and it'll never get corrected. Consider using the ntp-simple or chrony packages for more continuous synchronization instead. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Is sarge in a pre-release phase?
Michael Kahle [EMAIL PROTECTED] writes: I have a server that is dedicated to doing backups via Amanda. I have recently purchased a Intel Pro 1000 Server adapter for this machine. What kind of an adaptor is it? A network card? More recent 2.4 kernels should include a driver for this, as e1000. To get this adapter working, I rolled my own 2.6.2 kernel, but this introduced more problems onto my Woody machine. As you've noticed, 2.6 kernels won't run particularly well on woody. Given the Linux kernel's history, I wouldn't put much faith in 2.6 kernels running particularly well at all at this point, particularly for server applications. If I were going to build my own kernel, it'd probably be a 2.4.25 kernel at this point. My question is this. Is it safe to install sarge on a production server. I do not have must run flawlessly without intervention requirements on this server. It does take care of my backups, but I have a few spare DLT's lying around that I could use to get dumps of other servers if I was in an emergency situation. I have been interested in playing with the new sarge installer and I thought this would be a good opportunity if need be. I personally would still be a little wary of it, but it's your system. :-) The big thing to be worried about is probably that sarge doesn't get security updates; it can take a couple of days (or longer) for updated packages to trickle through from unstable, where an update in stable (security) or unstable would be available almost immediately. But if you really do need a 2.6 kernel, your best bet might be to go with unstable or testing and track that to keep up with package fixes. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Linux Compatibility Issue
Abdul Latip [EMAIL PROTECTED] writes: Once in a while, users are asking me about the compatibilty problem of Linux. They reasoning that M$Office can be installed anywhere like Win98, WinME, WinNT, et. al. Whereas it is hard to install a RedHat package into Debian, and so on. ...and it's hard to install Office on Red Hat, or a Debian package on WinXP. This isn't surprising. :-) The closer analogy is that it's almost as easy to install a Debian package on Debian potato, woody, sarge, etc. AFAIK, CMIIW, this is related with the compiler version, the glibc version, the kernel version, and the kernel configuration and modules. May I know what exactly the Linux compability problem is ? I.E.: It seems they are not so backward compatible? What are you really trying to do? There are some compatibility issues, which mostly break into three parts: (1) it's hard to use newer software against older versions of the C library; (2) other libraries might have changed interfaces, so you need a newer or older version; (3) a small amount of software depends on particular kernel versions. Within Debian, (1) and (2) are dealt with by dpkg's dependency system. If you want to install software that's only available as RPMs (which IME is pretty rare), you can use 'alien' to convert to a Debian package. (My understanding is that Windows has tried to avoid changing core interfaces so that (1) isn't a problem, but (2) still somewhat is [think about the number of games that prompt you to install DirectX]. (3) is, too, but more with hardware drivers, and They don't want you to think about that.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: What's the easiest way to move some files in a directory tree?
Martin Dickopp [EMAIL PROTECTED] writes: U N T E S T E D : - find olddir -type d | while read d; do mkdir `echo \$d\ | sed s,^olddir,newdir,`; done find olddir -type f -name \*.mp3 | while read f; do mv $f `echo \$f\ | sed s,^olddir,newdir,`; done I think my shell-scripting foo disagrees with yours. I'd do something like this (also untested): cd /old/directory/root mkdir /new/directory/root ln -s /new/directory/root new find . -type d -exec mkdir new/{} \; find . -type f -name '*.ogg' -exec cp {} new/{} \; rm new If /new/directory/root is something short (like /mnt), you could just as easily skip the symlink step. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Fonts, one more time
[EMAIL PROTECTED] writes: There have been a number of postings about fonts over the last few months. But the discussion seems to me to miss the point. Since this is kind of a meta-rant, a meta-answer seems to be appropriate... If you install Windows, you don't have to know anything about fonts, really. Text just looks good in all applications. Windows has a big advantage here in that there's only one way to draw things under Windows. Under Linux (and X in general), only very basic drawing primitives are offered; there are several higher-level toolkits, but this is where things start to break down. Under Windows, prettiness has happened once, and since every program uses the Windows toolkit, every program looks nice. Under Linux, both the GNOME and KDE people have tried to implement prettiness (differently), plus there are still lots of Xaw/Xt programs out there, and things that are based on GNOME 1 vs. GNOME 2, and things that ignore the widget kits entirely...it's not hard to see why the situation kind of sucks under Linux, but it's also very hard to fix it. If you install Debian Woody, text looks like shit, if you are used to Windows quality. I suspect some of this depends on what you install. If you go the all-KDE route, things will probably look fine. (Well, hopefully.) If you go the all-GNOME 2 route, you'll probably do almost as well. If you go with twm and xterm as your primary UI, it'll always suck. :-) Keep in mind that X didn't even have the concept of scalable fonts until relatively recently in its history, and still doesn't have standardized support for pretty things like transparency and antialiased fonts. (And I'm vaguely surprised that there's not an XFree86-specific X extension for transparency, or if there is, that applications don't use it. I imagine there's a good technical reason why the standard X font drawing routines can't do antialiasing, but I'm not enough of an X guru to know that. I'm still holding out for an Emacs built on top of a modern widget kit, either KDE or GNOME 2, with antialiased font support, preferably built out of XEmacs. :-) Select Linux-friendly devices when buying new computer Insert Woody CD in new computer Switch on computer Answer questions about hardware Select default answers to all other questions This also may be where the maybe Debian isn't right for you stock answer comes in. For example, Knoppix appears to be aiming for the installs itself without asking hardware questions and gives you pretty KDE market, while still having some Debian underneath. I'm sure other people can suggest other alternatives more aimed at this. What does it take to fix this problem? Fundamentally, solving a Linux religious war (GNOME vs. KDE), and porting every application to the winner. :-( -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: /etc/init.d/ - add/remove services
[EMAIL PROTECTED] writes: What is the preferred way to add or remove a /etc/init.d/ service from certain runlevels? 'ln -s' and 'rm', respectively. The Debian Policy Manual and man pages suggest I should use update-rc.d rather than manipulate the symlinks directly. They suggest that *Debian packages* use update-rc.d, not system administrators. I have searched around and in the past, people have suggested to just manipulate the symlinks directly. Is this still the case? Yes. The Debian infrastructure treats those as user configuration and doesn't change them, provided you've left at least one symlink around. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: recompilation and optimalization
Karol Czachorowski [EMAIL PROTECTED] writes: Is there any way I could recompile existing packages in my system? apt-build has no documentation and seems not to work... You might look through some of the generic Debian developer documentation... And if I use apt-get source and build manually, in many packages, I can't add optimalization flags to gcc/g++ (CFLAGS/CXXFLAGS aren't pass to compiler). Is there any (universal) way to pass flags to compiler while building debian packages? No, there isn't. You can edit debian/rules in each source package by hand, or possibly build using something like pentium-builder. Historically, though, Debian's position has generally been that processor-specific optimizations would add a lot of space to the archive and take up developer time, at very minimal performance gain for most users. There are exceptions, and things like crypto libraries and linear algebra packages *do* have optimized versions (I have a couple of files in /usr/lib/i686/cmov, for example). Search the list archives of debian-user and debian-devel for more information. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: .deb dependancy hell
Richard Hoskins [EMAIL PROTECTED] writes: [EMAIL PROTECTED]:~$ sudo dpkg -r libgphoto2-2 dpkg: dependency problems prevent removal of libgphoto2-2: libgphoto2-port0 depends on libgphoto2-2. [EMAIL PROTECTED]:~$ sudo dpkg -r libgphoto2-port0 dpkg: dependency problems prevent removal of libgphoto2-port0: libgphoto2-2 depends on libgphoto2-port0 (= 2.1.4-1). Much as when installing mutually dependent packages using dpkg directly, give both package names on the command line: dpkg --remove libgphoto2-2 libgphoto2-port0 Kind of reminds me of RPM: dpkg-the-program is in fact very similar to rpm-the-program. Trying to manage a Debian system using only an FTP client and dpkg is shockingly like trying to maintain an RPM-based system. :-) Debian has a consistent policy for packages so they interoperate better, and higher-level tools like dselect and APT, which, combined with a unified package repository, make the system much more usable. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: lmsensors with kernel 2.6.1
hanasaki [EMAIL PROTECTED] writes: Running sarge (+ xfree 4.3 from unstable) and kernel 2.6.1 Is there any lmsensors client available? xsensors doesnt run and results in the following in syslog kernel: xsensors: numerical sysctl 7 2 1 is obsolete. You'd need to get lm-sensors 2.8.2 or newer. Unstable only has 2.8.1 because the maintainer's been, uh, um, yeah. *blush* It looks like lm-sensors 2.8.3 has been released recently; you should be able to build that from source if you don't want anything besides the basic 'sensors' command-line program. (http://secure.netroedge.com/~lm78/) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: verifying kernel version
brfg3 at yahoo [EMAIL PROTECTED] writes: 1. (*) text/plain ( ) text/html (Please post to the mailing list in plain text only, and set your mailer to wrap lines at 72 columns.) One of the authors in this thread seems to suggest that Debian 3 already comes with 2.4.20 kernel. A 2.2 kernel is still the default one, but you can boot the installer from a 2.4 kernel, and install a 2.4.18 kernel with some security fixes backported after the installation. You should be able to install any 2.4 kernel on woody without problems, and you can probably find backported kernel-image packages online fairly straightforwardly. Is it possible the the uname -r command is returning the same version for the kernel image? I'm not entirely clear what you're asking. Generally the version that 'uname -r' will return matches an installed kernel-image package, except that the woody installer doesn't install a kernel-image package at all. It just seems hard to believe that Debian would ship woody rc2 with such an old kernel. Why? The kernel is just like any other piece of software in this respect; Debian's practice is always to keep the same version of software in stable, but backport security fixes as necessary. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: kernel update
Paul Morgan [EMAIL PROTECTED] writes: On Sat, 24 Jan 2004 03:56:13 +0100, knoppix wrote: Kernels work differently than other debian packages. Each kernel revision is a *different* package. So, do: apt-get update apt-cache search kernel-image apt-get install kernel-image-whatever Or even, 'aptitude', then within that, 'l kernel-image', pick one, '+', 'g', 'g'. Also, old kernels are never removed. To see what kernels you have hanging around, ls /boot To remove an old kernel (it won't silently remove your current kernel): dpkg --purge --force-remove-essential kernel-image-whatever Whoa, you passed a --force option to dpkg. You probably never ever want to do that. 'dpkg --purge kernel-image-2.4.18' should work fine (kernel packages generally aren't tagged essential). Or you can use '-' in aptitude to remove kernel image packages, just like anything else. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: using regex substitution on BASH variables
Matt Price [EMAIL PROTECTED] writes: Now, if this information were stored in a file, it would be simple to manipulate with regex's: sed 's/@/ -at- /' addressfile But I don't see an obvious way to get sed or awwk to take variable values as input. I can do the following: FROM=`formail -xFrom: $msg` ; echo $FROM tempfrom FROM=$(sed 's/@/ -at- /' tempfrom) but this strikes me as awkward. Is there a better way? You can use sed and friends in a shell pipe. So a typical invocation would look something like FROM=`formail -xFrom: $msg | sed -e' s/@/ -at- /g'` (Run formail, taking its input from the file named in $msg; then send its output through sed, and take the result of that and assign it to $FROM.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: graphic system setup
Marius Amado Alves [EMAIL PROTECTED] writes: I cannot get X to work properly. startx runs but shows in a 320x200 mode even though I select a higher mode in the configuration. Also, it shows only in the bottom half of the screen. The input devices seem to work. The problem is just the graphics. I've been configuring with xf86config and also hand editing. Is there another way? 'dpkg-reconfigure xserver-xfree86' will get you yet another configurator. Editing the XF86Config file by hand is mostly fine, too. I don't know my monitor frequencies exactly. The tag on the back says: Model 14A 14 color monitor 100-240 V 50/60 Hz 1.0 A LEO Systems Inc. This is a monitor, it can be used on US and European power. Sometimes applying Google for the manufacturer and model number can get you the manual if you've lost yours. My card is a S3 TrioV64+, but I don't know the memory size. I've tried From 256k to 2M. I think S3 cards in general lost support between XFree86 3.3 and 4.0; you might try using the older X servers in woody, e.g. xserver-s3. (II) Setting vga for screen 0. (**) VGA(0): Depth 8, (--) framebuffer bpp 8 ...you're using the (very generic, unfeatureful) VGA driver, not anything specific to your hardware. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Unable to find Ncurses libararies/Compile Kernel
[EMAIL PROTECTED] (Forest Fisher) writes: I'm trying to upgrade from the 2.4.18 kernel that installs with the bf24 flavor to hopefully a 2.6.1 kernel but at the very least, a 2.4.24 kernel. I put the zipped source code (i.e. linux-2.6.1.tar.bz2) into the /usr/src directory, checked to make sure there were no symlinks, and then did: #tar -xjvf linux-2.6.1.tar.bz2 Then, I used #ln -s linux-2.6.1 linux (That symlink is unnecessary on Debian, and potentially dangerous if you're installing other software that sees it and incorrectly assumes it corresponds to the running kernel. I also generally perform this steps as an unprivileged user, not root, and in my home directory.) So then I tried: #apt-get install ncurses As you've probably noticed by now, 'apt-get install' is only really useful if you know the exact package name. A more useful approach might be: (1) 'apt-get install aptitude' (2) 'aptitude' (3) 'u' if you need to update the package listing (4) 'l ncurses' to limit the display to only curses packages (5) Move down to not installed packages, press '[' to expand it (6) Notice that the package description for libncurses5-dev includes header files [and] static libraries; press '+' to install it. (7) Press 'g'. aptitude will display a listing of everything it wants to do. Press 'g' again to actually install. (8) Press 'q' to get out of aptitude. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Catches in upgrading to backports.org kernel?
Joel Konkle-Parker [EMAIL PROTECTED] writes: I'm considering adding a 2.4.24 kernel from backports.org to my pure Woody system to get support for my multitude of peripherals and such. Are there any catches that I should look out for before I go for it? I noticed that this will also update my initrd package. Is this safe? If you're comfortable with recovering from the usual ways kernel installs go wrong (forgot to run LILO, for example) this should be pretty harmless. Kernels are an exception to the normal way Debian packages work in that you can have multiple versions installed in parallel and, so long as your boot loader deals, they all work fine. You may want to install the most recent kernel-image-2.4.18 out of stable, just to be safe and have a backup around. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Versions
Pedro Hernandez [EMAIL PROTECTED] writes: --- Rus Foster [EMAIL PROTECTED] escribió: On Thu, 22 Jan 2004, Pedro Debian stable aims for well, stability so the old packages are known to be secure and work. If you want newer versions have a look at things Ok. I can buy that. How does things work with versions that are known to have security issues? Postfix 1.11 as well as 1.12 are known to have security flaws, are the debian version patched for those but still keep version 1.11? In general, Debian prefers to port fixes to older versions rather than push a new release, with new unknown bugs, on to unsuspecting stable users. If you look at the package's Debian changelog (generally /usr/share/doc/packagename/changelog.Debian.gz), it should include backported security fixes. like apt-get.org and backports.org Thanks, but secure and stable is what I am looking for right now, so woody-stable might be the solution. (Within Debian there are also two other branches: unstable is the branch developers upload to directly and always has the most bleeding-edge software, sort of; after some time in unstable without a serious bug appearing, software automatically migrates from unstable to another branch, testing. So in theory testing is the working parts of unstable from a couple of weeks ago. Starting with stable is definitely a good way to go, though.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: modules.conf equivalence for non-modules
Abdul Latip [EMAIL PROTECTED] writes: I would like to know where to get information about how to make aliases like in /etc/modules.conf for non modules (i.e. compiled in the kernel). Mostly you don't; the only way to configure drivers built into the kernel is by passing the right command-line options to the kernel (generally on your 'kernel' line in your GRUB configuration, or with an 'append=' line in the LILO configuration file). What are you actually trying to do? Aliases are generally used to mean when foo service is required, load bar driver, which isn't meaningful for things built into the kernel. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Debugging rsh
[EMAIL PROTECTED] writes: I am trying to rsh to my server. I am using kerberos authentication. I believe I have configured everything correctly, but rsh simply complains: hostname: Connection refused. That message generally implies that nothing at all is listening for an rsh server, at least on the port the rsh client expects. I checked my inetd.conf. What did you find? Peeking in /etc/services, it looks like krsh is probably running on the 'kshell' port (TCP port 544), not the normal 'shell' port (TCP port 514), so you might check that you have an rshd configured to listen on the right port. Also, there's the usual checks for Kerberized services: do you have a TGT? If you do, do you get a service ticket (for krb5, host/remote.host.name; for krb4, rcmd.remote.host.name)? (I'd guess yes and no from the symptom, but it can't hurt to check.) Which rsh server are you actually using? I think even around here, the standard for remote shell access is moving to ssh over Kerberized rsh and friends. You can configure ssh to do Kerberos authentication, and tunnel arbitrary things (including X programs) over the ssh connection. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ MIT Athena Frequently Asked Question #578: Why can't I use RSA authentication to connecto the dialup servers? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Automatically setiting mtu
glenn [EMAIL PROTECTED] writes: Firstly just because I used the tla MTU doesn't mean I undertand it, Maximum transmission unit, it's the largest packet size that can be sent out over some particular interface. For Ethernet the standard MTU is 1500 bytes; you can apparently get better interactive performance over very slow connections (modem) by setting a lower MTU, and some odd types of network connections may also require a lower MTU (e.g., IP-over-IP tunnels consume 40 bytes of the packet, so the MTU on a tunnelled connection is generally 1460 bytes). In theory IP can deal just fine with fragmenting packets larger than a given connection's MTU, but in practice some broken sites (slashdot.org is a good one) set flags that cause fragmented packets to be dropped. I look after a box where I need to set it, and would like it done automatically. You can set the MTU on an interface from /etc/network/interfaces: iface eth0 inet static address 10.20.30.53 netmask 255.255.255.240 broadcast 10.20.30.63 gateway 10.20.30.49 mtu 1460 After a bit of googling the only way, out of many, that seems to work on this box is to delete the default route and use the route command to reinstate it with the mss parameter. You could probably do that too; is this a particularly complicated router machine? I've experimented with the option interface-mtu parameter in dhclient.conf In my experience that option is completely ignored; it *might* work if the DHCP server is configured to send it and the client is configured to receive it, but there's no guarantees. Anyone ever had to play with this to get through the cursed ms internet sharing reigiem? (!) My (uninformed) impression is that that was just a NAT, which shouldn't require resetting your MTU. You might run Debian on your gateway machine instead; it's much more configurable, and much easier to find out what exactly is going on. 'apt-get install ipmasq' should get you going for straightforward things. What exactly *is* your configuration here? -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Release Management and Configuration Support
Antonio Rodriguez [EMAIL PROTECTED] writes: On Wed, Jan 21, 2004 at 03:49:59PM +, Colin Watson wrote: On Wed, Jan 21, 2004 at 10:10:49AM -0500, Timothy M. Spear wrote: If I choose to install the current Sarge Release, when Sarge becomes the stable release, how do I change the apt-get functionality to query the stable instead of the unstable servers? I have looked for and not found documentation on this. Refer to the distribution as sarge rather than testing. You will need to change the wording in /etc/apt/sources.list to reflect the name change. Right, but if you change it to 'sarge', you'll get testing until Sarge releases, at which point you'll have stable. After release -- potato woody sarge sarge+1 vvv APT tag vvvvv Release codename v potato potato potato potato potato woody woody woody woody woody sarge sarge sarge sarge stable potato woody sarge sarge+1 testingwoody sarge sarge+1 sarge+2 unstable sid sid sid sid sidsid sid sid sid That is, you can use a release codename directly, and stay at that release forever (even when that release actually happens, and after the next release happens). Or you can use a label like 'stable' for the current release or 'testing' for the next release. 'unstable' and 'sid' are the same thing, and is the area developers upload packages into directly; it will never get released directly, but packages from unstable trickle into testing. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: No keyboard input under X11
Curtis Vaughan [EMAIL PROTECTED] writes: I just installed Debian (woody) on a new box. Whenever I invoke X11 (kdm) the keyboard stops working. When configuring xserver, I have tried both pc101 and pc104 (I think it starts pc). But the same results. I have also tried xserver-xfree86 and xserver-svga. The keyboard is plugged into a serial port, btw. Do you mean that, or do you mean you have a USB keyboard? If your keyboard really is plugged into a real serial port (my only console is a dumb terminal) I don't think X will love you at all. Otherwise, Google (usb keyboard xfree86) suggests that USB keyboards should, in general, Just Work under X. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: problem with network configurations
[EMAIL PROTECTED] (enantiomer) writes: David Z Maze [EMAIL PROTECTED] wrote in message news:[EMAIL PROTECTED]... My laptop's /etc/network/interfaces file looks very loosely like iface eth0 inet dhcp iface net-home inet static address 192.168.1.3 netmask 255.255.255.0 broadcast 192.168.1.255 gateway 192.168.1.1 and if i declare the static part first, then it would load the static ip first and I would have to manually call ifup for the dhcp? would i have to name the static like: iface eth0 inet static and the dhcp like iface at-school inet dhcp? Is it the order that matters in the interfaces file? I am a little confused on the naming convention here. Do I even have to use the eth0 phrase? eth0 is the actual name of the hardware interface. (If you run 'ifconfig', you should see a listing for eth0.) If you declare 'auto eth0', at boot time 'ifup eth0' will happen, which will do whatever is declared in your 'iface eth0' stanza. The ordering is irrelevant. Names that don't correspond to interfaces can be anything, but I intentionally picked net-foo to (a) be consistent and (b) try to avoid stomping on potential network interface names in the future. I also have several net-foo stanzas for different places I might need a static IP, so I can 'ifup eth0=net-home' or 'ifup eth0=net-mit' or whatever. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Truetype font size
Antony Gelberg [EMAIL PROTECTED] writes: Anyone know how to change the font size in the following line: rxvt -fn -*-trebuchet ms-medium-r-*-*-*-*-100-100-*-*-*-* That's a font spec requesting a non-bold non-italic font named trebuchet ms at 100 dpi, at any size. Adding a point size might help; for 11 pt, -*-trebuchet ms-medium-r-*-*-*-110-100-100-*-*-*-* I like to fill in as many of the fields as is practical, though. I generally prefer scalable fonts, which X is somewhat reluctant to provide if there are any *s in the size fields; 0 has a similar meaning but prefers scalable fonts. Explicitly specifying the font encoding also avoids some class of surprises. So I'd actually use -*-trebuchet ms-medium-r-normal--0-110-100-100-*-0-iso8859-1 (Well, I'd actually use -*-helvetica-... for a sans serif font, but anyways. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: dhcp client setup
Attila Csosz [EMAIL PROTECTED] writes: How to setup a dhcp client? My computer connected to an another(server) the server act as a dhcp server. If I put the following lines to the /etc/network/interfaces auto eth0 iface eth0 inet dhcp Is it enough or should I start a client program too on my computer(client)? Which package is recommended to use: dhcp-client or dhcp3-client You need to have a DHCP client installed, but ifup should take responsibility for running it for you. Either of the packages you suggest should be fine; I generally use dhcp3-* because it's newer and shinier, but that's not a terribly good reason. :-) Other than that, it looks like you have done the necessary configuration and things should Just Work. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: problem with network configurations
[EMAIL PROTECTED] (enantiomer) writes: Recently I looked into setting up my laptop so that it could switch between static and dhcp. I looked at an article on this group and tried the suggestion and it didn't work. I restored my interfaces file back to normal (i had commented out what I had put in before) and thought all would be well, but now it doesn't seem to like even my old static ip network anymore. Well, first off: # /etc/network/interfaces -- configuration file for ifup(8), ifdown(8) iface eth0 inet dynamic The last word of that line wants to be static or dhcp, probably in your case static; see interfaces(5) or the example in /usr/share/doc/ifupdown. You should get an error message if you run 'ifup eth0' by hand. Is there any sort of reconfig tool for the network in debian? Anybody see what might be wrong? Also, what is the best way to be able to switch between static and dhcp for me? My laptop's /etc/network/interfaces file looks very loosely like iface eth0 inet dhcp iface net-home inet static address 192.168.1.3 netmask 255.255.255.0 broadcast 192.168.1.255 gateway 192.168.1.1 So at boot time, the network will try to come up (via ifplugd, but that's a detail) and get DHCP. If I know I want a static address instead, I can do 'ifdown eth0; ifup eth0=net-home' to use a different set of network parameters. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Multiple ip addresses with hardware router
Tony Baechler [EMAIL PROTECTED] writes: Hello all. This is my first post to the list, so sorry if this is off topic. I have a Dlink di604 hardware router. I am running Debian Woody. Is there a way I can get Linux to listen to multiple ip addresses (not the internal 192.168 block, but in the 68.x.x.x block)? Sure, just add alias stanzas to your /etc/network/interfaces file. For example: auto eth0 eth0:1 iface eth0 inet static address 18.101.2.50 netmask 255.255.255.240 broadcast 18.101.2.63 gateway 18.101.2.49 iface eth0:1 inet static address 192.168.1.2 netmask 255.255.255.0 broadcast 192.168.1.255 Does the hardware router block all but one ip address or is there a way around this? You'd have to look in its manual for details. Certainly, my router box (a somewhat sad P100 running woody) has three IP addresses across two interfaces, and works fine for almost everything. There's nothing intrinsic in the concept of router that prevents you from having multiple external addresses, forwarding some internal addresses unchanged but NATting others, etc. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: OT: How to match a substring?
Kent West [EMAIL PROTECTED] writes: if {the first word of uname -a is Linux} then echo You're logging into Linux else echo You're logging into something else, probably Solaris fi For yet another approach: case `uname -a` in Linux*) echo You're logging into Linux ;; *) echo You're logging into something else, probably Solaris ;; esac Which has the minor advantage of only using shell primitives, aside from the call out to uname itself. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: How to disk cleanup after apt-get upgrade/install/remove cycles?
(Please set your mailer to wrap lines at 72 columns.) Ramasubramanian Ramesh [EMAIL PROTECTED] writes: Often one is not sure of the choices and strength of packages/utilities and tend to install multiple utilies for the same purpose. For example I have about 3 or 4 cd players and assortment of mp3 palyers etc. Sooner or later you see that a lot of left over packages exists in your system that you no longer use. I am wondering if there is a fool-proof way of cleaning up that eliminates all unused packages. Specifically, if I installed package A that brought in packages B C and D due dependencies. Is there a way to find out that B C and D are no longer in the dependency list of any packages after removal of say A. This will allow me to systematically remove all unwanted packages. I'd suggest using aptitude as a package manager. If you mark a package for installation inside aptitude (or install a package from the command line using 'aptitude install'), aptitude will mark packages that are only installed because they're dependencies as auto-installed, and if you remove the package, it will later remove those. You can also add or remove the auto-installed mark by hand. So I'd: (1) Install aptitude. (2) Start aptitude. (3) Find packages that you think you don't need. Press 'M' (capital m) on each of them in the listing. They might turn purple and also gain a 'd'; that just means aptitude wants to uninstall it. You can also press '-' to remove (or '_' to purge) a package if you know you don't want it, '+' to install a package, or 'm' to clear the auto-installed mark. (4) Press 'g'. aptitude will display a summary of what it wants to do. (5) Press 'g' again; aptitude will actually do it. Pressing 'q' will always back you up to the previous screen; from the main screen, it'll exit aptitude. You probably want to mark every package in the 'devel', 'interpreters', 'libs', 'libdevel', 'oldlibs', 'perl', and 'python' sections as auto-installed, unless you know otherwise (you're probably a developer of some sort). (Some of those sections only exist in testing/unstable, not stable.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: installing kernel?
0debian user [EMAIL PROTECTED] writes: Hi I am running Debian unstable with kernel image 2.2.18 (! But I guess if you installed woody from a non-bf24 kernel and then updated, you could legitimately have something this ancient.) 1) What kernel should I install (a 2.4.24 stable kernel or a more risky 2.6.0)? I would suggest staying with the 2.4.x series until 2.6.x has proved itself a little more; YMMV. 2) How does one install kernel image package in Debian? Do I have to move ny old modules directory away so they are not overwritten if the new kernel fails and I must boot the old one? No, that shouldn't be necessary... 3) How does one install kernel from source in Debian? After it is compiled I should move the kernel image to /vmlinuz and copy over the System.map file to / ? I should run lilo -v before reboot? You should install the kernel-package package, and use that to build your kernel source and install it. There's a document linked to from http://newbiedoc.sourceforge.net/ which explains how to do this; you can also read kernel-package's documentation. But essentially, you run make-kpkg on a configured kernel source tree, it chews on things for a while, and eventually spits out a kernel-image .deb package. You install that with 'dpkg --install', which deals with making sure /vmlinuz points somewhere sane. Run 'lilo' if you need to and reboot. 4) What should I add to /etc/lilo.conf so it will let me select old or new kernel? Should work out-of-the-box, with options to boot /vmlinuz and /vmlinuz.old. 5) What is initrd and is it good to use? It's a system where the kernel boots from a ramdisk, loads some modules, and then goes on with life. It's useful if you don't know what needs to be compiled into the kernel, which is particularly important if you're building an official distribution kernel that everyone uses. It's probably more of a pain than it's worth if you're compiling a kernel for one specific machine. 6) How do I know what in the kernel config I should let the kernel load as modules and what should be compiled into the kernel image? If you're not using initrd, you must compile in drivers for your root disk and root filesystem. I'd suggest building modules for any removable device (so if you get a new USB mumble, you don't need to rebuild to have a driver for it), and not building modules for non-removable devices you don't have (e.g., ISA Ethernet cards). But building extra modules doesn't hurt, except in compile time and disk space. 7) Is there a good kernel install/config guide that is tailored to Debian and addresses 2.4.24 or 2.6.0 kernel? See earlier-referenced http://newbiedoc.sourceforge.net/ article. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: routing problem
Cosmin [EMAIL PROTECTED] writes: 1. (*) text/plain ( ) text/html (Please don't post to the list in HTML; plain text is fine.) (Summary: external router machine has external address 82.77.83.33/27, with routable internal network 81.196.166.97/29 and internal NAT network 192.168.0.0/24.) I have configured the file /etc/init.d/firewall like this: iptables -t nat -I POSTROUTING -s 81.196.166.96/29 -o eth0 -j MASQUERADE iptables -t nat -I POSTROUTING -s 192.168.1.0/24 -o eth0 -j MASQUERADE Both of the networks here use as a gatway the IP 81.196.166.97 given to eth1. There's two problems here: (1) You're telling the firewall to NAT both networks, where presumably you want the 81.196.166.97/29 network to be directly routed. You don't need special firewall rules for this, though you do need to have IP forwarding enabled. (2) You've told the machines on the 192.168.1.0/24 network that their gateway machine is on a different network, so they don't know how to reach it. You probably need to give the gateway machine an address on the NAT network (like 192.168.1.1) and tell the NAT machines to use that as their gateway. Do you have a solution to this problem?? ( I mention that all my computers are using WIN98 ) ...so install Debian on them. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: free cached memory
gmorais [EMAIL PROTECTED] writes: when I use free in command line I get: total used free shared buffers cached Mem:449300 219144 230156 0 5560 149084 -/+ buffers/cache: 64500 384800 Swap:0 0 0 my question is: there is any problem || related to the fact that my cached memory is so much occupied? No; if anything, you'll see a performance improvement, since in some cases Linux can use that memory instead of going to disk. Having actual free memory would be wasting that resource. and if there is, how can I free it? You don't, but if programs actually legitimately need the data space, the kernel will shrink the cache and buffer space. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: network configuration problem
Ritesh Raj Sarraf [EMAIL PROTECTED] writes: 1. (*) text/plain ( ) text/html (Please post to the list in plain text only.) i'm a newbie to Debian. I just shifted from RedHat. I've got two lan cards on my debian system. one connected to the internet and the other to my local lan. i'm not able to ping my ISP DNS server from my debian machine. Details: eth0 (Ethernet connected to ISP) IP 192.168.1.43 DNS 192.168.1.1 Gateway 192.168.1.1 Subnet 255.255.255.0 So, to be clear: you're logged in on this machine, you type 'ping 192.168.1.1', and nothing happens? Is the physical infrastructure working? (Are all the cables connected, do you have the right blinky lights everywhere?) If that all works, what do 'ifconfig eth0' and 'route -n' say? eth1 (Ethernet connected to my LAN) IP 10.0.0.1 DNS 192.168.1.1 Gateway 192.168.1.43 Subnet 255.0.0.0 Gateway isn't something you want to define on more than one interface; if you do define it, if needs to be an address on the same network (so, in this case, a 10.x.x.x address). How exactly are you specifying gateway and DNS here? It looks like the setup you want is: -- eth0 is on 192.168.1.43/24 -- eth1 is on 10.0.0.1/8 -- The default route is via 192.168.1.1 (on eth0) -- The DNS server is 192.168.1.43, which happens to be on eth0's network -- (Optional) ipmasq from eth1 to eth0 And none of the settings you've given so far contradict this. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Debian install on Virtual PC / OS X
Dave's List Addy [EMAIL PROTECTED] writes: I am trying to install the latest Debian with the ISO bf2.4-3.023-netinst. I am doing this with Virtual PC 6.1 on an Apple iMac 17 Flat Panel with a G4 1.25 MHz. I can get the install to boot and start, my problem is on selecting the correct Network driver for the iMac. I tried Tulip and no go. If I understand what Virtual PC does properly, the actual hardware inside the Mac is somewhat irrelevant; you'd need to figure out what network hardware Virtual PC is emulating and use a driver for that. You'll almost certainly get noticably better performance, and not have an intermediate layer of Microsoft, if you installed Debian (PPC) on the Macintosh directly. The downside is that you won't be able to use Debian and Macintosh programs at the same time, since you can only be booted into one OS or the other. Maybe there's something like VMWare that's a PPC virtualizer out there? -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: www.gnu.org, glibcbug, bug@gnu.org, dpkg-reconfigure locales
s. keeling [EMAIL PROTECTED] writes: What a strange loop I've landed in. I started out trying to make en_CA locale work. dpkg-reconfigure locales everybody said, but that didn't do it. Once I edited /etc/locale.gen and /etc/locale.alias, then ran locale-gen, it was finally fixed. From what little I know about this, that sounds like the way it's actually supposed to work. In /etc/locale.alias it says to use the glibcbug script to report it to gnu. That sends to [EMAIL PROTECTED], which bounces. ... What do I do now? [Note: cc:'ed to Woody's libc6 maintainer.] If it's a problem in the Debian packaging ('dpkg-reconfigure' doesn't do what I expect generally is) then you probably don't want to report it to GNU in any case. It's easy to report a bug to the Debian bug-tracking system, though, probably using the 'reportbug' command in the 'reportbug' package. ('reportbug libc6' or 'reportbug locales') If it really is a problem in the upstream source, rather than the packaging, the Debian maintainer should take reponsibility for forwarding it appropriately. See also http://bugs.debian.org/ and, for example, http://bugs.debian.org/libc6. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: switching user in X without logging out previous user
Micha Feigin [EMAIL PROTECTED] writes: Is it possible to switch the active user in X without actually logging out using the graphical interface( (x|g|k)dm ) like with the change users option of m$ XP ? Not as such. A common thing to do is to switch to an X console (Ctrl+Alt+F1), log in as another user, and have them run 'startx -- :1'; then, with default settings, Ctrl+Alt+F7 will get you the first X session (on :0), and Ctrl+Alt+F8 the second (on :1). I think most display managers can be configured to manage multiple X servers as well. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: having trouble with abcde
Scott Berry [EMAIL PROTECTED] writes: I am having a problem converting tracks from my cdrom which is /dev/cdrom. This is the error message I get when I try to do this: abcde Getting CD track info... cd-discid: /dev/cdrom: CDROMREADTOCHDR: Input/output error abcde error: CD could not be read. Perhaps there's no CD in the drive? Are you using cdparanoia underneath? (Probably.) A useful exercise is to run 'cdparanoia -v 1' to try to extract the first track from the current CD. If you're using IDE-SCSI emulation, you'll probably need to install the 'sg' module too. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: eps to page of labels
David Purton [EMAIL PROTECTED] writes: Does anyone know of a simple way to print a page of labels from a single eps file? I could always import the eps into scribus lots of times, but I'm hoping there is a simpler way like eps2labels... or some such thing. Any ideas? Not simple, but you could write a Postscript file that did the work. If I was writing it, and I knew the label was 2in wide (144pt) and 0.5in tall (36pt) I might do something like: %!PS-Adobe-3.0 %%Pages: 1 %%EndProlog %%Page: 1 1 36 144 324 { % x: first at 36pt, last at 324pt, incr 144pt 36 36 360 { % y: first at 36pt, last at 360pt, incr 36pt save translate /showpage {} def 0 setgray 0 setlinecap 1 setlinewidth 0 setlinejoin 10 setmiterlimit [] 0 setdash newpath userdict begin dnl THIS BIT IS PROCESSED BY m4 %%BeginDocument label.eps changequote(,) include(label.eps) changequote %%EndDocument dnl END m4 DIRECTIVES end restore } for } for showpage %%EOF This needs to be preprocessed with m4, but should come close to doing what you want. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: OT: bash scripting question -- passing values to ls
Matt Price [EMAIL PROTECTED] writes: here's something that ocmes up a lot for me: I use locate to find a bunch of files: % locate charter | grep -i font /usr/share/texmf/fonts/afm/bitstrea/charter /usr/share/texmf/fonts/tfm/bitstrea/charter /usr/share/texmf/fonts/type1/bitstrea/charter /usr/share/texmf/fonts/vf/bitstrea/charter then I want to ls -l each of these files... IMHO, the simplest way to do it is locate charter | grep -i font | xargs ls -l but that will fail if the filenames have spaces or other whitespace in them. The very slow but reliable equivalent would be find / -name charter -path '*charter*' -print0 | xargs -0 ls -l or even find / -name charter -path '*charter*' -ls but that doesn't do the caching that locate does. (Other people have given good suggestions about using backticks to do this as well.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Package manager tar.gz apps
Alphonse Ogulla [EMAIL PROTECTED] writes: Just wondering, what happens to the system when you remove an installed package or component of a package and replace/update with a tar.gz compiled From source application? Can this break the system -- leading to dependancy issues? Don't do that. :-) If you're going to install things from source, do it in a place not controlled by the package manager, such as /usr/local or $HOME. In some cases you can leave the Debian version installed but just not use it; in other cases (MTAs come to mind) you can use the equivs package to create an empty package that causes dpkg to believe that a package is installed. You definitely should NOT './configure --prefix=/usr; make install' a random source package, especially one that exists as a Debian package. If dpkg or APT decides the package should be upgraded, it'll blindly overwrite your version, which it has no idea exists. I'm asking because I had to install module-init-tools, e2fsprogs and procps (requisite for kernel 2.6.0) from source simply because I couldn't find the respective deb package or probably the deb package failed on my system. All of those are in unstable, and rumor is that you can run 2.6.0 kernels under unstable without too much trouble. If you're brave enough to run a Linux kernel that young, you're probably brave enough to run unstable too. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Package installing / Kernel Upgrade
Jan Albrecht [EMAIL PROTECTED] writes: after my debian is working without problems on my firewall for a few months, I have two more or less simple questions: #1 While updating or installing a package the computer needs a lot of time. For example updating 2 packages took about 45 minutes. Okay, the computer is only a P133 with 16 MB RAM, but is this a normal behaviour? RPM never took so long to install. Is there any way to speed up this process or just the normal way? My firewall machine seems to do okay with APT and installing packages; it's a P100 with 56 MB of RAM (recognized, there should be 64 MB but I never bothered to find out where the last 8 MB went). It's probably the lack of memory that's hurting you. Can you find some cheap/free used 72-pin SIMMs for the machine? #2 Is it possible to install a new kernel with apt-get upgrade? Or is the only way to upgrade the kernel to recompile a new one with the dpkg options? No, and no. Generally new versions of the Linux kernel come in differently-named packages. So 'apt-get upgrade' will see that there's not a newer kernel-image-2.4.18-386 and not upgrade anything, even if there is kernel-image-2.4.23-386 available. I'd normally recommend using 'aptitude' to look through the package list ('/ kernel-image', then '\' until you find one that looks good, then '+' and 'g' to install it). On a machine that limited, you might want to search for packages on a separate machine, or use the lower-level command-line APT tools. You can also compile a kernel on a separate machine, and it will generally work fine. I also do this with my firewall machine, since I'd prefer the kernel compile to take less than a week. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: High DPI values and how text and grfx loose scale.
Mike Mestnik [EMAIL PROTECTED] writes: On my system I have small monitors that I drive at high resolutions. This makes my DPI about 130. With this DPI it's vary obvious that text is bigger, while EVERYTHING else remains small. This has created a lot of bugs dealing with fonts being bigger than there containers. I also run with dual heads, so I have 2 valid DPI settings. Gnome cares not of these things, I think it must. Sucks, don't it. :-( I think the current best buzzword for this is SVG (scalable vector graphics). But a lot of the current world seems to be based around bitmapped graphics, which are designed assuming a 75dpi display. Web pages, which were initially intended to be completely platform-independent, have gathered a lot of unfortunate dependencies on a particular pixel size; I bet that, say, MIT's top-level Web page (http://web.mit.edu/) would look terrible on your display, especially if you're using a sensible Linux browser. You might be able to find a window manager that believes in higher display resolutions. (I think modern openbox does, though it's stopped believing in non-XINERAMA multi-head displays.) A Web browser that scaled images and pixel sizes up to your advertised display resolution from 72dpi would be clever. But otherwise, yeah, a lot of the world breaks this way. (...and I've thought for a while that having a 300dpi LCD display would be very nice, aside from software support sucking in pretty much exactly this way.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: X nie dziaa
ukasz Roej [EMAIL PROTECTED] writes: Can anybody tel what should I do to get my X Window working? Here's what it says when I try to run it: (...) Could not init font path element unix/:7100, removing from list Fatal server error could not open default font 'fixed' (...) What X packages do you have installed? In particular, do you have the various xfonts-* packages that xserver-common suggests (xfonts-100dpi or -75dpi and xfonts-base) installed? -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Public / private IP addresses
Antony Gelberg [EMAIL PROTECTED] writes: I've dealt with quite a few LANs over the years. I'd like to try something that I never have done before... I work with ADSL providers who allocate 5 public IP addresses (sometimes 1) to a connection. If I have a LAN of, say, 20 workstations, I can use NAT, and give them private addresses - no problem. I usually have an ADSL router / modem, hooked up to a Linux box configured as a bridging firewall, which connects to a switch. But if they wanted to run a public email server as well, clearly that needs a public IP address. Fine, but how does the routing aspect work? Do I need to ditch the bridging configuration on the firewall and reconfigure it as a router with 3 NICs? You can run two IP networks on the same physical network; I do that here for arcane and esoteric reasons. :-) If your ISP gives you static IPs, this is easy; you arrange for the mail server to have an externally visible IP address (either by configuring it directly or having your local DHCP server hand it an external address), and tell your firewall machine that that address is internal and that it should route it directly without NATting. I don't know if this is a solvable problem if you need to get the second address by DHCP, though; I could envision some cleverness wherein the gateway machine acts as a transparent bridge if it sees traffic from the specific MAC address of the mail server, but I'd have no idea how to set it up under Linux. In that case, having three NICs probably would help, since you could bridge from DMZ-external and NAT from internal-external. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Booting Caper.
Jonathan Lassoff [EMAIL PROTECTED] writes: I want to boot the debian woody install on the second drive, and have been with a boot floppy for a few months now. [...] Well the original disk reprted all kinds of bad sectors while writing it, so I found a floppy that works, and it still fails to boot. How are you making the boot floppy? I'd probably try to do this sort of thing by using a real bootloader... So then I thought I might have my first go at using GRUB on the command line. So I boot into my Redhat 9 install and switch to single user mode (init 1) and run grub. I set the root partition and specify my kernel with all the right options. Then I specify my initrd image and then run boot and the thing just just quits, it doesn't boot or do anything. Well, yeah, you've already booted the machine, the command-line grub isn't going to magically reboot your running kernel. You need to install grub on to some media (your hard disk or your known-good floppy) and boot from that, then this incantation would work. Read the GRUB manual. (I find a GRUB floppy to be a great rescue tool, BTW: if you have some clue of what's on the machine, you can use it to boot even if your MBR is broken, you can boot from partitions that the local boot loader doesn't know about, and if your system is really hosed, you can connect a null-modem cable to another machine, tell GRUB to use a serial console, and start catting files from the GRUB prompt. Not that I've had flaky hardware that requires this or anything. :-) ...so my recommendation would be to follow the procedure in the GRUB manual, and make a GRUB floppy, and either use that to boot your Debian partition or use it to install GRUB into your MBR. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Woody: TaskSel - XWindows; startx fails
Paul Morgan [EMAIL PROTECTED] writes: If one is setting up a server, one may only want to run an X server (XFree86, for example) on that machine, and not an X client. Other machines on the network would be running X clients and connecting with the server's X server. One doesn't need stuff like window managers running on the server. That sounds backwards and confusing. An X server talks to the keyboard and display on the machine it's running on; an X client runs on some machine and has its display on some X server not necessarily on the same machine. xterm is an X client, for example. So a server (sitting in a rack) probably wouldn't have an X server, but it might have client programs installed that people could log in and run remotely. A window manager happens to be a special case of an X client, and it's possible to run it remotely, but it's rarely what you actually want. :-) Desktop environments like GNOME and KDE in my experience tend to be happiest if they're running on the same machine that the X server is on. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: modem external
Paul Morgan [EMAIL PROTECTED] writes: On Sat, 27 Dec 2003 00:44:50 -0200, Bruno Vane wrote: someone can help me to setup an us robotics courier v.everything external modem on debian 3.1? my linux is detecting the modem, and the modem dials to the provider, but cant establish the connection. What's the exact model number of the modem? Is it a Winmodem? Traditional wisdom is that it's hard to go wrong with an external modem, since they plug into a normal serial port which Linux always knows how to talk to. There's no such thing as an external winmodem. :-) This might be less true in the modern age of USB, though. Is the connector a 9- or 25-pin serial connector, or is it USB or something else? -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Deb commands
Stephen Liu [EMAIL PROTECTED] writes: Where can I find a list of Deb command corresponding to RPM. Is there similarity/equivalent The 'dpkg' command is most similar to the 'rpm' command; if you have an individual .deb file, 'dpkg -i foo.deb' will install it, replacing any previously existing version, and will complain if its dependencies aren't available. To actually download and install packages, most people use some variant of the APT tool set. I recommend using 'aptitude', a console-mode UI that will list the packages and allow you to mark packages for installation and removal, resolve any dependency conflicts, automatically remove unused packages, and finally do a bulk download/install/remove. The 'apt-get' command-line tool is also used fairly widely, though it's harder to resolve dependency conflicts and that tool doesn't track automatically-installed packages. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: how to access certain part of memory
j smith [EMAIL PROTECTED] writes: i want to write a program to read array of Java applet running in browser. I'm unclear that there's a good way to do that; if you actually needed something that complicated, I'd probably write a stand-alone Java application (or a stand-alone program in your language of choice with fewer dependency/freeness issues than Java). in other words, how are memory test utility implemented? That's a completely different question; you'd want to look at how tools like memtest86 are implemented, but the basic answer is in assembly, outside of the control of any operating system. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Synching volumes on logout
David Baron [EMAIL PROTECTED] writes: The problem that I had, several times, was a shutdown hangup on something like deconfiguring inetd. The script that's getting run at that point is /etc/init.d/inetd; if this is happening repeatedly, you might look at that script further. If I journal (or mount the linux partition with a synch option ??), how much of a performance loss is this? I'd believe that running ext2 with the sync option is a major performance hit, though I don't know for sure. I haven't noticed ext3 being particularly slow. Is ext3 a better system? XFS? I created the partition using PartitionMagic and it recommended the ext2. ...and now you're asking a religious question. :-) Changing an ext2 partition to ext3 is easy (using tune2fs(8)), changing to anything else involves reformatting the partition. I use ext3 on my personal machines and have been happy with it; the other popular option seems to be reiserfs. Would a real quick unmount/mount in the bash_logout be safe and do the job? It wouldn't be possible. (Unmounting your root partition involves being root, closing any temporary files bash might happen to have open, and having the shell, its libraries, and the mount program all vanish.) Running sync(1) might work, though. You'd want to make sure you ran it in the right place, though; if you're running under an X display manager, your .bash_logout file probably won't get run when you log out of that. There are also changes to the filesystem that happen between logout and shutdown (like removing the pid files in /var/run, and updating the logs in /var/log) that could still result in a corrupt ext2 partition if the system were powered down prematurely. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: trouble installing kernel image
Sam Rosenfeld [EMAIL PROTECTED] writes: I am having trouble installing a kernel image which seems to depend on initrd. The error message below is a verbatim message which follows the halting of the installation. For some reason after I add 'initrd=/initrd' I cannot run lilo. Do I need initrd? You need an initrd line in your lilo.conf (or grub.conf) if you're using a kernel built with initrd support. This includes all of the prebuilt Debian kernel-image packages, yes. If so, how can I satisfy the request from the kernel image installation request? I wouldn't worry about it at that particular point, just make sure that you've updated your lilo.conf and rerun 'lilo' before rebooting. If I don't need initrd, how do I identify a kernel image that doesn't require initrd? Build your own. (Which isn't necessarily a hard or bad thing to do, but it is effort and does take time. :-) The message: You are attempting to install an initrd kernel image (version 2.4.16-k6) This will not work unless you have configured your boot loader to use initrd. As a reminder, in order to configure lilo, you need to add an 'initrd=/initrd' in your /etc/lilo.conf That doesn't quite look right. I'd look in / to see what the file is actually named; /initrd.img sounds more right to me. I repeat, You need to configure your boot loader. If you have already done so, and you wish to get rid of this message, please put `do_initrd = Yes' in /etc/kernel-img.conf. Do you want to stop now? [Y/n] (It's safe to answer n to this prompt, let the kernel installation continue, update your LILO configuration, run 'lilo' by hand, and then reboot.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: upgrade from Corel Linux
David G. Schlecht [EMAIL PROTECTED] writes: I've tried recompiling a new GCC but can't because the current one is too far out of date. That surprises me a little; I thought a goal of gcc was to be compilable with random crufty vendor KR C compiler. But this isn't actually something I'm knowledgable about. :-) I've tried loading the GCC binaries as an rpm but that failed miserably with missing /bin/sh errors and more. (Using alien, or rpm directly?) Of course all the new versions of the tools (like apt-get) need the new libs and I can't compile them. apt-get doesn't work because it's too old and doesn't recognize the Cache-Limit config and the new packages are too big to fit. Can you try something like this: # make sure sources.list points at Debian 'stable' or 'woody' apt-get clean apt-get update apt-get install apt # now have new APT, add Cache-Limit apt-get install aptitude aptitude # and upgrade as normal Alternatively, you can download the .deb files directly and install them with 'dpkg --install', but you do wind up chasing down dependencies by hand that way. Sounds to me like it's time to scrap the old beast and reinstall debian. Or -- is there an easier path that I haven't tried? If you're stuck with a bad partitioning scheme and really can't get anything else to work, you might need to reinstall, yeah. :-( -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: How that dependency turned up?
Shaul Karl [EMAIL PROTECTED] writes: I am trying to upgrade apt-show-versions. I have installed 0.04 and am going to have 0.05. What bothers me is that the dependencies for apt-show-versions are perl | perl-5.005 | perl-5.004, apt. ...for which version of apt-show-versions? {24} dmaze% apt-cache show apt-show-versions Package: apt-show-versions Version: 0.05 Depends: perl | perl-5.005 | perl-5.004, apt, libapt-pkg-perl (on current unstable) Are you maybe looking at the dependency list for the older version? -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: compile driver file problem, different version of kernel image and kernel source, any clue?
kean [EMAIL PROTECTED] writes: trying run debian in new dell poweredge server rather redhat. everything workout fine until I get stuck in the networkcard problem. Couldn't find the e1000 network card binary driver for the kernel 2.4.20 image. Which version of Debian? Which kernel, exactly? Then I'm trying to compile one for it. The result ... compile fine but it's give me error while probe the module and after look into, this is because the kernel image is 2.4.20 and the kernel source I have in /usr/src/ and /usr/include is 2.4.18 You basically want to never ever use the kernel headers in /usr/include/linux and /usr/include/asm; if you are, your build process is doing something wrong. See /usr/share/doc/libc6/README.Debian.gz for more information. What driver, exactly, are you trying to build, and how? Does it have some way to tell it where the kernel headers are? (An e1000 driver has been included in the kernel since at least kernel 2.4.21, and quite possibly earlier; http://lxr.linux.no/ suggests that it was added to the kernel in 2.4.20. If you're using a prebuilt Debian kernel, I'd be a little surprised if you didn't already have the driver hanging around.) For the effort you're going through, it almost sounds like you might be better off building a 2.4.23 kernel using kernel-package that you know contains the driver you want. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Compiling kernel on a different computer
Piers Kittel [EMAIL PROTECTED] writes: I've got an ancient 486 which needs its kernel recompiled, but it is extremely slow, and the hard drive isn't big enough - and I think it's possible to recompile the kernel on my main PC and transfer the kernel and modules over - how to do this? Build the kernel normally using kernel-package, configuring it for the target machine. (http://newbiedoc.sourceforge.net/ has what seems to be the canonical Debian kernel compilation how-to these days.) When you're done, you should have a kernel-image-*.deb file; copy it to the target machine using scp or some such. Then log in as root on the target machine and use 'dpkg --install' to install the kernel. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: antialised fonts in emacs
Micha Feigin [EMAIL PROTECTED] writes: On Mon, Dec 22, 2003 at 10:33:04PM +1300, Paul William wrote: How can I use antialised fonts in emacs? I am using unstable. My understanding of the situation is that doing this would entail modifying Emacs's rendering engine to use something that did antialiasing. If I was going to bet, I'd bet that XEmacs successfully did this before FSF Emacs. But both Emacsen have historically been against prebuilt widget kits; the easiest way to get antialiased text from a programmer's point of view is to build your program as either a GNOME (2) or KDE program, and Emacs is explicitly neither. I am not sure about anti aliased, but try moving the type1 fonts to the bottom of the fontpath list in XF86Config and if you have truetype fonts move them to the top. Significantly improved my situation. This isn't going to help the problem at all, but I'm kind of curious why everyone keeps suggesting it. Is the X Type 1 font renderer really that terrible? In my setup I've actually gone out of my way to use Type 1 fonts and been happy with the results. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Changing bootloader and FS question
Stephen Liu [EMAIL PROTECTED] writes: My Debian 3.0 box, newly installed, is running LILO and Ext3. Can I use 'dselect' to change the bootload to GRUB and FS to Reiser without making another clean installation. OR is there any other way?. Not as such. You can use dselect (or aptitude) to install the 'grub' package, which will give you the GRUB files but not actually install them. Going through the GRUB manual will help you get this set up. In general, you can't change filesystem types in place. (ext2 to ext3 and back is an exception.) You could nuke an existing partition and create a new, empty reiserfs partition there, but that's probably not what you're after. It looks like modern GRUB does at least have reiserfs support, which is handy for you. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Need Help in compiling
sajid hameed [EMAIL PROTECTED] writes: 1. (*) text/plain ( ) text/html (Please post to the mailing list in plain text only.) i have started using unix recently i have an problem using fortran i have write a program in fortran namely test.f that uses a subroutine displaymessage.f and this subroutine uses another subroutine display.f i compiled all of them in single library say libt.a when a make an executable file of it the error is No Main Fortran Program to execute What commands, exactly, are you using to do this? This is a little unusual way to compile programs. In C, at least, I happen to know that putting main() in a library would work. Playing around with g77, though, it looks like this fails. That is, if main.f is C main.f: pull some FORTRAN code... PROGRAM MAIN CALL PRINTMESSAGE END and printmessage.f is C printmessage.f: define the PRINTMESSAGE subroutine SUBROUTINE PRINTMESSAGE V = 17 WRITE(*,*) V END I can use 'g77 -c main.f' to get main.o, 'g77 -c printmessage.f' to get printmessage.o, and 'g77 main.o printmessage.o' to get a binary (a.out). I can use 'ar cru libpm.a printmessage.o' to get a library containing the PRINTMESSAGE subroutine, then use 'g77 -L. main.o -lpm' to get the same a.out binary. But putting main.o in a libary doesn't work at all. I suspect that some of this is the behavior of the system linker with respect to libraries. But really, the best answer is don't do that: if it doesn't work on Linux, it seems unlikely to go on any other Unix-like system either. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: xfree86 4.3 installing on unstable
Bill Kalebaugh [EMAIL PROTECTED] writes: David Z Maze wrote: You probably don't want a full XFree86 4.3 install, just the X server for hardware support. On my laptop, I downloaded the Xxserv.tgz and Xmod.tgz binary tarballs from xfree86.org, unpacked them in /usr/local, repointed /etc/X11/X at /usr/local/bin/XFree86, and edited the XF86Config file appropriately. It looks like I still have the Debian xserver-xfree86 package installed, but I don't really use it. I tried following your instructions to the letter, now X will not come up, had to re point to /user/X11R6/bin. What did you change in the XF86Config-4 file It looks like the only substantive change was changing ModulePath from /usr/X11R6/lib/modules to /usr/local/lib/modules. [And having used this setup successfully from a couple of months, I switched to the packages in Debian experimental, and they seem to work fine. Installing xlibmesa-dri, also out of experimental, seems to have given me direct rendering support vs. the modules in the 2.4.23 kernel; with some hacking I was able to build xlibmesa-drm-src, but I'm not actually using it. This would seem to suggest that there's another tarball from xfree86.org you'd need for this functionality.] -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Filesystem Mounted On The Wrong Mount Points
Joseph A. Nagy, Jr. [EMAIL PROTECTED] writes: If I drop out of X (how do you kill X once it is going?) Ctrl+Alt+Bksp will kill your X server; if you're running under a display manager [gdm, kdm, xdm, wdm], that will probably restart the server, so you'd need to log in on the console as root and run /etc/init.d/xdm stop (or equivalent). and make temp folders and start moving stuff around, can I just rearrange everything (while making the appropriate changes in FSTAB)? Or can I just change FSTAB and reboot and everything is miraculously where it should be? Just changing /etc/fstab won't cause the data to magically move between partitions. If I were doing this level of shuffling, I'd probably shut the system down to single-user mode ('shutdown' or 'telinit 1' as root). If you're trying to move the root partition, I'd do it while booted from your favorite restore media; remember in that case to also update your bootloader configuration to know where the new root partition is. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: floating screen
Gruessle [EMAIL PROTECTED] writes: I did goggle for floating screen and floating display I don't know what else to look for you tell me what you would call this. If I change resolution with Ctrl Alt + or - to a smaller resolution I do get bigger everything, but my display bust out of the screen boundaries and when I move my mouse to one side it will float over so that I can see the side which I moved the mouse too. Right, XFree86 works that way. So how do fix this at the moment I have too small fonts, can't read the stuff, need to get to 1024x768 resolution If that's what you really want, you can edit your XF86Config file (or use 'dpkg-reconfigure xserver-xfree86') to not use higher-resolution display modes. I personally generally go for using larger fonts on higher-resolution displays. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Right ftp server
Gruessle [EMAIL PROTECTED] writes: Usually I chouse to make my own mind up, but man there is so match to read on all this if I don't get help I will be never done. Also usually I like to get the best, so I was thinking proftp at first. But I need something basic easy to config., something I don't have to read two books on before I can get going. Don't need anonymous I like X config stuff if there is something Which is the right ftp server for me? None, especially if you don't need anonymous access. Use scp to copy files to and from your machine (part of the ssh package, which includes a server). It's more firewall-friendly than FTP and doesn't involve sending your password over the network in the clear. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: xfree86 4.3 installing on unstable
Lukas Ruf [EMAIL PROTECTED] writes: is there any way to install xfree 4.3 on my Debian unstable box without switching to experimental? Thanks for any explanation on how I could achieve that! You probably don't want a full XFree86 4.3 install, just the X server for hardware support. On my laptop, I downloaded the Xxserv.tgz and Xmod.tgz binary tarballs from xfree86.org, unpacked them in /usr/local, repointed /etc/X11/X at /usr/local/bin/XFree86, and edited the XF86Config file appropriately. It looks like I still have the Debian xserver-xfree86 package installed, but I don't really use it. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Argument list too long
Dobai-Pataky Balint [EMAIL PROTECTED] writes: i have a webcam logging all motion into jpgs, my php code does lots with it, except after i convert selected jpgs copied into the workin' folder, and after conversion i'd like to rm them from php or by hand it'll give: rm *.jpg su: /bin/rm: Argument list too long so i have to enter mc, enter the directory and F8 them out. how could i do a right rm? (without generating a shell pattern to rm them in separated groups) You could use find/xargs; xargs' purpose in life is to break up command lines like this. Maybe find . -name . -o -type d -prune -o -type f -name '*.jpg' -print0 | xargs -0 rm (untested) to find JPEG files in the current directory, but not subdirectories, and remove them. Alternatively, find . -type f -name '*.jpg' -print0 | xargs -0 rm will find JPEG files in the current directory and all subdirectories and remove them. With fewer files, I'd advocate trying something like find . -type d -prune -o -type f -name '*.jpg' -print first to non-destructively check that it's getting the right list of files, then tack on the 'xargs rm' afterwards; sanity checking the find command's output can't hurt here. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: getopt (1) example scripts?
stan [EMAIL PROTECTED] writes: I see that my Debian machine has a getopt utility installed to use in parseing options to shell scripts. The man page mentions example scripts, but I cna't seem to find them on my machine. I also can't figure out the upstream source of this program, else I would just grab the sources, and unpack to see the scripts. Can anyone tell me where the source of this module is? {29} dmaze% which getopt /usr/bin/getopt {30} dmaze% dpkg -S `which getopt` util-linux: /usr/bin/getopt Separately, bash(1) documents a getopts (plural) builtin which looks to do something similar. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: dial-up to ethernet
mmissett [EMAIL PROTECTED] writes: /etc/network/interfaces has the following: first, it says the connection is static, which it isn't, it's DHCP. First off, I'd try fixing that. Run 'ifdown eth0', replace the current eth0 stanza with 'iface eth0 inet dhcp', then run 'ifup eth0' and see what happens. So, do I need to rewrite this file to make it correspond as much as possible to what the Mac is telling me? You could also assign yourself a static address on the same netblock. But it would need to be a different address from your router address and from your Mac. Did I cover everything? One (hopefully simple) question I'm still left with is: how do internet apps, from ping to Mozilla, know I'm not using ppp/ttyX any more but eth0 over a network? They don't have any idea at all. At some level, they ask the system, I'd like an outgoing connection to 192.25.206.10, TCP port 80, and the kernel deals with sending packets out over the most appropriate network interface. You can look at the relevant kernel configuration using, among other tools, 'ifconfig' and 'route -n'; if nothing ppp-related is listed there (especially in the route output), you're not trying to use a ppp connection. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Start Xserver problem
Stephen Liu [EMAIL PROTECTED] writes: I just finished another clean installation of Debian 3.0 with net-installer (Sarge) and encountered starting xserver problem Vedio card - WinFast GeForce2 MX RAM - 32MB driver chosen - nv ... (--) Chipset GeForce2 MX/MX 400 found (II) NV(0): Not using default mode 800x600 (insufficient memory for mode) (II) NV(0): Not using default mode 1024x768 (insufficient memory for mode) (WW) NV(0): Mode pool is empty (EE) NV(0): No valid modes found What I take from this: you're using the right driver for your card (GeForce2 MX does sound like an NVidia card, and 'nv' would be the right driver), and the X server finds your card, but doesn't realize there's memory available. (1024x768x32bpp would require exactly 3 MB of memory, which you should easily be able to handle.) If you look in /var/log/XFree86.0.log, does it list the amount of memory that's being detected? If it's detecting wrong, you might need to add 'VideoRam 32768' in the Device section of /etc/X11/XF86Config-4; you also might be able to specify video memory manually in 'dpkg-reconfigure xserver-xfree86'. (And if it asked you how much memory and you answered 32, that would cause this, though that's just speculation.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: yanvp
Dobai-Pataky Balint [EMAIL PROTECTED] writes: yet another nvidia problem. i got a 2.4.22-1-686, and the kernel headers for it, for wich nvidia-linux-1.0-4496-pkg2.run said that there is no file in include... From just recently running the NVidia installer, it does have a prompt for where the kernel includes are (maybe only in advanced mode); if you look in /usr/src, the right directory is going to be something like /usr/src/kernel-headers-2.4.22-1-686. so i got kernel-source, and symlinked it to linux, as it should, but than the compiled nvidia.o is not compiled with the right source, with which the image was. This isn't useful; kernel-source-2.4.22-1 is used to compile all of the kernel-image flavors, given the right .config file, but the -686 .config file isn't installed by default. It might be buried somewhere in the tree, but you'd also have to regenerate some other files. In any case, I'd delete the /usr/src/linux symlink, since nothing maintains it and so it's not necessarily guaranteed to link to the source for the running kernel. Alternatively, you might be able to do something with the Debian-packaged nvidia installer, using either a magic kernel-package invocation (against the -headers package would probably work) or using module-assistant. I don't have an incantation for either available right off hand, though. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Logbook - RCS
Gruessle [EMAIL PROTECTED] writes: I found following in the Configuration HOWTO manual. My question is what software do U use as Logbook and what in RCS? 2.2. Start the Logbook! To keep your installation in shape, it's essential that you know exactly what happened to your machine, which packages you installed that day, what you removed or modified, and so on. If I kept a logbook file, I'd do it using vi. Your favorite text editor should be more than adequate for this. (Though my Debian machines tend to be pretty close to stock; the only software on my laptop that's not either from a Debian package or in $HOME is an X server in /usr/local.) Make a backup copy of the system files you touch. Better still, use RCS; Install the rcs package, then read rcsintro(1). (Again, though, this is something I don't generally do, though I understand why it's Good Practice (TM). I do use version control for my Emacs dotfiles, if that makes things better.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: man files to text editor
Gruessle [EMAIL PROTECTED] writes: Is there a way I can open man files in a text editor. Emacs has a major mode for editing 'roff files, like man pages, if that's what you're looking for. I like to print one but have not configured my printer jet. 'man -t' will produce postscript from a man page; see man(1). This might be easier to move around or manipulate or print on other machines than the raw 'roff file. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Web server Partitions
Braxton Neate [EMAIL PROTECTED] writes: I know this is a question that gets asked a lot, but googling around I can't seem to find a good answer. I'm re-installing a web/sql server which currently has one large root partition and a swap partition. This is obviously not the best setup. I'm wondering what other people would recommend in the way of partitioning? The server is a 2x 800mhz PIII with 512MB RAM and a 40GB hard drive. It depends on your exact needs. Assuming you have no normal interactive users, I'd probably set it up as /var/www -- big enough, maybe 10-15 GB /var/lib/postgres (or whever) -- big enough, maybe 10-15 GB swap -- 0.5-1 GB / -- Whatever's left On this sort of system, the main benefits you get from partitioning are fault isolation: if something gets confused on your root filesystem, and fsck can't recover it, you haven't lost your data. Alternatively, if you decide to reinstall the system, this partitioning scheme lets you reinstall software but keep data. If you have a substantial amount of built-from-source software, you might also consider a partition for /usr/local for similar reasons. Adding more partitions, in my experience, doesn't make the system more manageable; a common thing to happen is to install the system with a small /var partition and then later realize that using APT is painful because it wants lots of space in /var/cache/apt. Or you make /usr too small, or /tmp too small, and run into problems later on. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: right kernal for AMD
Gruessle [EMAIL PROTECTED] writes: How do I find out what the right image is for a AMD 700? apt-get install kernel-image-2.4.18-bf2.4 apt-get install kernel-image-2.4.18-1-686 386, 586, k6 and k7 systems That's an Athlon chip, right? Probably 2.4.18-1-k7 is the best pick, since it also includes a patch for the recent local-root exploit that affected Debian, and it's the closest match for your processor. -386 should always work, though. (ObNitpick: AFAIK, only the Commodore 64 had a kernal; you could install VICE to get one. :p JSR $FFD2, baby!) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: XP-LINUX Dual Boot
Abhay Watwe [EMAIL PROTECTED] writes: I have a Dell Dimension 8300 which came preloaded with XP on the first hard drive. I added a second hard drive and installed Debian Woody on it. I installed LILO with MBR on /dev/hdb1 which is my / partition. The machine does not have a floppy drive, so I rebooted and used the Debian CD to boot into linux in rescue mode. Then I used the following command to create a linux.bin file [and more on trying to make the NT bootloader work] Both of the standard x86 Linux bootloaders, LILO and GRUB, support chain loading non-free operating systems; I've had excellent luck using GRUB to boot other operating systems (it's just for the games, honest!). GRUB in particular is a very capable bootloader; it understands filesystems, so you don't need to reinstall it if you install a new kernel, and you can do some amount of recovery work from the bootloader prompt if your system winds up hosed. From /boot/grub/grub.conf: title That thing on /dev/hda1 rootnoverify (hd0,0) makeactive chainloader +1 -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: changing roots on: dpkg -i
Rob Benton [EMAIL PROTECTED] writes: I'm trying to install the xfree86 packages with the /opt directory as root. I've tried using --instdir but the install fails on the pre/postinst scripts. Is there an easy way to do this without having to build my own package? No. In general, dpkg's options to change the root are useful if you have a chroot environment, or if you somehow otherwise have a complete working system installed somewhere other than / (e.g., you're booted off of a rescue CD and your hard disk is mounted on /target or something). The best you could do with this approach is install X stuff in /opt/usr/X11R6/..., and even that wouldn't work because the X server will do things like look for its configuration file in /etc/X11 (and has, in the Debian build, never heard of /opt). As far as X goes, IMHO the easiest way to get an XFree86 4.3 X server (because that's what you're really after, right?) is to download the Xxserv.tgz and Xmod.tgz binary tarballs from xfree86.org, unpack them somewhere like /usr/local, and repoint the /etc/X11/X symlink to point to them. There are also various backports, plus the ~official Debian experimental packages; search the list archives for details. Debian in general doesn't believe in /opt, and relocatable binaries are a hard problem that's not real high on the dpkg feature list. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: ICMP redirect
[EMAIL PROTECTED] writes: I would like to ask about the icmp messasages sending in linux 2.4.x. I have two subnets: 192.168.0.0/24 and 192.168.1.0/24 on the same ethernet segment. There is a gateway in each subnet (192.168.0.1 and 192.168.1.1). Clients use netmask 255.255.255.0. Routers have their own default routes (through DSL) and additional route to neighboring subnet: For example, router on 192.168.0.1: ip route add 192.168.1.1 dev eth0 ip route add 192.168.1.0/24 via 192.168.1.1 The problem is that client get redirect only for 1 hop: for instance 192.168.0.x client accesses 192.168.1.x, receives redirect from 192.168.0.1 that advises to send through 192.168.1.1. Unless I'm confused about what ip(8) does, there shouldn't be ICMP redirects issued at all. Reading RFC 777 makes it clear that, in this case, a redirect is inappropriate: a message from 192.168.0.17 to 192.168.1.34 goes to 192.168.0.1 first, which forwards it on to 192.168.1.1, but that's not on 192.168.0.0/24 and so the source and next hop aren't on the same network. How can I configure routers so the clients could send traffic directly to each other? You don't. You could configure the clients so that they know they're on both networks, though; it's probably easier if you gave each machine an IP address on both networks so things don't get confused. (Which leads to the question of why you're doing this. I have something similar set up at home, but one network gets NATted and the other goes through an IP-over-IP tunnel. My desktop machine has addresses on both networks; my laptop only on the tunnelled network, the wireless access point only on the NAT network. I can ping the WAP's IP address just fine, with everything going through the router, but that's not a big deal.) Is it necessary to add additional address'es to gateway so they would have their own addresses for each subnet? Probably wouldn't hurt; I'd guess it is necessary but can't say for sure. Or you could do all of this with just one gateway, still on both networks. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: fun stuff - dvd - 802.11g
Alvin Oga [EMAIL PROTECTED] writes: i've just gotten the distro to recognize the Netgear WG311 pci 802.11g card ( have a linksys card too but havent tested it yet ) - note that its g 54Mbps ... vs the slower 802.11[a/b] Nitpicking, 802.11a is also advertised as 54mbps, but runs on a different frequency than 802.11b/g. So on a mixed b/g network, you can wind up with the g clients slowing down to b rates to accomodate the slower hosts; since 802.11a runs on a separate frequency, it always runs at full speed. I also know there are some interoperability problems with 802.11g (I had trouble earlier this summer trying to use a Dell-brand 802.11b card vs. some big-name 802.11g APs), but those might be cleaned up with newer hardware. The downside of 802.11a under Linux is that - i got it working with the athros drivers from madwifi http://www.sourceforge.net/madwifi there's only one manufacturer of 802.11a chipsets (Atheros), and one driver for it, which works well enough for most purposes but isn't DFSG-free. - now the trick is to setup another box just like it and the two machines should be able to talk to each other over 802.11g ... I don't know whether things besides 802.11b support ad-hoc networking. But access points are comparatively cheap these days, and they pretty much all advertise 802.11g support. - not that WEP is NOT secure ... ( its been cracked ) Google-search for end to end argument; this doesn't concern me terribly (though my housemates think I shouldn't have a completely open 802.11b AP). and the other fun stuff... i just bought my first dvd .. w/out having a player or system setup ... - played w/ ogle, xine, mplayer, few others.. - after a few hours of fiddling ( installing various packages ) - dvd player is now working .. so now i need a new set of real speakers and real audio amps Which software package did you wind up using? -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Function of make dep : newbie :-?
Marcos José Setim [EMAIL PROTECTED] writes: Please, which the function of make dep? It depends on how the makefile is set up (search for a line that says dep:), but usually what this does is look through the source files for a particular program and generate extra Makefile lines such that, if one file changes, every file that includes it is rebuilt. Where meeting some article/documentation easy about make? Someone already pointed at the (make) Info file; you also might look at the GNU coding standards if you want to know what some of the common make targets are. Most of this documentation is aimed at developers, though; if you're just trying to build software and not modify it, reading individual packages' INSTALL files is probably a better way to go. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: I2C support
Jeffrey L. Taylor [EMAIL PROTECTED] writes: Is the I2C support only in 2.4.x? I would like to use the lm-sensors package. The packages in Debian claim to support kernel 2.4 and not kernel 2.6. The upstream Web page (http://secure.netroedge.com/~lm78/) claims that there's now userspace support for 2.6 kernels; IIRC, there are kernel modules for both i2c and lm-sensors in the 2.6 kernel tree. If you want to use lm-sensors with a 2.4 kernel, you need to build your own kernel at this point. Make sure you have everything i2c-related disabled in the kernel configuration, and build lm-sensors and i2c modules separately ('make-kpkg modules-image'; lm-sensors-source's README.Debian file should have information if you haven't done this before). If you don't completely disable i2c in the kernel configuration, you will get random lossage and kernel panics. It's really sad. I don't know a good way around it. If you're in the situation of needing other i2c-related drivers, there's a kernel patch at the same place that claims to update the in-kernel i2c support. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: I2C support
Jeffrey L. Taylor [EMAIL PROTECTED] writes: Quoting David Z Maze [EMAIL PROTECTED]: Jeffrey L. Taylor [EMAIL PROTECTED] writes: Is the I2C support only in 2.4.x? I would like to use the lm-sensors package. The packages in Debian claim to support kernel 2.4 and not kernel 2.6. I was thinking about 2.2, not 2.6. i2c-2.8.1's README says that it only supports kernel 2.4.9 and newer. i2c-2.6.5's README (which is the version in stable) doesn't contain this admonition; lm-sensors-2.6.5's README says that it requires kernel 2.2.0 and i2c-2.6.1 (or kernel 2.4.13). I mistakenly booted from CD 1 and ended up with a 2.2 kernel. (Actually, IIRC this was a network install). The box in question is a key server and I'd rather not have to monkey with installing a completely different kernel. You can get an updated 2.4.18 kernel fairly painlessly via APT; searching aptitude's package list for 2.4.18 should help you find it. There still aren't prebuilt kernel modules, though; if you get the Debian kernel source for whichever kernel you have installed (maybe just the -headers package is sufficient) you can probably build modules from the {i2c,lm-sensors}-source packages. This doesn't appear to be a terribly well-documented process, alas. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: make-kpkg
Michael Montagne [EMAIL PROTECTED] writes: I use Grub as a bootloader. After making a kernel .deb using make-kpkg, I'm running dpkg -i Near the end you are asked to if you want to make a boot block. What is this? Is it just an entry in Grub or LILO? What I'm most concerned about is being able to boot to my old kernel if I screwed this one up. I think that's specifically a hook to LILO; I always say no, and if you're using GRUB, you don't need to reinstall the boot block when you update kernels. You might leave an entry in /boot/grub/grub.conf that boots 'kernel /vmlinuz' and 'initrd /initrd.img', and a second one that does /vmlinuz.old and /initrd.img.old; that way you should, in principle, always be able to boot your current and previous kernels without ever touching your GRUB configuration. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Buggy Kernel How-To?
Wolfgang Pfeiffer [EMAIL PROTECTED] writes: So, in short, what I found by googling about for some time: The correct way seems to be to put the kernel-source in my (non-root) home directory, and then cd /usr/src/ ln -s /home/someuser/kernel-sources linux and then, as non-root, compile the kernel in /usr/src/linux/ (And then forget about some of the stuff I read in the Kernel-HowTo ?) I'd certainly believe that the Kernel-HOWTO isn't the best source of information for compiling kernels on Debian. Unpack, build, and install everything as root will *work* on every Linux out there, even if it's unsafe. I'd look at the kernel-building documentation on http://newbiedoc.sourceforge.net/ (specific to Debian). None of my machines have a /usr/src/linux. I don't miss it. On my laptop I build kernels in /home/dmaze/src/kernel/kernel-source-$KVERS; my home desktop machine builds kernels for both itself and my firewall machine in /usr/local/src. Real root privileges are only involved in building the kernel when I install the kernel-image packages using dpkg and the subsequent reboot. :-) The background to all this is that I tried to get the kernel sources as non-root while being in /usr/src/some.kernel.directory with rsync: Which, IIRC, isn't possible. A non-root doesn't have the permission to download stuff to this dir, right? Add yourself to the 'src' group to get write access to /usr/src; the 'staff' group for /usr/local/src. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: shell script question
Han Huynh [EMAIL PROTECTED] writes: I know this isn't a bash/korn shell script news group, but the fact is I can't find one. Since bash/ksh is the default linux shell, I was hoping someone could answer a few pretty simple questions. Is there any way to export a variable for one parent shell to a different parent shell? I know that export will work to a subshell, but I can't find any process to return a variable to a different parent shell. Can't be done. What are you really trying to do? The . command will read in a child script in the current shell, so you can do things like write a generic .variables script and then run . $HOME/.variables in both your shell startup file and your .xsession to read those in. Another useful alternative for really short things is shell functions; I have some shell code that looks like ATHENA_SYS=i386_linux24 add() { if test -d /mit/$1/arch/$ATHENA_SYS/bin; then PATH=$PATH:/mit/$1/arch/$ATHENA_SYS/bin; elif test -d /mit/$1/bin; then PATH=$PATH:/mit/$1/bin; fi } so I can run 'add outland' or whatever to add an MIT locker to my shell path; there's an automounter running on /mit. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: IP addresses.
Geoff Bagley [EMAIL PROTECTED] writes: My firewall captures the IP addresses of people who ping my TCP ports. They consist of the standard four eight-bit integers. The firewall whois command is able to back-track some of these, others not. Is there a programme, Linux or Windows, or a Web Site, where I can look up such numbers, and convert them to normal internet addresses ? Those *are* normal internet (IPv4) addresses. Do you mean you want to reverse-resolve these into hostnames? Any standard DNS tool (probably either 'host' or 'dig', depending on how much information you want) can do this. Of course, there's no guarantee any particular IP address will reverse-resolve. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: How do I temporally disable gdm? (or is it gpm)
Mark Healey [EMAIL PROTECTED] writes: I've received a number of suggesions for my X problem. Before I try any of them I'd like to disable my graphical login screen. It is either gpm or gdm. I can't remember which one is the login screen and which is the console mouse driver. gdm is the GNOME Display Manager, much like xdm (X), kdm (KDE), and wdm (Wings). gpm is the console mouse driver. Anyway, I'm hoping that there is a line in a file I can comment out. Delete /etc/rc2.d/S99gdm, which is a symlink telling the init scripts start gdm as the last thing in runlevel 2. Or, if you just want to stop it for this session and have it start again at the next boot, run '/etc/init.d/gdm stop'. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Ethernet card not detected - where is YAST2 ?
[EMAIL PROTECTED] (Sebastia Altemir) writes: My LINEX (Spanish Debian with 2.4.20 kernel) instaled ok, but the ethernet card was not detected, so nothing (eth0) was instaled. lspci -v says 02.0c.0 MYSON Tech Inc : unknown device 0803. If I run modprobe fealnx, it runs OK (lsmod displays (unused)) Two ways to do this: either add 'fealnx' on a line by itself in /etc/modules, which has a list of modules that get loaded at boot time. Or, add 'alias eth0 fealnx' in a file in /etc/modutils (maybe create a file there named with your hostname) and run update-modules as root; this causes the driver to get loaded when you try to bring up the et0 network interface. then ifconfig eth0 1.2.3.4 ... goes ok ( lsmod display (1)), and I can ping near computers. You need to edit /etc/network/interfaces; read interfaces(5) for details, or see /usr/share/doc/ifupdown. Having set up that file correctly, 'ifup eth0' will bring up and configure the interface, 'ifdown eth0' will take it down. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [DSA-403-1] Kernel update?
Harshwardhan Nagaonkar [EMAIL PROTECTED] writes: So, does this mean that I can compile my kernel without initrd, and it will still not break on debian? I understand that this will involve editing /etc/lilo.conf and getting rid of the initrd line. Any other gotchas that I should know about? Yes, that's right. The important thing is that you need to make sure the drivers for your root disk and filesystem (probably IDE disk and ext2, but these could both be other things) are built into the kernel, not built as modules. Just make sure you have a LILO entry around to boot your old kernel if you get stuck. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [DSA-403-1] Kernel update?
Bill Moseley [EMAIL PROTECTED] writes: On Wed, Dec 03, 2003 at 10:43:43AM -0500, David Z Maze wrote: Yes, that's right. The important thing is that you need to make sure the drivers for your root disk and filesystem (probably IDE disk and ext2, but these could both be other things) are built into the kernel, not built as modules. Just make sure you have a LILO entry around to boot your old kernel if you get stuck. So is the purpose of initrd to have a small kernel but be able to load whatever modules might be needed for the currently running hardware? Almost. You also get to have a kernel that's completely generic; if your machine has a SCSI hard disk with reiserfs, and mine is IDE with ext3, the an initrd/kernel pair could boot both, without having any of the drivers compiled in. If so I often wonder why that's needed -- that is, why not just build a kernel with everything compiled in? If, say, installing from CD then size of the kernel isn't critical (not to mention that the modules take up space, too), and most machines have quite a bit of RAM these days, so I would not think that an issue either. My fuzzy memory is that there are a couple of factors. One is that memory for kernel drivers absolutely can't be used for anything else, so if you're trying to get the last megabyte out of your system, an unused module is cheaper than an unused in-kernel driver. Another is that there are a couple of limits on the size of the kernel, and so building everything in blows you over that limit pretty quickly. For a distribution kernel you also might want to install it on floppies, which gives you a hard limit on the size of the kernel. (But yes, all of these are becoming less of an issue with more modern hardware.) Which brings me back to the point that maybe I don't really understand the need for initrd... It makes sense for a distribution kernel (Debian's default kernel); IMHO it's more of a pain than it's worth for home-built kernels, so kernel-image-2.4.23-$HOSTNAME for me is never an initrd kernel. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Debian Investigation Report after Server Compromises
(Not speaking for Debian at all.) Dr. MacQuigg [EMAIL PROTECTED] writes: 1) What is a sniffed password, and how do they know the attacker used a password that was sniffed, rather than just stolen out of someone's notebook? It sounds like someone's personal machine got broken into, and a keystroke logger installed. Then they did something like upload a package and typed their password on the Debian machines, and the attacker was able to capture the username and password. 2) Was the breakin done remotely, or by someone with physical access to the machine or network? I thought that sniffing required physical access to a network over which unencrypted data was being transferred. Are the remote logins to Debian servers unencrypted? I think there's only ssh. (So if you broke into the machine and installed a compromised ssh binary, that could work to steal a password too.) Captured password might be more correct than sniffed. But I haven't heard anything that suggests the attacker had physical access to anything. 3) How does an attacker with a user-level password gain root access? In this case, there was a bug in the kernel that let a user process do pretty much anything it wanted to, assuming I understand its implications correctly. I understand you can call system services that have root access, and provide bad data in those calls that will cause buffer overflows, maybe even a machine crash, but how does a buffer overflow allow root access? I know there is a deep technical explanation for this, but I'm hoping someone can explain it in simple terms, or maybe point me to a good article or book chapter. The usual way this happens is that you have a daemon running as root. Somewhere there's data being read, and past the end of the data is a pointer saying where the function should go when it returns. So a typical buffer overflow attack knows where it expects to be in memory, and overwrites a fixed-length buffer with more than the expected amount of data, rewriting the return pointer to point to some code that also lives on the stack; when the read_input() function returns, instead of returning to its normal caller, it returns to the attack code, which is now running as root. (But note that this is different from the exploit used to gain root on the Debian servers; there are multiple sorts of vulnerabilities and therefore multiple exploits.) -- David Maze [EMAIL PROTECTED] http://www.mit.edu/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Alias help needed
Bob Tilley (ATT) [EMAIL PROTECTED] writes: I wish to use the expression catl filename to execute the commands cat filename | less. Is it possible to use 'alias' in the definition? alias catl='cat $0 | less' seems like a good idea, but I don't know if the '$0' works in a simple alias. I think in Bourne-style shells (bash, ash, ksh, zsh), you almost always want to use shell functions instead of aliases: catl() { cat $@ | less } (Where $@ is sh-ish for all of the rest of the command-line arguments, quoted correctly.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Kernel Security Update - 2.2?
Jacob S. [EMAIL PROTECTED] writes: Ok, I haven't seen anyone else ask it, so I'll ask the dumb question I couldn't find an answer for. :-) Is the 2.2 kernel series affected by the bug found in the 2.4 and 2.6 kernel tree? My assumption would be yes, but if not, it would save me some work. In other words, I'm trying to figure out how soon I need to learn how to setup iptables to replace my aging ipchains. :-) Note that, independent of everything else, you can still use ipchains under kernel 2.4; my home gateway machine does this. Alas, the configuration is sufficiently complicated that I actually need to learn how it works, rather than just doing something like installing the ipmasq package. :-) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: unchecked 31 times
Monique Y. Herman [EMAIL PROTECTED] writes: On Mon, 01 Dec 2003 at 16:55 GMT, Alan Shutko penned: Nick Welch [EMAIL PROTECTED] writes: I suppose mke2fs(8) is where that comes from specifically. Easy to disable the periodic checks, though: tune2fs -i 0 -c 0 /dev/hda6 That's a very bad idea. Wait, wait; I'm confused. I thought one of the perks of running a journalling file system was that you can speed up the boot process by disabling boot-time fsck? The relevant perk is that, if the system shuts down abnormally, the boot-time fsck gets to replay the journal, which is fast, rather than actually having to go through and do the full check. If you have bad hardware, no filesystem is going to be completely safe against random lossage; running periodic full fscks just in case is good practice. (But turning the check frequency down is almost certainly a practical thing to do.) -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: installing new latex document type - bst file not found by documents
Micha Feigin [EMAIL PROTECTED] writes: On Wed, Nov 26, 2003 at 10:53:04PM -0500, David Z Maze wrote: Micha Feigin [EMAIL PROTECTED] writes: I just installed the files for a new document type from cpan (latex8). I put them into /usr/share/texmf/tex/generic/latex8 and ran texhash. (/usr/local/share/texmf is probably a better pick; in this case, it's easier to distinguish locally installed from dpkg maintained, but there's also a little more of a guarantee that various Debian packages won't step on it.) The document type is recognized properly (the sty file is found by latex), but when trying to use the bibliography style, (latex8.bst) without a full path it is not found. You need to use one of the directories reported by 'kpsepath bst', or customize your texmf configuration. /usr/local/share/texmf/bibtex/bst is probably the best directory on the list. Thanx, moved the whole package to /usr/local/share/texmf/latex8/..., linked the bst file from /usr/local/share/texmf/bibtex/bst and ran texhash. Now the bst file is found but the sty file isn't found by latex. Similarly, 'kpsepath tex' will tell you where .sty files need to go; probably /usr/local/share/texmf/tex/latex8/ (or even tex/generic/latex8, going with your original directory layout) would work. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Xresources Xdefaults documentation
Micha Feigin [EMAIL PROTECTED] writes: I picked up on most of the syntax, but there are a few points, such as it appears that using XTerm. affects all exterms even if they were called through a link (x-terminal-emulator for example) while xterm. affects only when called as xterm or whats the difference betwin xterm*fonts and xterm.fonts (some apps appear to prefer it one way and some the other some take both). Every widget in the resource tree has a name ('xterm', 'x-terminal-emulator') and a class ('XTerm'). Generally the man page will list both. Dots separate items in the tree; a star between items indicates that zero or more items may be skipped. So in the particular case of xterm, that has a widget named vt100 (class VT100), which lets you specify a font (class Font). So these are mostly equivalent: XTerm.vt100.font: fixed XTerm.VT100.Font: fixed XTerm*Font: fixed (...and all apply to all xterms, regardless of the name they were started under.) These make a difference if you have multiple things with the same class: xclock.foreground: black xclock.hands: gray40 ! or, would set both of the above: xclock.Foreground: black There's a resolution order documented somewhere if you have conflicting resource settings, which mostly works out to the most specific thing takes effect. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Automatic debsums generation
Brian McGroarty [EMAIL PROTECTED] writes: 4. Any debs you built yourself will need manual sum generation with a command similar to step 2 pointing to the directory with the debs. If you're going off building (or rebuilding) .debs yourself, and the package is debhelper-based, it's easy enough to add the relevant md5sums file using dh_md5sums, which (at least on unstable) is part of the standard debhelper distribution. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Poll: who's apt-get upgraded sid this week?
Tom [EMAIL PROTECTED] writes: How many people have apt-get their SID boxes since the indicdent? Everything seems to be back up... Is it sitnorm again? I'm continuing to run aptitude ~daily, but my mirror hasn't seen any updates since the compromise. -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Package question
Vanh Phom [EMAIL PROTECTED] writes: Where can I find package Berkeley DB3? Start aptitude, and search for packages that have both 'Berkeley' and 'database' in their descriptions, as in / ~dberkeley~ddatabase -- David Maze [EMAIL PROTECTED] http://people.debian.org/~dmaze/ Theoretical politics is interesting. Politicking should be illegal. -- Abra Mitchell -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]