Weird Routing/IP-Masquerading issue
orwell has two ethernet cards and serves as the router for my home network. eth0 connects to a cable modem with IP 24.x.x.x.x (assigned via DHCP). eth1 connects to the home network with ip 192.168.1.1. The routing works fine, and I never have any problems getting to the outside world from any of the other systems on the network...except for orwell. When I'm actually physically logged into orwell, I'm sometimes unable to establish TCP connections with the outside world. I just ran fetchmail on magellan (192.168.1.2) and it connected to my ISP's POP server fine. But, even as I speak, fetchmail on orwell is blocking on the connect() call because it cannot establish a TCP connection with the outside world. I'm totally bewildered. Twenty minutes from now, it could work fine. Instead of using the ipmasq package, I setup the network and some special IP-Masquerading hacks (for Napster, DirectX, and ICQ) through /etc/init.d/network, which I've attached. I also use portfw to forward orwell:81 to magellan:8080 to let others access my Zope server, but I don't think that that is an issue either. I'm running Debian potato and kernel 2.2.12 on orwell. Does anyone have any ideas? Please CC me as I'm not currently subscribed to debian-user (I can only handle one high-traffic list, and zope takes the cake!) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org #! /bin/sh ifconfig lo 127.0.0.1 # internal network is hard-coded; external is setup by DHCP ifconfig eth0 > /dev/null ifconfig eth1 192.168.1.1 netmask 255.255.255.0 ipchains -P forward DENY ipchains -A forward -s 192.168.1.0/16 -j MASQ echo "1" > /proc/sys/net/ipv4/ip_forward # for DirectPlay games :-) ipmasqadm autofw -A -r udp 2300 2400 -h 192.168.1.2 ipmasqadm autofw -A -r tcp 2300 2400 -h 192.168.1.2 ipmasqadm autofw -A -r udp 47624 47624 -h 192.168.1.2 ipmasqadm autofw -A -r tcp 47624 47624 -h 192.168.1.2 # for napster ipmasqadm autofw -A -r tcp 6699 6699 -h 192.168.1.2 # for ICQ ipmasqadm autofw -A -r tcp 31000 32000 -h 192.168.1.2
Please help a poor gaijin!
Hi! I'm trying to figure out what level of Japanese input support exists in Debian. Ideally, I'd like to be able to receive messages written in Unicode with mutt (via the mutt-ja package), and type messages/documents in kana/kanji with vim and have them stored as Unicode. Being able to print Japanese text would also be really nice. I'm a bit overwhelmed by how all of the packages work together. Do I need kinput2, canna, mutt-ja, and jvim-canna? Will I need some type of kana/kanji dictionary? Do I need to 'enable' kinput somehow, perhaps by editing my XF86Config? The English documentation is a little confusing and my Japanese fluency isn't enough to begin to try and understand the stuff written in Japanese. Thanks to any and all who can shed light on this for me. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Fetchmail Errors in Potato
> > > I was wondering if anyone has received errors when receiving mail through > fetchmail (both version 4.6.4-1.1 and 5.0.5-1). When I upgraded to potato > a couple days ago I having been getting the following error message from > the fetchmail log: > > > fetchmail: Starting fetchmail 5.0.5 daemon > fetchmail: 5.0.5 querying uclink4.berkeley.edu (protocol IMAP) at Thu, 26 > Aug 1999 13:24:08 -0700 (PDT) > fetchmail: Authorization failure on [EMAIL PROTECTED] > fetchmail: socket error while fetching from uclink4.berkeley.edu > fetchmail: Query status=2 > fetchmail: All connections are wedged. Exiting. > What's the output of 'fetchmail -v'? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Help with compiling
On Thu, Aug 26, 1999 at 01:41:20PM -0700, Dan Smith wrote: > I have downloaded and installed g++ and binutils and > all the packages they depend on. How do I compile and > link a C++ program? e.g. what is the proper syntax of > the c++ and ld commands? > Please send answers to me directly > ([EMAIL PROTECTED]) as I am not > subscribed to the list. Thank you. You shouldn't have to invoke ld directly. To compile a program consisting of foo.c, foo2.c, and foo3.c which depends on the gtk and mysql libraries (hypothetical), you'd probably want to do something like this: g++ -O2 -Wall -pedantic -ansi -c foo.c g++ -O2 -Wall -pedantic -ansi -c foo2.c g++ -O2 -Wall -pedantic -ansi -c foo3.c g++ -o foo_program foo.o foo2.o foo3.o -lgtk -lmysqlclient or this: g++ -O2 -Wall -pedantic -ansi -o foo_program foo.c foo2.c foo3.c -lgtk -lmysqlclient -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: z/bzImages, shell scripts
On Thu, Aug 26, 1999 at 08:00:57PM -0600, Tom Lineman wrote: > NOTE: I sent this message five minutes before I got my mailing list > subscription confirmation. Since the confirmation message said all my > submissions would be returned, and I didn't get this mail back, I am > assuming that my mail was discarded...HOWEVER if it was, in fact, accepted > and you have already read this message before, please ignore this copy as it > is just that: a copy (albeit with this note added). > > > > Hello there. I have two questions regarding Debian v.1.3.1 > > First of all, I'm running a 486 that has two partitions, one for MSDOS > and one for linux. Currently, my system boots off of the DOS partition. To > boot to linux, I have to insert my custom boot floppy and boot up. What I > would really like to do is use loadlin to boot to linux, but I really have > no specific information concerning bzImages or zImages. If someone could > give me some info or a URL, I would really appreciate that. One question: > would it be possible to use a line somewhat like this: > loadlin linux root=/dev/hda2 /dev/initrd=root.bin > so that loadlin boots up using the "linux" and "root.bin" files? Or would > it be better to use a zImage? I'm not sure what you are asking. The 'linux' file is a zImage, a compressed copy of the kernel. A bzImage is a copy of the kernel that is large enough to warrant special code to get it in memory. Also, you don't need to specify /dev/initrd=root.bin, because the kernel will load its root filesystem from /dev/hda2. > > My second question has to do with small shell scripts. I was reading a book > on Unix the other day, and it talked about "aliases," which it stated only > ran under Berkely *nixes. I'm guessing that means FreeBSD, NetBSD, OpenBSD, > or BSDI, but can Debian linux use aliases, too? Not to be pedantic, but aliases are a function of the shell (usually bash, but sometimes tcsh or zsh). AFAIK, bash (or whatever shell you are trying to use) runs on all of the above operating systems, so you should have no problem. > The reason I'm asking is > that I wrote some small shell scripts to make life a little bit easier. For > example, let's say I wanted to make a program called "delete" that would act > like this: > rm -iv > so that I could "delete" a file instead of "rm"'ing it, like so: > $ls > deletethis.zzz > $delete deletethis.zzz > Are you sure you want to delete deletethis.zzz? y > deletethis.zzz deleted > $ls > $ > I did that, but every time I ran "delete" I would get something like this: > $delete deletethis.zzz > Parameter not good (not verbatim, but it was close to this) > use 'rm --help' for more details > The same thing happened if I tried making scripts containing mv -iv and cp > -iv. What is the problem here, and just as importantly, what is a posssible > solution? What _exact_ commands are you using to set the aliases? > P.S. I know this is probably a silly question, but if my CD-ROM drive is the > second device on my first IDE Controller, would it be /dev/hdb ? Yup, you're right on the money. BTW, The only silly questions are the ones left unasked. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: removing kdm
On Fri, Aug 27, 1999 at 11:12:20AM +0900, [EMAIL PROTECTED] wrote: > Hello, I have a problem of dselect message. > I was trying to remove kdm. However the following error message appeared, and > I can not go further now. Please somebody give me any suggestion. > > ---error message- > The following packages will be REMOVED: > kdm > 20 packages upgraded, 0 newly installed, 1 to remove and 15 not upgraded. > Need to get 0B/9822kB of archives. After unpacking 435kB will be freed. > Do you want to continue? [Y/n] > (Reading database ... 30212 files and directories currently installed.) > Removing kdm ... > dpkg - warning: while removing kdm, directory `/usr/X11R6/lib/X11/kdm' not > empty so not removed. > dpkg - warning: while removing kdm, directory `/etc/X11/kdm' not empty so not > removed. > /var/lib/dpkg/info/kdm.postrm: line 15: syntax error near unexpected token > `kdm' > dpkg: error processing kdm (--remove): > subprocess post-removal script returned error exit status 2 > start-stop-daemon: stat /usr/bin/kdm: No such file or directory A quick and dirty hack is to look over the postrm script (/var/lib/dpkg/info/kdm.postrm), and either fix it or do what it does manually and delete it. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: modules loading (possibly zip driver quesiton too...)
On Fri, Aug 27, 1999 at 08:47:20AM -0400, Jon Hughes wrote: > I just posted a question last night about some problems with a ZIP drive. > One thing I read that could be a possible solution is the module lp is > loading before ppa (at least, that's what the book says...). I haven't yet > figured out in Debian where to set up which modules load when in what > filecan anyone point me in the right direction? > The best way is to have the kernel load modules on demand via kmod. Unfortunately, there are times (ide-scsi comes to mind), where there is no elegant way to do that. /etc/modules contains a list of modules loaded at boot time. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Working with mail in /home not /var
On Fri, Aug 27, 1999 at 05:17:15PM -0400, Felipe Alvarez Harnecker wrote: > > Hi, is there a simple solution to move folders in > /var/spool/mail/$user to /home/$user/mailbox, i'm working with exim and > qpopper and symlinks seems not to work. > > Thanks. > One way is to implement a per-user ~/.forward file. Check /usr/doc/exim/filters.txt.gz. This has the added benefit of being able to allow your users to sort their email as well. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: kde
On Fri, Aug 27, 1999 at 12:48:22PM -0500, jason wrote: > anyone have a working apt entry to get kde? snowcrash doesn't seem to like > me anymore.. > deb ftp://kde.tdyc.com/pub/debian potato kde contrib works fine, snowcrash merged with a KDE rpm site and they got a new server. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: hardware...
On Tue, Aug 24, 1999 at 12:43:18PM -0500, Don Custer wrote: > To the good folks at Debian... > > I have two short questions which I hope can be answered with two short > answers. I failed to find pertaining answers in the FAQs. > > 1. I am going to buy a new computer (to replace my 1991 dinosaur). >I want it to be powered by Debian Linux. Are there any particular >brands that are more friendly toward Linux? If it would make this an >easier question to answer... are there any particular brands to stay >away from? I don't want to end up with a pile of incompatible junk. Stuff to consider: * drivers don't exist for sound cards with Aureal chips and the new Soundblaser Live. * most PCI modems are winmodems, and some ISA ones are, too. * video cards: stay away from 3dfx stuff, I'd recommend Matrox or NVidia based products because both companies are actively working on 3D drivers for Linux and have been open about specs * USB stuff other than keyboards, mice, and some printers doesn't work and won't work until at least kernel 2.4, due at the end of the year * Make sure your printer either speaks Postscript or PCL! Others might be 'winprinters' * SCSI scanners work and are supported the best If you want to support Linux-based businesses, order a system from VA Research (http://www.varesearch.com). > > 2. In regard to this Y2K frenzy... do you folks know of any reason >(other than price) for waiting until after Y2K to buy a computer? >I am referring to any problems in today's systems that I may not >be aware of. Nope..most of the problems seem to be in Microsoft's OS. Most BIOSes have been compliant for quite some time. Because of the way Unix OSes store time, a 32-bit computer is safe until 2038, and once we move to 64-bit processors, we'll be fine for a couple trillion years :-) > > Any insight will be greatly appreciated. I am not a Linux Guru. I have > used RedHat. It's nice, but reminds me of Microsoft. I have also played > with SlackWare, no complaints on it. But, after seeing the Debian system > my son has setup on his computer... well... you got me hooked! I've used Slackware, RedHat, Caldera, and Debian and I've been a happy Debian user for 3 years. Debian is cool and with the Corel and Stormix alliances, it can only get better :-) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: considerations for kernal upgrade
On Wed, Aug 25, 1999 at 03:12:52PM +1300, [EMAIL PROTECTED] wrote: > Hi > > i am running debian ver 1.2 which uses libc5. I need to upgrade to libc6 > in order to install the latest smail package so that i can control relay > spamming. > > Would i need to install a more recent version of debian so that the > packages using libc5 will now run on libc6? > > Whats the best way for me to upgrade?? Forget a libc upgrade. Why not just grab the source to the smail packages, use dpkg-source to extract them, and run 'debian/rules binary' to build yourself a libc5 deb of the newest smail? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Libraries
On Thu, Aug 26, 1999 at 11:21:35AM -0300, Guilherme Balena Versiani wrote: > Hello, > > I want to know how to make the _pic.a format libraries found in > Debian distribution. I want to create reduced shared libraries with > these ones to make a small filesystem image for a boot/root disk. > I need to know at least where I could find these files: > - libncurses_pic.a (version 3.0); > - libc5_pic.a; > Or, maybe, can someone send libraries to me? > Look at the scripts in the source of the boot-floppies package. Shared libraries (.so) would probably be a better idea than static if you are looking to conserve disk space. By definition, no one else could send you libraries that are optimized for YOUR specific task, so I'm not sure why you want others to send you libraries. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: mounting problems
On Thu, Aug 26, 1999 at 04:43:31PM +0200, Jocke wrote: > Hi all, > > Been recompiling my 2.2.10 kernel to be able to use hp 8100+ CDRW. > I think I got it all right but some strange things did happen. > > I can't mount my ordinary cd anymore! > mount: wrong fs type, bad option, bad superblock on /dev/hdc, >or too many mounted file systems > > Where do I start looking? > > Here is some information. > in fstab: > /dev/hdc/cdrom iso9660 ro,user,noauto 0 0 > > cdrecord -scanbus > Cdrecord release 1.8a23 Copyright (C) 1995-1999 Jörg Schilling > scsibus0: > cdrecord: Warning: controller returns wrong size for CD capabilities page. > 0) 'SONY' 'CD-ROM CDU4011 ' 'UY04' Removable CD-ROM > 1) 'HP ' 'CD-Writer+ 8100 ' '1.0g' Removable CD-ROM > 2) * > 3) * > 4) * > 5) * > 6) * > 7) * > > Can anyone help me ? > You compiled in SCSI emulation for IDE devices, right? That means that, to user-level programs, your CDROMs are /dev/scd0 and /dev/scd1. You'll have to 'modprobe ide-scsi' before using the CD-ROM devices, or add its name to /etc/modules so that it is autoloaded on startup. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Wierd compile error with libpthread.a
On Thu, Aug 26, 1999 at 12:13:34AM -0400, Rupert Brooks wrote: > Hi, > > Please note to CC replies to [EMAIL PROTECTED] and [EMAIL PROTECTED] I > don't follow this list all that regularly - thanks > > I am having problems compiling a c program under Debian-Potato. I have > traced it this far: If I compile any C program, and link in the library > /usr/lib/libpthread.a then an executable will be produced, but when > executed will simply lock up. Not quite. From what I can tell, you aren't linking pthread to your program, you are including it as one of the object files in your program. So, gcc is shoving the whole libpthread.a in your binary. This works better: cc -funroll-loops -fomit-frame-pointer -pipe -Wall -g main.o pppoe-discover.o -L. -lpcap -lnet -lpthread -o pppoe Instead of directly specifying the path of libraries, the best way is to link to them with '-l'. The linker looks first for a shared (.so) library (because the -shared option is implied), and then for a static (.a) library. '-L.' tells the linker to also look for library files in the current directory. The resulting executable is dynamically linked with pthread (a standard libc component..no need to statically link this) and libc, and statically linked with everything else, per 'ldd pppoe'. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: how to create a dummy /dev/dsp
On Thu, Aug 26, 1999 at 02:01:19AM -0400, Stefan Langerman wrote: > Hi, > > I have slink installed on an old Pentium 75 box, WITHOUT a sound card. > When I installed the system, it apparently didn't create a /dev/dsp. > How do I create one that doesn't do anything? > > Thanks, > cd /dev; MAKEDEV audio -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: any gif animator for linux!!
On Thu, Aug 26, 1999 at 01:56:01AM -0700, ZEN MYSTIC wrote: > > is there any gif animator available for linux...if so > pls tell where to get... > Try grabbing a newer (1.1.x) version of the gimp from http://www.gimp.org. I'm pretty sure it has support for some types of animations. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Logitech Mouseman / X
On Wed, Aug 25, 1999 at 09:18:11AM -0400, Fraser Campbell wrote: > Hi all, > > Just got a new computer and cannot get the mouse to work with X no matter what > I try. The mouse is 3 buttons, PS/2 and labeled as a Logitech Mouseman. > > gpm works fine with it. X does not. I have tried every mouse type with it > and > the mouse cursor simply jumps to the top left corner of the screen as soon as > it > is touched. > > I have disabled gpm so it is definitely not a conflict with gpm. > > Any ideas? Reply with the contents of the Pointer section of /etc/X11/XF86Config (start with 'Section "Pointer"' and end with 'EndSection') -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: more RAM = more speed?
On Wed, Aug 25, 1999 at 08:14:06AM -0700, Patrick Olson wrote: > 1. hdc has been a CD-ROM since the second I booted the Debian install CD. > Will anything bad happen if hdc is suddenly a hard disk and hdd is now the > CD-ROM? I haven't actually used the CD-ROM since I finished installing, > so the only possible trouble spot might be booting. Booting won't be a problem, since you aren't moving around hard drives. Check /etc/fstab to make sure there is no reference to hdc. Otherwise, you are fine. > > 2. Is there a a how-to or some doc's for this? I made a swap partition > with cfdisk in the beginning and let the (excellent) install program > figure out the rest. Thus, I never learned much about swap. The 'mkswap' program makes a swap partition. Make a partition (/dev/hdc1) and run 'mkswap /dev/hdc1'. Use -c to check for bad sectors. Use 'swapon /dev/hdc1' to enable the swap partition. Add a line to /etc/fstab similar to your existing swap line to automatically enable this swap partition on boot. 'swapoff /dev/hdc1' will turn off your swap partition. > > 3. Since any drive I add will be old and slow (<200MB), is it worth it? Try it out and see. I'd say yes, because one of the drives wouldn't be going back and forth, moving the head all around. If it doesn't make any difference, you can use it to back up data or something. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: network firewalling situation
On Wed, Aug 25, 1999 at 03:43:10PM -0400, Dan DeMond wrote: > Hi all, > I'm setting up a machine here to take over the job of our firewall > temporarily. Both machines have the same network cards in the same > configuration. The current firewall is a debian 2.0 machine running > kernel 2.0.29 on an i486. The temp machine is an i386, debian 2.1, kernel ^^ > 2.2.10. I'm using the same setup scripts in the same order to make the ^^ > computer route packets between two networks, but it just isn't working on > the i386. Can anybody tell me what I'm doing wrong? The firewalling stuff changed in 2.2. ipchains is now used instead of ipfwadm. Either run 2.0.37 instead of 2.2.10, or convert your scripts to use ipchains. For more info, Check out the IP-CHAINS-HOWTO or the IP-Masquerading-mini-HOWTO, both available in /usr/doc/HOWTO. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: slink --> potatoe (hi dan quayle! :)
On Mon, Aug 23, 1999 at 03:06:10PM -0700, Seth R Arnold wrote: > hi all, > > I recently bought slink from cheapbytes ($12 for 4 CDs! wow! :) so I could > install it before I get to school... but, it seems to me that it comes with > ancient versions of some software (windowmaker 20.x I think...) so I would > like to upgrade to potato. > > Is there a nice easy way of doing this? I remember seeing a link on the > debian homepage that discussed what was involved in moving from 2.0.x to > 2.2.x (I did that manually with SuSE, an experience that .. I would rather > avoid again, but I did learn from it. :) -- but I haven't seen it lately. Install the APT package with dselect (last one you'll ever need to install :-). Add this to /etc/apt/sources.list (change the mirror if necessary): deb http://http.us.debian.org/debian unstable main contrib non-free deb http://nonus.debian.org/debian-non-US unstable/non-US main contrib non-free #deb-src http://http.us.debian.org/debian unstable main #deb-src http://nonus.debian.org/debian-non-US unstable/non-US main You can uncomment the deb-src lines once you upgrade to the potato version of APT. It lets you download, modify, and rebuild packages with very little work. But that's off the subject :-) After that, run 'apt-get update' to get a list of slink packages, and 'apt-get dist-upgrade' to grab everything. Once you've done that, and upgraded your APT, then you can uncomment the deb-src line and rerun 'apt-get update' to grab a source index too. Be sure to read man apt-get and apt-cache. APT can do lots of neat tricks :-) > > Without further suggestions, it seems that the way to do it is point dselect > to the potato directory of a nearby mirror and use that thing to upgrade all > my packages. Any caveats I should watch out for, or prepare against? > (Something about bash?) It seems a bit .. hitting the + key too often. I'm pretty sure that the bash/libreadline problem has been fixed. If you are really paranoid, download zsh or ash and run the APT install from that. > > And, wrt to the install process -- it tripped me a bit that i needed the > second bin cd first -- my fault, I just booted the CD and didn't even look > for a readme file (sitting there plain as day :) -- but it would be nice to > see documented in the setup. Also, asking questions during the install of > packages threw me off a bit. Maybe this should be brought up in > debian-devel, but... it would be nice for packages to ask their questions at > the beginning, or at the end. as for byte-compiling emacs code > (dangerous waters here...) vim doesn't have bytecode to compile. :) You can have my VIM when you pry it from my cold dead fingers :-) > > Thanks all. > > btw -- the reason why I decided to try debian was its users -- many of the > users evangelize debian, absolutely lauding it. So I had to try it. Agreed..debian is cool because of the people factor. I've been a RedHat convert since '96. > So far > it seems like the perfict fit for me. the users I have seen mention it > though didn't carry a holier-than-thou attitude with them -- users that like > their way of doing things, since many did things the other way before. You > all seem like a nice bunch of folks. If you keep it up, you will probably > keep getting more converts. :) (Heck, I liked the idea enough that *I* want > to be a debian developer -- just no clue what to develop. heh heh. :) I was a developer a couple of years ago, but stopped doing it. Right now, there's a major backlog, and the rumor is that they won't accept new ones until after potato's release in November. Check the debian-devel archives if you are interested. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: compiling help
On Fri, Aug 20, 1999 at 11:13:11PM -0400, Tom wrote: > > But the file does not use libcgic it uses libcgi (here is the file) > When I compile with the suggested: cc -o cgitest cgitest.c -lcgi > it works fine, but when I change the cc to g++ I get: > Ahhh..you could have mentioned that in the post! I assumed you were using the cgic stuff that you mentioned in the first post to your other message. Honestly, I have no idea what is wrong. Perhaps someone who is better familar with the guts of gcc can help you. It appears that if cgi.c is compiled by g++ and cgitest.c is compiled by g++, then they link. If they are both compiled by gcc, then they link. Otherwise, not a chance. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Learning ZSH
On Fri, Aug 20, 1999 at 10:20:34AM -0700, Sean 'Shaleh' Perry wrote: > > > > The part I miss most is the context-sensitive TAB-completion, like > > bash can not do this. However a coworker and fellow Debian developer uses zsh > and from what I have seen *NO* shell matches it. > > He has tab complete for apt, dpkg, telnet, and other fun things. > Whoa! sounds really cool..I could think of lots of uses for that! I've used bash for my whole 'linux life', can anyone recommend a tutorial/web page for zsh? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Small faux pas with LILO
> got a Win95 boot disk? try running 'fdisk /mbr' and rebooting... it should > replace the MBR with a Win95 compatible one. (you'll have to use a Linux > boot disk to boot your linux system and attempt to configure LILO again > though) > Correct me if I'm wrong, but the MBR is fine, the partition boot record is what's corrupted. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Small faux pas with LILO
On Sat, Aug 21, 1999 at 12:08:31PM +1000, Andrew Pollock wrote: > Hi, > > I have a PC with two hard drives, /dev/hda and /dev/hdb > > I installed Linux on /dev/hdb and inadvertantly set boot=/dev/hda1 instead > of /dev/hda in /etc/lilo.conf and now I'm having trouble booting that > partition (a Windows 95 partition). Have I done inrepairable damage to that > partition or can I recover from this boo boo? > I've made this mistake too! LILO should have backed up the boot sector of /dev/hda1 to /boot/boot.0301. So, this incantation of dd should get things back the way they were: dd if=/boot/boot.0301 of=/dev/hda1 bs=512 count=1 -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: compiling help
On Fri, Aug 20, 1999 at 09:58:35PM -0400, Tom wrote: > > Thanks to the group for pointing me in the right direction with the CGI > info, I installed the CGI package from WWW > > In the cgitest example that comes with the package, it says to compile > the program doing > > cc -o cgitest cgitest.c -lcgi 'gcc -ocgictest cgictest.c -lcgic' works fine. The library is called 'libcgic', so you link with -lcgic. Simple as sthat. Brute force way is something like this. You get 'undefined reference to cgiOut', so you do 'cd /usr/lib; grep -l 'cgiOut *.so' and voila, you know what library cgiOut is in! > > How to I convert that to g++ ? Using g++ to compile a C program is just the same as using gcc to compile a C program. Unless 'gcc -v' and 'g++ -v' say different things, then you can use gcc/g++ interchangably. cc is a standard unix name for the C compiler, and is symlinked to gcc on my system. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: fetchmail: SMTP connect to localhost failed: Connection refused
> Correct me if I'm wrong, but Exim is supposed to be run from inetd by > default (thus the 'big exit 0'). There are only certain cases where > you would want/need Exim running all the time... ok, you are right. Sorry, I've since switched to qmail and I'm used to seeing the qmail stuff running all the time. I never could get exim to work without uncommenting the 'exit 0' line, when I did use it. > > > What does "ps aux | grep 'exim'" say? > > On a system where Exim is run from inetd, it won't show up as running > in the process table unless there's a connection in progress... > You are right..I am wrong Please forgive me :-) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Installing Debian 2.1 problem
On Fri, Aug 20, 1999 at 09:00:31PM +1200, Darran Provis wrote: > hi, I'm trying to install 2.1 (i386). > I have purchased Linux on CD's but I have a sound blaster CD-rom drive so the > install won't recognise the CD drive. I know I need to make a boot disk with > the correct drivers on it but, I need to know where to find the drivers etc > and how to make the floppy. I have already made a rescue floppy using rawrite. > Thanks In order to rebuild the kernel with new drivers, you need access to a running Linux machine, but I don't think that's what you want to do. You shouldn't need to so that, IMHO. What modules have you tried to load in the install program? Hoe do you know that it hates your CD-ROM drive? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: PCL 5 and Magicfilter
On Fri, Aug 20, 1999 at 05:19:47PM -0700, Lazar Fleysher wrote: > Hi Everybody, > > I have a Winprinter :( and it can simulate PCL5 with 600dpi resolution. > The printer is connected to a win machine and I use samba to print to it. > The question is which device should I use for magic filter configuration? > Currently, I have set up the filter to use laserjet (which is, I think, > PCL2 with 300 dpi).Definitely, it is not 600dpi. > Setup magic filter for laserjet, acting as if it is a local printer. Then, add a script called /etc/magicfilter/samba-ljet3-remote with this in it: #!/bin/sh /etc/magicfilter/ljet3-filter | /usr/local/sbin/smbprint $* Make sure it is executable. Next, add the attached smbprint script to /usr/local/sbin. Make sure it is executable. It is from /usr/doc/samba-doc/examples/examples/printing. magicfilterconfig should have made a spool directory for your printer called something like '/var/spool/lpd/hplj3'. Add a .config file in that directory with lines like this: server=kennedy service=hplj password=foobar Finally, Add a printcap entry like this: lp|hplj3|HP Laserjet III:\ :lp=/dev/hplj3:sd=/var/spool/lpd/hplj3:\ :sh:pw#80:pl#66:px#1440:mx#0:\ :if=/etc/magicfilter/samba-ljet3-filter:\ :af=/var/spool/lpd/hplj3/acct:lf=/var/log/lp-errs: Restart lpd (or lprng..it seems to be better and not die randomly on me) and everything should work! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org #!/bin/sh # This script is an input filter for printcap printing on a unix machine. It # uses the smbclient program to print the file to the specified smb-based # server and service. # For example you could have a printcap entry like this # # smb:lp=/dev/null:sd=/usr/spool/smb:sh:if=/usr/local/samba/smbprint # # which would create a unix printer called "smb" that will print via this # script. You will need to create the spool directory /usr/spool/smb with # appropriate permissions and ownerships for your system. # Set these to the server and service you wish to print to # In this example I have a WfWg PC called "lapland" that has a printer # exported called "printer" with no password. # # Script further altered by [EMAIL PROTECTED] (Michael Hamilton) # so that the server, service, and password can be read from # a /usr/var/spool/lpd/PRINTNAME/.config file. # # Script further modified by Richard Sharpe to fix some things. # Get rid of the -x on the first line, and add parameters # #-t now causes translate to be used when sending files # # In order for this to work the /etc/printcap entry must include an # accounting file (af=...): # # cdcolour:\ # :cm=CD IBM Colorjet on 6th:\ # :sd=/var/spool/lpd/cdcolour:\ # :af=/var/spool/lpd/cdcolour/acct:\ # :if=/usr/local/etc/smbprint:\ # :mx=0:\ # :lp=/dev/null: # # The /usr/var/spool/lpd/PRINTNAME/.config file should contain: # server=PC_SERVER # service=PR_SHARENAME # password="password" # # E.g. # server=PAULS_PC # service=CJET_371 # password="" # # Debugging log file, change to /dev/null if you like. # logfile=/tmp/smb-print.log # logfile=/dev/null # # The last parameter to the filter is the accounting file name. # Extract the directory name from the file name. # Concat this with /.config to get the config file. # eval acct_file=\${$#} spool_dir=`dirname $acct_file` config_file=$spool_dir/.config # Should read the following variables set in the config file: # server # service # password eval `cat $config_file` # # Some debugging help, change the >> to > if you want to same space. # echo "server $server, service $service, file $acct_file" >> $logfile ( echo translate echo "print -" cat ) | /usr/bin/smbclient "$server\\$service" $password -U $server -N -P >> $logfile
Re: burn a cd from an iso file
On Fri, Aug 20, 1999 at 05:46:24PM -0700, Kenneth Scharf wrote: > If I already have an iso file, how do I burn a CD under linux? > Xcdroast seems to want to build it's own iso file in a dedicated > directory or partition. > Not necessarily. Symlink (or move) your ISO image into the directory that you tell xcdroast. Then click 'Master image' and click 'Write image'. Key in the name of the ISO file. Image Size and Image-ISO9660-Label should show something once you key in the image. Then, click 'Start Write Image.' Alternatively, check out the man page for 'cdrecord'. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Migrate them slowly
On Fri, Aug 20, 1999 at 06:53:54PM -, Pollywog wrote: > I have used Mahogany and have it installed, but it still does not work as > well as xfmail. I will try it again when the next package update takes > place. Too buggy now. > Yeah, but it looks _really_ promising. The embedded Python is a cool idea. Once it does threading as well as mutt, I'll look into it seriously. To make it easy for newbies to migrate, we replace their components one a time under Windows using a windows port of dpkg: Outlook Express -> Mahogany Word-> AbiWord Excel -> Gnumeric Command.com -> bash (or zsh) with the cygnus tools Photoshop -> Gimp IE -> Mozilla a lot of other programs exist for unix that look and function like their windows counterparts (like icq stuff, xmms, IglooFTP/gftp, et. al). and then the kicker: Explorer's UI -> LiteStep (Afterstep/WindowMaker for windows) Gradually, a little at a time, we get them used to it. Then, we replace the kernel. An experienced user helps them set it up, and they already know how to use the programs. Voila! We have a new Debian user! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: restricted shell
On Fri, Aug 20, 1999 at 12:03:52PM +0200, J Horacio MG wrote: > Hi, > > I know of the existance of a restricted shell, though I'm not sure > whether this is what I'm after. Nope, correct me if I'm wrong, but you just want a way to remotely login to your computer. > > In time, I'll be moving away to another country, and I intend to leave > my home computer with the current users configured, ie. root (of > course), my sister, and I. I'd also like to create a "guest" user with > some little (hence restricted) access. Is rsh what I should be looking > for? I'd say to use telnetd, because there are some nice telnet programs out there. But, telnet is insecure, and by default the only way to log in as root on telnet is to login as a normal use and 'su' to root. > > Also, I'll need to access the computer both as root and as normal user > remotely, via the inet, as I cannot give my sister root control (good > ...! it took me a lot to convince her to give up windozing, then > customize her account and installing some nice looking programs, etc... > sadly enough, mutt doesn't offer a flowers bouquet with each new mail, > so I had to give her XCMail ;-) ); > is, in this case, ssh what I should be installing? If you can, do ssh and grab the unix version or the freeware windows program 'Tera Term Pro'. If not, telnet is a good second choice, because it comes standard on all Windoze machines (you didn't mention what kind of computer you will be connecting from). -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: web db suggestions please
On Fri, Aug 20, 1999 at 06:35:12PM +0100, richard wrote: > Hi all. > I'm kinda having a go a developing an intranet at work and I need > some opinions from any experts out there. > We have a document register that is obviously used register any > documents written, written in vb and > accessing an Access db. Now I would like to move this to this intranet. > At the moment our server is a > 486 running debian 2.0. What I'd like to know is what would be the most > favoured choice of front and back > ends? I'm thinking cgi *should* be fairly decent for the front end but > I've never done this before so I'm no > expert. But I have absolutely no idea what db to use. I need something > small and fast initially. Suggestions? > Definately go with PHP3 and MySQL. I've had great luck with it. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: uploads over modem are SLOOOOOW!
On Fri, Aug 20, 1999 at 09:00:55AM -0500, Brian Servis wrote: > Hi all, > > dowloads. I don't set mtu or mru in the ppp config files, after the > link is up ifconfig shows that mtu is set at 1500. My latest upload >From my past experiences, that could be the culprit right there! I've had connections go from what you experience to near-perfect after I dropped the MTU to 552 (512 + 40 byte header) or even 296 (256 + 40 byte header). Check your PPP config files (can't recall where Debian puts them..haven't needed PPP in over a year) and add 'mru 552' and 'mtu 552'. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: debian installation
On Fri, Aug 20, 1999 at 09:57:06AM -0600, Julian Taylor wrote: > Debian, admittedly, is one of the most primitive Linux > installations I've ever used. Among the problems are > the fact that no matter what you tell it during > installation about your intended configuration, it > sets everything up not to work. You'll find the same > problem with ppp which is most likely set up to deny > all access and printing which is most likely not set > up at all. > I've had a completely opposite experience. A newbie friend wanted to install Linux, and he ordered a RedHat 5.2 CD from Cheapbytes, much to my chagrin. I tried to help him install it, but after 4 hours of trying to configure ppp, (me with 3+ years of experience with all sorts of distributions), we gave up. Two weeks later, I got him to try Debian and all it took was one run of 'pppconfig' and 'pon' and we were up and running. Doesn't get any easier than that! As far as your printer problems are concerned, have you tried the magic filter package in combination with lprng? The Printing-HOWTO suggests that, and it works like a charm. Even in my odd situation (printing to an HP Laserjet III attatched to a Windows machine), I got it to work without much trouble following instructions in the Samba and Magicfilter documentation. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: CGI Help
On Fri, Aug 20, 1999 at 09:49:06AM -0500, Keith G. Murphy wrote: > Stephen Pitts wrote: > > > [cut] > > My next project..developing an admin system using Python and wxWindows. > > I can develop under Linux and deliver it to my client running > > Windows without changing the code..do that with any other language > > (including the much-acclaimed JDK/Swing combo!) > > Hmmm. Couldn't you readily do this in Perl also? I'm not going to > argue the point on this list, nosirree. Just asking. I don't even know > Python. > > Anyway, if anyone wants to get in on a definitive ongoing flamewar on > Perl vs. Python, go to comp.lang.perl.misc right now. Slurs are hurled, > subjectivity parades as objectivity, everything you could want. :-) > Oh..God..I saw that one the other day while searching on Dejanews...looks really ugly! I'll stay out of that one. The only cross-platform tool that I know of that exists for Perl is Tk. >From a cursory glance at the Tk docs, it doesn't have some of the neat controls that wxWindows has (TreeView and ListView). Also, wxWindows looks like the native GUI (GTK or Windows). Finally, most of the guts for wxWindows exist in C++ and from what I understand, a lot of the Perl/Tk implementation is still in Perl and the speed difference is noticable. Then again..GTK has been ported to Windows, so maybe the bindings work on Windows and Linux? Also..Python just looks cooler, and feels write. To paraphrase from the flamewar. programming in Python is more fun (IMHO) and irregardless of what other benefits I get out of it, I write programs because it is fun. If it ceased to be fun, then none of the other rewards would be worth it. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Multiple installation question.
On Fri, Aug 20, 1999 at 12:08:15PM -0500, Miguel A. Figueroa wrote: > Hello all, > > First, Debian kicks a**!! Agreed...it just doesn't get any better than this! > Second, is there a deselect command that will give me a list of packages > that I can then use it to install an identical box using apt/deselect? I > know that RedHat has a quick installation type utility and I was > wondering if Debian has something similar. This way a base installation > can be done on the box and then tell dselect that these are the packages > I want without intervention from the user. > > Thanks > Miguel A. Figueroa At this point, all you can do is synchronize the packages on multiple systems: >From the machine that has the list of packages installed you want, do dpkg --get-selections > packagelist Then, move packagelist to the destination computer, and do dpkg --set-selections < packagelist and then '[I]nstall' from dselect, or 'apt-get dselect-upgrade' And both machines will have the same set of packages installed. Unfortunately, you still have to babysit the install process and the configure scripts. Some people are working on a system to let you do a 'kickstart' install similar to RedHat. Check the debian-admintool list for info (http://www.debian.org/Lists-Archives/). Until then, hire monkeys to do the gruntwork :-) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: NFS and file locking
On Fri, Aug 20, 1999 at 02:13:39AM -0400, William T Wilson wrote: > On Thu, 19 Aug 1999, Stephen Pitts wrote: > > > doesn't work. I think that it is trying to lock the files in some > > non-NFS compatible way. > > You have to run lockd if you want locking to work over NFS. Even so, > locking still doesn't work right (i.e. it still leaves open the > possibility of race conditions) and it only works with flock(), as far as > I know. fcntl() won't work because fcntl() style locks are supposedly > "mandatory" locks and the NFS protocol has no support for mandatory > locking. Thanks for the indepth info..thats just what I need. I'm now running kernel-space NFS server with patches from the knfs package and Wordperfect (evidently using flock()) works fine. An strace on kexpress, however, reveals that gdbm uses fcntrl! Perhaps I'll look into it and submit a patch at some point. Since you seem to be in the know, does Coda offer any better locking support? > > > Are there any workarounds to these NFS locking problems? > > Yeah, create the locks on a local filesystem :} > Yup, I'll continue to symlink. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Eureka!
On Fri, Aug 20, 1999 at 10:19:13AM -0700, Sean 'Shaleh' Perry wrote: > *DANGER* 2.2.11 has memory leak issues and other fun nits. Make sure to > upgrade any and all boxen running it to the next release as soon as it comes > out. Oh, wonderful! I upgraded to 2.2.10 and it killed all my filesystems around mid-June. forcing a reinstall. Now I'm running a ticking time bomb called 2.2.11! Should stable be redefined to a.b.c, where b is even and c >= 30? What were the > 100 beta releases in 2.1 for? I had a wonderfully stable system for > 6 months until I got the 2.2.x series to fix some DHCP problems and because of IPChains support... now I've got this to worry about! I built 2.2.11 with the NFS patches that come with the knfs package and installed new versions on the client and the server. Server is now running kernel-space daemon instead of user-space daemon. I no longer get any boot-up errors, and lockd seems to be running OK. However, kexpress/gdbm still can't generate locks. Wordperfect works OK, though. Whatever..I'll go back to symlinking ~/.kde/share/apps/kexpress to /usr/local/smpitts-kexpress and move on! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Eureka!
On Thu, Aug 19, 1999 at 07:17:37PM -0700, Sean Perry wrote: > > > > Are there any workarounds to these NFS locking problems? > > kexpress can be fixed by mailing the author and helping him fix it -- source > is > a great thing. > > WP8 I am surprised fails, we used it on Suns where everything is nfs. The problem is on my end. In deperation I upped the server to the latest potato and upgraded it to 2.2.11. When I rebooted (third time it has ever been rebooted), I saw the message: Starting NFS client services: rpc.lockdlockdsvc: Function not implemented Evidently, my NFS server CAN'T do file locking. Loading the 'lockd' moduel and restarting the service doesn't do anything either. Should I try kernel-level NFS server support? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: configure for SoundBlaster cdrom sbpcd.h ?
On Fri, Aug 20, 1999 at 12:39:46PM +1000, Alexander Jankowsky wrote: > > > As the system is booting up, it eventually gets to this... > > sbpcd-0 [01]: sbpcd.c v4.6 Eberhard Moenkeberg <[EMAIL PROTECTED]> > sbpcd-0 [02]: Scanning 0x340 (LaserMate)... > sbpcd-0 [03]: Scanning 0x230 (SoundBlaster)... > sbpcd-0 [04]: Drive 0 (ID=0): CR-563 (0.75) at 0x230 (type 1) > sbpcd-1 [05]: You should read linux/Documentation/cdrom/sbpcd > sbpcd-1 [06]: and then configure sbpcd.h for your hardware. > sbpcd-0 [07]: data buffer size: 8 frames. > > I tried adding, > sbpcd=0x230,1 and sbpcd=0x230,SoundBlaster into /etc/lilo.config Ahhh..but the stock Debian kernel ships with sbpcd as a module! /etc/lilo.conf only affects stuff compiled into the kernel. The correct lilo.conf syntax would be 'append = "sbpcd=0x230,1"', anyways, if it was builtin to the kernel. Better is adding 'options sbpcd sbpcd=0x230,1' to /etc/modutils/sbpcd, and running 'update-modules'. Then 'rmmod sbpcd; modprobe sbpcd' should work, and it will load automatically without autoprobe the next time you reboot. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Installing Netscape
On Thu, Aug 19, 1999 at 07:35:56PM -0400, Adam Krell wrote: > "/usr/lib/netscape/netscape: error in loading shared libraries > libstdc++.so.2.8: cannot open shared object file: No such file or directory" > > I checked and I do have the package libstdc++2.9 installed{HYPERLINK > "libstdc++2.9.html"}. Can anyone > point me in the right direction to solve this? You've answered your own question. You have libstdc++2.9 installed, netscape is built against an older version (2.8). Searching the Contents file (a great thing to have around...check the CD or the FTP site you got Debian from): bash-2.02$ zgrep 'libstdc++.so.2.8' Contents-i386.gz usr/lib/libstdc++.so.2.8 oldlibs/libstdc++2.8 usr/lib/libstdc++.so.2.8.0 oldlibs/libstdc++2.8 So, you need to install the libstdc++2.8 package in the oldlibs section to complement your existing libstdc++2.9. If you've got apt setup, then 'apt-get libstdc++2.8' should do it. Otherwise, go into dselect, press [S]elect Packages, '/' to search, then type 'libstdc++2.8' and press enter. Hit '+' to select it, Enter to get out of that menu, and [I]nstall to install it. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Scratchy sound during IO operations
During heavy-duty IO operations, the xmms sound becomes so scratchy and garbled that it is unbearable. This occurs when doing things like 'apt-get dist-upgrade' and while ripping cds with cdparanoia. I don't remember this happenning before. The sound card is an ISA Soundblaster 16 PnP and the motherboard is a Pentium II motherboard with two relatively new 6 GB IDE hard drives. I don't remember this happening before, but I realize that the system bus is probably being overloaded by the data going to/from the hard drives ahd the memory to adequately deliver data to the sound cards. Is there any setting I can tell something (BIOS, kernel, etc) so that the bus is shared better? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
NFS and file locking
I mount /home over NFS from another Linux machine. Most programs are fine with it, albeit mutt is a little slow reading a 1000 message debian-user in Maildir format. However, I have two programs in particular that don't seem to like this. kexpress, a KDE-based newsreader, and Wordperfect 8. WP8 complains that it can't access files in ~/.wprc because they are 'in use'. I rmed that directory and it still doesn't work. I think that it is trying to lock the files in some non-NFS compatible way. The same thing happens with kexpress. The strace shows a line like this: fcntl(4, F_SETLK, {type=F_WRLCK, whence=SEEK_SET, start=0, len=0}) = -1 ENOLCK (No locks available) and it prompty bombs out. Linking ~/.kde/share/apps/kexpress to a local directory inside of /usr/local solves the problem, so it is definately NFS related. But, that's an ugly solution. Are there any workarounds to these NFS locking problems? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: About rlogind, rshd and terminal information
On Fri, Aug 20, 1999 at 12:02:03AM +0200, Christoffer Hall-Frederiksen wrote: > Now here is where the plot thinkens: > > This is only a problem if I rsh from an HPUX to linux, not > from OSF1 or SunOS. And another thing that mystifies me is > this. If i rsh to the machine from an HPUX machine and do > a eval `resize`. Then the ternimal is OK as long as I don't > resize the xterm. But if i have done the eval `resize` and > then visit nvi, then the terminal is ok AND the information > gets updated when I resize! A god guess is, that ncurses > does a bit for initialisation of the terminal (since nvi > uses ncurses), but does anybody have a clue og what it > might be?? > > I am puzzeled, to say the least! > What is the value of $TERM in these two situations? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: CGI Help
Another cheer for Python! I ported a C++ daemon that wouldn't compile cleanly on Pair Networks' FreeBSD systems (old version of GCC and the STL) to Python in two days (my first Python program!) and: * code was 200 lines of code vs 500 of C++ and all in one file instead of 5 files of C++ * code ran _much_ faster, partially due to this feature: my_list = ["Alpha", "Beta", "Gamma"] if "Epsilon" in my_list: <-- 'in' is a godsend print "Uh-oh, something's really wrong here" (I was too lazy to use a binary search routine in C++, so I hit the database server several hundred times) * things like data = urllib.urlopen("http://www.debian.org";) replaced 20 lines of C++ code that used the GNOME-specific libghttp1 (another thing to compile/install on their system) * the regexp .{5}\s+(\d{3,4})p?\s+(?:\d{3,4}p?)?\s*\w{2}\s+(?:(?:( \d{2})-(\d{2}))|Life)\s+(?:\s+.{5}\s+(\d{3,4}) \s*(?:\d{3,4}p?)?)? saved me almost a hundred lines of text parsing. Did I mention I like Python a lot? My next project..developing an admin system using Python and wxWindows. I can develop under Linux and deliver it to my client running Windows without changing the code..do that with any other language (including the much-acclaimed JDK/Swing combo!) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Strange (?) boot-up messages
On Thu, Aug 19, 1999 at 10:40:46AM -0300, Guilherme Soares Zahn wrote: > Hi there, > > here I am, still trying to fine-tune my PC and understand it a bit > further... That attitude seems to be the minority these days, judging from the recent flamefests on Slashdot. > > Now, I've been puzzled by some messages I find in the boot sequence > and by a particular message I find in the syslog: > > When I boot the system, as soon as the klogd is started here comes the > following messages: > > Aug 18 16:45:27 groo kernel: klogd 1.3-3#31, log source = /proc/kmsg > started. > Aug 18 16:45:27 groo kernel: Cannot find map file. > Aug 18 16:45:27 groo kernel: Loaded 25 symbols from 4 modules. > > OK... I'm puzzled by the 2nd one... what does he mean by 'Cannot find > map file'? I guess he means that he can't find the /boot/map file, No, its looking for a System.map file, which lists where all the functions live when the kernel is in memory. (useful for debugging kernel panics among other things). > > but > it's there (together with a boot.0300 and a boot.0302 - btw, what are > those files anyway?) Is it somehow serious or dangerous for the system? /boot/boot.0300 is a backup of the boot sector for the master boot record of /dev/hda. /boot/boot.0302 would be a backup of the partition boot record of /dev/hda2. (ls -l /dev/hda shows that hda is major device 3, minor device 0; ls -l /dev/hda2 shows that hda2 is major device 3, minor device 2) > > A bit later, when setting up the serial ports, it states: > > Aug 18 16:45:27 groo kernel: Serial driver version 4.27 with no serial > options enabled > Aug 18 16:45:27 groo kernel: ttyS00 at 0x03f8 (irq = 4) is a 16450 > Aug 18 16:45:27 groo kernel: ttyS03 at 0x02e8 (irq = 3) is a 16450 > > OK... whay ttyS00 and not ttyS0?? At home it happens just the same way, > but there I have RedHat 5.2 installed and because of that I can't seem > to find a way to get rid of the old /dev/cua0 mouse device, because if I > link /dev/mouse to /dev/ttyS00 can't find the latter and if I link it to > /dev/ttyS0 it won't find a mouse... ;-/ What is the output of 'ls -l /dev/ttyS*'? You are right, /dev/cua? devices are obsolete now. > > And then, by the end of the boot-up procedure, I get this one (just like > that, doubled): > > Aug 18 16:45:32 groo /usr/sbin/gpm[152]: Skipping a data packet (?) > Aug 18 16:45:32 groo /usr/sbin/gpm[152]: Skipping a data packet (?) > > What's that all about? ;-? Perhaps it has something to do with your other problems? > > Last, in my syslog I will find, from time to time: > > (...) kernel: floppy0: perpendicular mode not supported by this FDC. > > What is 'perpendicular mode'? Do the massage mean that my FDC is wrongly > configured for Linux or what? ;-? Can you use your floppy drive just fine? Looking at the source in drivers/block/floppy.c, it has something to do with seeking and getting high transfer rates (128k/s) on special floppy drive controllers. > > Thanks for your attention!!! Any time -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: fetchmail: SMTP connect to localhost failed: Connection refused
On Thu, Aug 19, 1999 at 08:52:48AM -0700, root wrote: > Hello, > I'm doing a fresh slink install on a friend's old '486. > I've installed fetchmail, but keep getting the error msg: > fetchmail: SMTP connect to localhost failed: Connection refused > I edited /etc/exim.conf and added 'localhost' and my hostname to > 'local_domains=' and restarted exim, but no change. Are you sure that exim is running? The default /etc/init.d/exim has a big 'exit 0' at the very top of the script? What does "ps aux | grep 'exim'" say? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: apt with smbmount
On Thu, Aug 19, 1999 at 09:08:07AM -0700, David Karlin wrote: > Hello, > I'm installing slink on a friend's old '486, with a CD rom which is > smbmounted from a drive on his win98 box. > I can access the .deb files okay, and can do 'dpkg -i file.deb', but > 'apt-get install ' is broken because the symlinks on the CD > aren't followed. > Is there any way to get apt-get to work on an smbmounted CD rom? > I'm considering installing personal web server on the '98 box and > serving up the CD as http or ftp if smbmount won't work. Yes, but that wouldn't work either, because no microsoft filesystem supports symlinks. > I'd like > to avoid making any changes to the winbox because he uses it for work > and I don't want to be responsible for fixing anything on it that > might break. > Any ideas? Have you tried just using dpkg-multicd with dselect, instead of APT? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Default Dir in X11
On Thu, Aug 19, 1999 at 04:37:03PM -0400, Kelly Lesperance wrote: > Hi, > > Since I installed Debian, I've had one problem. The default directory for > anything (an xterm, a Save dialog, etc) is /usr/bin, not the user's home > dir). Any ideas would be greatly appreciated. How do you start X? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Fonts and Wordperfect 8
Since several of you on the list seem to be having better luck with WP8 than I am, I'd like to ask a couple of questions: 1. How do you get the included fonts on the CD-ROM to show up in Wordperfect? I've heard rumors of a 'font installer', but I can't seem to find anything and the 'Fonts' menu in the Preferences dialog is unhelpful. For what its worth, I can't change the path to the Type 1 fonts in the 'Files' section. 2. From what I can gather, X supports Type 1 fonts. Is there a utility to generate the required fonts.scale file for them (like ttmkfdir for Truetype fonts or mkfontdir for PCF fonts) ? Perhaps..if nothing else..I can use my WP fonts in Abiword or something else. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Gnome+IceWM
On Tue, Aug 17, 1999 at 09:12:27AM +0800, A. M. Varon wrote: > On Tue, 17 Aug 1999, Bek Oberin wrote: > > > I'm not sure of the proper incantations to run both IceWM (the > > gnome-compliant version) and Gnome. My instinct was this: > > > > /usr/bin/gnome-session & > > exec icewm-gnome > > > > But I end up with -two- toolbars on top of each other, which > > is rather dim. What's the proper? > > Edit /usr/X11/window-managers > The topmost content should be: > > /usr/bin/X11/icewm-gnome > Correct me if I'm wrong, but that changes the window manager for all users, and it doesn't even start the GNOME panel! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: glibc 2.1.2 and slink
On Wed, Aug 18, 1999 at 05:28:26PM +0400, Alexander Zhuckov wrote: > Hi! > > Can I safely compile and install glibc 2.1.2 (and respective dev > packages) from potato on slink? Why not just install the glibc 2.1 package from potato instead of building it yourself? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Best 2.2.x for Slink?
On Wed, Aug 18, 1999 at 08:43:25AM -0700, tjm wrote: > It's time to upgrade. Is there a recommended > version of the 2.2.x kernel to use on a slink > system for the least amount of problems, or > should I just go for the latest that I can > find? At this time, stability is more important > than leading edge. > > Get 2.2.11! I had 2.2.10 wipe out a hard drive and had funny NFS problems with 2.2.7. Because stable releases only have bug fixes, newer is better. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Can I get new, but *not* updated packages from "unstable"?
On Thu, Aug 19, 1999 at 08:50:25AM -0400, Evan Van Dyke wrote: > Joe Emenaker wrote: > Potato is based on glibc2.1. Slink is glibc2.0... so from what I've read > on many other posts here, it's a BadIdea(tm) to try and mix these due > to the different glibc dependencies. Errr..not exactly. As long as you upgrade your libc6 and a few other core packages to the potato versions, you can keep everything else as a slink version. slink packages work fine with the potato libraries, but the converse isn't true. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: knfs success?
On Mon, Aug 09, 1999 at 10:14:22AM -0700, Jinsong Zhao wrote: > I was playing with knfs. I compiled the kernel 2.2.10. Then I > installed the knfs package from the unstable. But looks like knfs > cannot export a directory. I explicitly issue commands: > > exportfs /usr1 machine1(rw) > > then it prints "Function not implemented." > > Then the basic function of NFS is not there. How useful can it be? > I'm assuming you are running the kernel that you just compiled (check with uname -a). Personally, I've always edited /etc/exports when using NFS. Here's what it looks like on one of my boxes: /home ender(rw) /mp3 ender(rw) The server is orwell..on the client, ender, I have entries like this in my fstab: orwell:/home/home nfs defaults0 0 orwell:/mp3 /mp3nfs defaults 0 0 and 'mount -a' works perfectly. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: XFree86 3.3.3.1 on Debian 2.1
On Mon, Aug 09, 1999 at 12:36:36PM -0500, [EMAIL PROTECTED] wrote: > I am attempting to install Debian 2.1 on a Compaq Presario 1255 laptop > computer with Neomagic video. I noticed a "News and Announcements" item > from May 21 on www.debian.org/~branden <http://www.debian.org/~branden> , > that indicates a XFree86 3.3.3.1 package that could be installed for Debian > 2.1. Would I download xfree86-2_3.3.3.1.orig.tar.gz or > xfree86-3_3.3.3.1.orig.tar.gz? Thank you for the help. > > > -- > Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] < /dev/null > Wow..this question is coming up once a week now! Short answer: go to ftp.netgod.net/x and get the updated packages. The two .tar.gz files that you are referring to are source packages.. they contain the human-readable source code that can be compiled into executable programs. Don't grab them unless you know that you must, because building debs will take several hours. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: SCSI Emulation
On Fri, Aug 06, 1999 at 03:13:49PM -0400, Quesada, Juan wrote: > I need help in configuring SCSI Emulation under kernel 2.2.5 > I've turned on SCSI Emulation option, I cannot see it in /proc/devices > You won't see it under /proc/devices, because the 'virtual' SCSI controller for your IDE drives is not a real device! Look at /proc/scsi/ide-scsi/0 or /proc/scsi/scsi to verify that it is enabled. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Esound and vmware
On Fri, Aug 06, 1999 at 06:35:53PM +0200, Andy Spiegl wrote: > Hi! > > I tried to start vmware with the esddsp wrapper, but vmware still > complains that it can't access /dev/dsp. > I've used ESD with VMWare fine. What are the permissions on /dev/dsp? What user is running VMWare? What user started esd? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: network audio system?
On Sat, Aug 07, 1999 at 06:55:17AM +0100, John Gay wrote: > > > I've set up an old 486DX33 as an X-Terminal and I plan to put a sound card > into > it. After a few questions and searching, it seems that nas, network audio > system > is what I need, but I can't seem to find much information about it on the web. > Does anyone know if there is a home site for this? I tried www.ncd.com, but > this > only talks about thin clients for X-Terminals, even though they are listed as > the original creators for nas. I would just like to have some more info on > this > one. Thanks for all the help and info. > > Cheers, > > John Gay > At this point, I think the more popular approach is to use esd (package esound). Some programs (xmms comes to mind) have builtin support for it, while others can be convinced to use it with 'esddsp programname'. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Disk drive: going, going, ...
On Sat, Aug 07, 1999 at 08:48:06AM -0400, Jim Foltz wrote: > Hi, > > I an going to do a re-installation of Windows and Debian and planned on > backing up the /etc directory. My /home and /usr/local are on a second > drive so will remain intact. What else on the / partition might need > to be saved? I don't have any databases or anything out of the ordinary > (that I know of, or remember) > If you want to get back up and running fast, do dpkg --get-selections > /usr/local/dpkg_packages. Then, when you reinstall, skip all of the initial dselect stuff until you do dpkg --set-selections < /usr/local/dpkg_packages. All of the packages that you had on your old system will be automagically selected in dselect for you! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: what's the difference of netscape and navigator browser
On Thu, Aug 05, 1999 at 10:04:09PM -0500, Daniel Yang wrote: > I looked around web browsers and found Netscape and Navigator. I don't what > is the difference between them. Eventually I installed Navigator, but I still > don't know if I installed the right one. > Is there a document describing web browsers and configuration? Netscape was a company that made two web browser products: Navigator and Communicator. They were bought by AOL a while back. Navigator can only view web pages, while Communicator can also do mail, news, web page editing, etc. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Video Card
On Tue, Aug 03, 1999 at 01:39:42PM -0300, Mario Olimpio de Menezes wrote: > On Tue, 3 Aug 1999, Raymond A. Ingles wrote: > > > > > Well, nVidia has submitted patches to XFree86 and Mesa, and I know > > they've been incorporated into XFree86 3.3.4. I think the latest Mesa > > > some idea when XFree86 3.3.4 will be in .deb format? > Try the X Strike Force homepage for some beta packages: http://www.debian.org/~branden/ Alternatively, do what I did with my TNT2: 1. install the Debian xserver-svga package just like normal 2. grab the archives with glx built into the X server from nVidia (not the ones with -dyn at the end) from nVidia 3. put XF86_SVGA in /usr/local/bin and edit /etc/X11/Xserver 4. Put libGL.so.* in /usr/local/lib and run ldconfig -v / Quake2 and Quake3 look wonderful, as does any/all other GL stuff! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: QT 2.00 and Slink
On Tue, Aug 03, 1999 at 12:19:22PM +0200, Tadas wrote: > Hello, > > I don't know. When I run ./configure, I see the message: > > --- > Build type:linux-g++-shared > Compile flags: -I$(QTDIR)/src/3rdparty/zlib -I$(QTDIR)/src/3rdparty/libpng > Link flags: > GIF supports: no > > Creating makefiles... > > Qt is now configured for building. Just run make. > To reconfigure, run make clean and configure. > Try using "rm config.cache; CC=g++ ./configure" and see if the build works. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: apt-get blocked by firewall what ports does it use?
On Tue, Aug 03, 1999 at 10:22:07PM +1000, Gareth wrote: > G'day all, > after some heavy network abuse our intranet admin has changed the > firewall setup and now apt-get does not work. Does anyone know what ports > apt-get it uses so I can ask him to enable them? > > First..a pet peeve...if you are posting a new thread to the list, please send a message to [EMAIL PROTECTED] Please don't reply to an existing message. To my mailer, it looks like you are adding to the "Powered by Debian Logo" thread. apt-get either uses http or ftp connections, depending on how your /etc/apt/sources.list is set up. It sounds like your admin has blocked outgoing FTP connections. If you are just using the standard Debian mirrors..most of them serve HTTP as well as FTP stuff. Here's an example from my /etc/apt/sources.list: deb http://http.us.debian.org/debian unstable main contrib non-free deb http://nonus.debian.org/debian-non-US unstable/non-US main contrib non-free -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: xemacs in color in rxvt
On Tue, Aug 03, 1999 at 12:05:03AM -0600, John Galt wrote: > > BTW any POSIX shell and termcap system will respond the same to a "echo > $TERM", HTH > Even inside of an xterm? My xterms give me 'xterm-debian' and on a virtual console, I get 'linux'. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: xemacs in color in rxvt
On Mon, Aug 02, 1999 at 09:02:26PM -0400, Arcady Genkin wrote: > Hi all: > > I use rxvt for an X terminal. How can I get xemacs show colors in it? > Midnight commander uses color by itself... > What is $TERM set to? ("echo $TERM" should show you if you use bash). Try "export TERM=xterm-debian", and then run xemacs. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: keyboard problem in X/started after restore from backup
On Mon, Aug 02, 1999 at 09:47:12PM -0400, Steve Stancliff wrote: > Hi all, > > I have been running Debian slink for about 4 months now. > Tonight I decided > it was time to take some more disk space away from Windows > and give it to > Linux, so I tarred (-cvvzf) my Linux partitions, > repartitioned the drive, and > untarred (-xvvzf) Linux. > > I had previously done this same operation in cloning my > Linux to another > computer, and had no problems. > > Tonight, after untarring, I have some problems: > > 1) the permissions for the devices in /dev apparently > weren't restored > correctly - I couldn't run wterm until I changed some of > them. I've attached 'ls -l' from my /dev directory. Use diff to compare it to yours and see if there are any differences (other than fd0, dsp, and mixer, which I've changed myself) > > 2) My keyboard isn't working right in X. I have the > backspace-as-del problem, and > also it doesn't recognize the "windows" keys at all. > I realize that there are some standard fixes for the > backspace problem, but I think > my situation is a little different. My guess is that there > is a file somewhere which the > permissions are screwed up on, which is causing the keyboard > problems. Hopefully > someone out there can suggest where to look. > xterms communicate using /dev/ttyp?, perhaps you should check those permissions. > > Finally, should I have done something different with tar in > order to make it properly > restore permissions? Seems strange, since as I said I have > done essentially the same > thing before with no problems . Taking a cursory glance over the tar man and info pages, '-p' causes tar to preserve permissions. Also, there are options to preserve the owner of the file. IMHO, you got lucky the first time. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org devls.gz Description: Binary data
Re: newbie² in trouble: installation fails
On Tue, Aug 03, 1999 at 01:38:50AM +0200, damiaan wrote: > hello there, (please forgive my poor English) Your English, especially the spelling, seems much better than a lot of Americans I know :-) If only we in the States had such a good grasp of foreign languages as you Europeans do :-) > I tried to install Debian 2.1 on a 486 with floppies.I managed to boot > from hard disk, and started doing some linux. Soon i got tired of > downloading the packages each time i wanted to try something new and > ordered the cdroms. First wanted to configure the cd-driver for my > creative labs CR-563 B in linux, it never worked out(also in dselect). dselect wasn't the right place to configure the CD-ROM. You need to load a kernel module (a device driver) for your CD-ROM. From experience with a system with that type of CD-ROM, I'm pretty sure you needed the matcd module. 'modprobe matcd' would have loaded it, and adding 'matcd' to /etc/modules would have caused the system to load it automatically on bootup. > After a while i convinced myself it would be better to reinstall (what > would allow me to repartition as well). I still had a rescue and a boot > disk. I repartitioned, and wanted to continue with a cdrom-installation. > > I couldn't configure my cdrom-driver. I tried a few things, and gave it > up. When i wanted to reboot, i found out my hard disk was empty(i > partitioned it), and i tried to reboot with my boot disk. Linux stopped > booting with an 'unable to open an initial console' message. Is this the boot disk from the previous installation? >Then i > could only try to reinstall with floppies. When i started the whole > installation once again everything seemed allright, till i came at > 'install the OS kernel and modules' where you have to insert the rescue > disk (it was already inserted). i got 'this is not the rescue > floppy...'message. tried it again and again, nothing helped. I threw the > > floppy away, took another computer, rawrited2 resc1440.bin to it, this > one didn't work either, nor did any other floppy.(all formatted 14.4 > disks that allowed me to start the installation but stopped at the same > point). > > So these are my questions: how can i get Linux to understand that i'm > using the right floppies? Try different floppies. Perhaps your floppy drive needs calibration? I've had installations where it took several times. With my dedicated Linux file server, I created a 20 MB DOS partition and used a DOS based CD-ROM driver to copy these files to a "debian" directory: base2_1.tgz drv1440.bin linux loadlin.exe install.bat resc1440.bin root.bin Then it just cded to the debian directory and ran 'install'. Much faster (and less error-prone) that loading from floppies! Once your system was up and running, you could replace the DOS partition with a swap partiton. > how can i configure my CR-563 > creative labs in the installation program? Wait for the section "Configure Device Driver Modules". Pick the matcd module. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: 2.2.x kernel and PCI eth0 problem
On Mon, Aug 02, 1999 at 09:25:14PM -0400, Rupert Brooks wrote: > Hi, > > I have configed and compiled both the 2.2.10 and 2.2.5 kernels, and my > ne2000 clone PCI ethernet card does not work with either. > > With my 2.0.34 kernel, the card worked fine. The module ne2k-pci loaded > without a hitch and dhcpcd configured the ip information flawlessly. ( > I use an ADSL connection which requires DHCP to be used to set up the IP > address.) > I have a cable modem over here and also use DHCPCD to get my IP address, etc from the provider. The old dhcpcd was doing Bad Things (TM) in regards to the new kernel networking code (i.e. setting the IP address to 0.0.0.0 IIRC). A newer dhcpcd-sv package is part of potato and it has worked fine for me. If you don't run potato, you can fairly easily grab the source and recompile it against glibc 2.0. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: XScreensaver locks up whole system.
On Tue, Aug 03, 1999 at 11:06:18AM +1000, Dan Everton wrote: > On Tue, Aug 03, 1999 at 10:22:49AM +1000, Dan Everton wrote: > > On Mon, Aug 02, 1999 at 01:21:03PM -0500, Stephen Pitts wrote: > > > Kinda offtopic suggestion, but rebuild your kernel with > > > the "magic SysReq" key option, then Alt-SysRq-K will > > > > Well, unfortunately, magic sysrq is already compiled in and did nothing. > > Tried everything from Alt-SysRq-K to Alt-SysRq-l. Nothing. Wow..it really crashed hard. > Well, after following the debugging procedure outlined above I've got > nothing. Basically, the program refuses to crash unless it's being displayed > by the xserver (xserver-svga 3.3.3.1-10). It can run just fine for ages > while the X server is not on screen, but as soon as I switch back to the > xserver, the whole systems locks (if I happen to catch the penetrate program > in it's end of level display). > > Any other hints? > Try running gdb over a telnet connection. If you want to get really into it, printout the source for penetrate and start setting breakpoints. Narrow down what is actually happening and where (in the code) it locks the system. Then, trace that back up the chain of command into the xserver (possibly rebuild X with -g..damn that would take a long time!). Because the Linux kernel is not known for causing system lockups, I would use Norton Diagnostics or something to run a full memory check. Personally, I can't reproduce your bug. I'm using kernel 2.2.7 and the custom-compiled X server provided by nVidia. Penetrate runs quite nicely, though I prefer xmatrix :-) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: gtk-config
On Mon, Aug 02, 1999 at 10:00:04PM -0400, Kevin Cramer wrote: > I wouldn't upgrade to potato. I used to run a system from unstable > and it worked fairly well but I finally got burned. > How so? I've been running unstable since last Dec and I have yet to have any serious problems. glibc2.1 broke java, and I had to rebuild liburi-perl to get netscape to install, but other than that, my system has not been severely hosed. On the contrary, I've been able to play with the latest and greatest software and had to compile from source much less stuff because Debian is extremely up to date. You see posts from people on -user at least once a week wanting to install stuff from potato, everything from xmms to kernel-package to the new X. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Unstable and xmms
On Mon, Aug 02, 1999 at 01:43:18PM -0400, Kevin Cramer wrote: > I was forced to figure it out when I tried installing the xmms package > from xmms.org and not only didn't it work, but it uninstalled my > x11amp. Since I don't have the x11amp package anymore and it is not > available at Debian's site, I was forced to with xmms. I learned that > if you aren't installing using apt-get, then you should try the > --no-act switch of dpkg before trying it! Errr...the x11amp package is not available in the "stable" section of the FTP site? > > I found that with some work you can install xmms on a stable system. > What I did might affect some packages but I don't think so. I'm not > sure it will work for everyone and it might cause damage but here's > what I did: > > 1. Get the xmms .deb package from xmms.org > 2. Get the libgtk1.2 and libglib1.2 packages from ftp://ftp.netgod.net/x. > 3. Get the libaudiofile0 and libmikmod1 packages from unstable. > 3. Install the libaudiofile0 and libmikmod1 packages with dpkg: > dpkg -i --ignore-depends=libc6 4. Install libglib1.2 and then libgtk1.2 with dkpg. > 5. Install xmms with dpkg -i --ignore-depends=libgtk1.2. IMHO, here is a much safer idea: 1. Grab the source for potato xmms, libaudiofile0, and libmikmod1 2. Grab the libgtk1.2, libgtk1.2-dev, libglib1.2, and libglib1.2-dev packages from the netgod.net site. 3. Build your own libaudiofile0 and libmikmod1 packages with "debian/rules binary" and install the resulting packages. 4. Build your own xmms package and install it. Why I like this better: * using the force options on dpkg is a _bad_ idea. The only time I've used it is to replace my mailer (going from exim to qmail). As the man page says, it could severely damage your system. * glibc is not designed to be forward compatible (i.e. glibc2.1 apps work on 2.0). Its risky pushing a critical system library to do something it wasn't intended to do. * you don't have to play with things or worry about hosing your system, you just make the packages the same way the Debian maintainers do, linking against slightly different libraries. * rebuilding a package from source is an (IMHO) good rite of passage, just like recompiling your kernel. It helps you find bugs (wmifs segfaulted on me...I rebuild the package from source with -g, ran it in the debugger, and found the problem), run potato stuff on slink, and shows you the flexibility of the Linux system. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: QT 2.00 and Slink
Is it using gcc or g++ to compile? -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Newbie question : kernel-package vs perl5
On Mon, Aug 02, 1999 at 04:28:35PM -0400, [EMAIL PROTECTED] wrote: > > > (I originally posted this message in the debian newsgroup, but > figured a lot of people would read the list and not the group, so I'm > reposting here. Sorry for all of you who read this twice.) > > Using slink, I'm trying to install kernel-package v. 6.21. My problem > is that dselect will not let me install it, saying that it depends on the > perl5 package. > > Perl *is* installed (Perl 5.004.04-7), but the package name in the list is > "perl", not "perl5". > Yet another victim of the Great Perl Upgrade (TM) :-) Grab the source to kernel-package (apt-get source kernel-package works, but I can't remember if that feature is in the slink release of APT. If not, grab the .dsc, .diff.gz, and .tar.gz from your nearest Debian mirror) Edit the debian/control file and change the perl5 depends to 'perl'. Type "debian/rules binary" and install the resulting package. Learning how to rebuild a Debian package from source can be a great asset, especially if you intend to use potato packages (where some of the core system libraries, like libc and libstdc++ have been upgraded) with slink. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: How to make a boot disk?
On Tue, Aug 03, 1999 at 12:54:34AM +0800, Liu Chung Him wrote: > Dear All, > > I have forgotten to make a boot disk during installation. > Is there any method to create it again ? > Try liloconfig -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: XScreensaver locks up whole system.
Kinda offtopic suggestion, but rebuild your kernel with the "magic SysReq" key option, then Alt-SysRq-K will kill all prorams on the current VC. Running xscreensaver in the gdb debugger will let you find out where it crashed and what might have caused the crash. Try this: 1. Use "apt-get source xscreensaver" to download the xscreensaver source. You'll also need the lesstif dev package. 2. Rebuilt xscreensaver with debugging info Adding 'export CFLAGS=-g' right above the first ./configure line in debian/rules should work. should work. 3. Startx and switch to another virtual console. Log in as the user that started the x server, and type "export DISPLAY=:1". Then start xscreensaver with "gdb xscreensaver". 4. Type "run args", where args are the command line arguments to xscreensaver. Wait until it crashes, then use the SysRq combo to kill the xserver. Go back to the gdb console and type 'where'. That will tell you where it crashed. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Three networking questions
I've got a P-150 that is running Potato and kernel 2.2.10. It acts as a IP-Masquerading router, a Samba server, an NFS server, and serves other misc roles. I've been having some problems with it lately: 1. I get the message "eth1: Something Wicked Happened! 2008" printed on the console. This also happens on another box I've got. Both have D-Link ethernet cards, and use the via-rhine module. I grepped the kernel source, and that message seems to indicate some type of status code that the driver doesn't recognize. Has anyone seen this message before and/or had it cause any adverse effects? 2. I can't open TCP connections to any host on the Internet when directly logged into the gateway box. Its uptime is around 27 days, and this has just recently began happening. Things work fine from the Debian and Win98 PCs I've got that use it as a gateway. I can ping hosts OK from the gateway, so the route information is OK. Any ideas here? 3. I use qmail as the mailer for the gateway box and the Debian box hooked up to it, and also store mail in my home directory in Maildir format. The workstation mounts its /home over NFS from the server, and folders load incredibly slowly with mutt. (15-20 seconds for around 1000 messages). Are there any performance tweaks for NFS and/or Maildir that anyone knows about? (Other than not to use them in combination :-)) Thanks for any info on the above problems. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Init takes long time and what is lspci?
On Sun, Aug 01, 1999 at 02:31:15PM +0200, Jonas Steverud wrote: > > Debian Potato. > > When I boot, init prints out "INIT 2.74" (?) and then it stops for > 10-20 seconds and then continues with "NET4: Unix domain sockets 1.0 > for Linux NET4.0." which takes another 5-10 seconds. Anyone who knows > why this happens? Is it trying to get some information from somewhere > or what? Before I changed to the 2.2.10-kernel (see other thread) this > did not happen. Or is it perfectly normal? > Unix domain sockets are a special type of file that can be read and written to, to gain all of the advantages of sockets without the overhead of TCP/IP. X and its font servers use them, as well as xmms and mysql (right off the top of my head..probably lots of others). In the Debian kernel_image_2.2.10, unix domain sockets are compiled in as a module and kmod is loading them as a module because some daemon that is starting needs them. On my system, I see that 108 things are using unix domain sockets. My smallest system here is a P-150, but I'll agree that there is a noticable boot-time delay when loading this module. In discussion right now on debian-devel is the idea of using ash instead of bash for /bin/sh. Since ash is faster/uses less memory than bash, it could significantly decrease the bootup time of your system. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: alsa modules
On Sat, Jul 31, 1999 at 10:30:05AM +0200, Lex Chive wrote: > Hello, > > I have just installed and compiled alsa. In alsa-base there is a script > /etc/init.d/alsa which preloads all the sound modules. Doesnt that ruin the > purpose of modules, which is after all to save memory when it's not needed > (well, at least part of the purpose)? Or is there some issue specific to alsa? > kmod seems to handle the autoloading well enough. > > -Lex Agreed. Someone a while back posted a "clever" script to this list that fixed all of the sound problems people were having by preloading everything. I'm with you: I have 32 modules right now in /lib/modules/2.2.7 and the only one loaded at boot time is for my ethernet card, via-rhine. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: alsa modules
Earth to Shao..is anyone out there? What you are doing is the same as what "make install" for ALSA does..preloading the modules at boot time and defeating the whole point of modules. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: scripting a potato upgrade
On Fri, Jul 30, 1999 at 03:11:33PM -0600, Cheshire wrote: > Hey all, I'm in the process of grabbing the lib/base/admin and a few > other dirs of files for a potato upgrade from slink. By nature of > dependancies, it can take a few times through dpkg -i to get everything > installed properly and I was thinking of scripting the process but I've > never really written any scripts before. I was wondering if someone > could point me in the direction of somewhere I can look to learn--at > least enough to have dpkg run through a couple directories a few times. > > |cheshire| > > > -- Why not use APT? Here's my /etc/apt/sources.list: deb http://http.us.debian.org/debian unstable main contrib non-free deb http://non-us.debian.org/debian-non-US unstable non-US/main non-US/contrib n on-US/non-free To upgrade: "apt-get update; apt-get dist-upgrade". It handles everything and installs all of the packages in the right order. Because of the glibc2.1 upgrade, I'm not sure how to do it your way. I've used APT since before slink and it did everything in the right order. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: C++ functions help
> Strange. I thought *BSD are using EGCS as well. > Nope, moving to EGCS is on the FreeBSD 4.0 feature list. Its in the ports collection for 3.2, but my ISP (understandably) probably wouldn't want to install it. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Default editor at install
On Fri, Jul 30, 1999 at 11:59:39AM -0500, Robert Rati wrote: > It's been a while since I've last installed Debian from scratch. Is ViM > the default vi clone installed? I thought it was another clone. Anyone > know? Nope, nvi is, because it is bug-for-bug compatible with VI. Also, I believe that ae is the default editor, although there was a heated debate on debian-devel a while back about the default editor. Our beloved VIM lost out because people said it wasn't intuitive and that it was too hard to learn . You can take my VIM when you pry it from my cold, dead, fingers! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: C++ functions help
On Fri, Jul 30, 1999 at 07:50:52PM +0300, Micha Feigin wrote: > Where can i find help for the c++ functions? > exept for the info page on iostream i couldn't find any info on the other > libraries or functions of c++. > C++ implementations vary greatly between platforms, and even between different compilers on the same platform. One library that seems to be present on some newer compilers is the Standard Template Library (STL). Some information is available through Yahoo: http://dir.yahoo.com/Computers_and_Internet/Programming_Languages/ C_and_C__/C__/Class_Libraries/Standard_Template_Library__STL_/ Also, C functions can be used perfectly well in C++. Check out the glibc2.1 info pages in the glibc-doc package. A word of warning: In my (albeit limited) experience, C++ and the STL has been more of a pain than a blessing. I wrote several programs that used the STL for linked lists and sorting, and found out that FreeBSD is not up to par with Linux in terms of C++ compilers and lacked some STL features I needed, and ended up rewriting the programs in Python to get them to port over cleanly. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: debian vs redhat
On Tue, Jul 27, 1999 at 02:29:34PM -0400, Bryan Scaringe wrote: > I never did get the dhcpcd client (the new one, with 2.2.x support) under > either RedHat or Debian. But dhclient is a much cleaner program. Its in the dhcpcd-sv package in potato. The init script is really clever, it won't start dhcpcd on 2.2.x kernels. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Setup a Proxy or Ipmasq
On Wed, Jul 28, 1999 at 09:31:54PM -0400, Donna Lopolito wrote: > I have two ethernet cards. One connected to the cable modem the other to > my internal network. > I want all the boxes on the internal network to be able to use netscape, > real player etc. What is the best way to go about it and how is it done? > I've got the exact same setup. The IP-Masquerade mini-HOWTO at /usr/doc/HOWTO/mini/IP-Masquerade.gz was very straightforward, and I've been happily using IP-Masquerading for about a month. Read it over, you'll need to recompile the kernel (instructions are in the HOWTO as to exactly what to compile in), and post to the list if you don't understand anything. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: C2 Certification
On Tue, Jul 27, 1999 at 05:56:21PM -0400, Jasmine Chan wrote: > > > > Hi, > My name is Jasmine Chan and I was wondering which packages of Linux is C2 > Certified. And if they are not, is there any steps taken to make Linux C2 > certified. Thanks in advance for your help. > > Jasmine Chan Why does it matter? NT is only C2 certified when it is not connected to a network :-) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Potato
On Tue, Jul 27, 1999 at 04:39:06PM -0700, Account for Debian group mail wrote: > > Anyone using the Potato release on a machine that needs to be up. How > stable is it at this point? I'd say go for it. I've been running it on both of my home machines with no problem since before the slink release. The last major fiasco was the perl upgrade, and all that entailed for me was grabbing the source for a couple of packages and rebuilding them to depend on perl5 instead of perl. Before that, there was a mysql reorganization, but I just put all of the mysql* packages on hold and waited for it to sort out. Before that, I survived the glibc2.1 upgrade with the only casualty being Java. I've never had a severe problem with potato itself, but the 2.2.10 kernel ate a filesystem on my workstation box. On the advice of this list, I'm back with 2.2.7. As far as uptimes go, my potato based server (Apache, Zope, NFS, Samba, IP-Masquerading, et. al) with 10 GB of Ethernet traffic has a 20 day uptime (since the last power outage over here). IMHO, the unstable dist is Debian's greatest strength (work of over 100 developers upgrading/improving in tandem..always bleeding edge..everything you need is in there), and the stable dist is Debian's greatest weakness (can't focus and organize well enough to produce releases more than twice/thrice a year). Also, if you look at it from an ideological POV, you can help the Debian dist out by reporting any/all bugs you find. I've found a couple, and the maintainers are very courteous about fixing them, especially on the rare occasions I included a patch. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: mail list abuse? (was Re: sendmail/fetchmail how?)
On Tue, Jul 27, 1999 at 09:36:49AM -0400, Samuel R. Scarano wrote: > On 27 Jul, Geocrawler.com wrote: > > This message was sent from Geocrawler.com by "Joakim Svensson" <[EMAIL > > PROTECTED]> > > Be sure to reply to that address. > > Hm. So this guy's not really subscribing to the list, but wants us > subscribers to help you out? I think this is considered bad form -- I > don't think the mailing list is intended to allow people to mooch help > without actually being part of it (and putting up with the volume). > Perhaps I'm just being self-righteous? What's the consensus? Totally agree with you. People do this on USENET,too. They post through Dejanews and then say "I don't have time to check this newsgroup, so please email me." The list is not your personal help desk. > On the other hand, the volume is quite overwhelming, so I can > understand the desire to just get quick help that way. I've always > thought that this list should be replaced by a newsgroup, but I guess > that's already been discussed to death. I just wonder how much more the > volume can grow before the mailing list is considered impractical Its already on my news server as muc.linux.debian-user or something like that. I know it has been archived on Dejanews as well. The primary reasons include maintaining a good S/N (signal to noise) ratio. Quite honestly, with fetchmail in the background running every 10 minutes, maildrop/qmail filtering the list mail, and mutt, my trustly threaded mail reader, it feels like a newsgroup and looks like a newsgroup, so it must BE a newsgroup, right? :-) > BTW (off-topic): since I'm adding to the volume anyway, I thought I'd > just thank those who responded to my video card question. I'm going > with the Riva TNT, just so you know. And that Linux hardware link > (http://www.math.sunysb.edu/~comech/tools/CheapBox.html) was *very* > informative -- highly recommended. Cool! I'm the happy owner of a Riva TNT2. Have you checked out the GLX stuff that NVidia released? Running Q3Test in a window with decent performance was Very Impressive! -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Using apt with downloaded deb
On Mon, Jul 26, 1999 at 12:05:08PM -0500, Brad wrote: > Just out of curiousity, what's up with the apostrophies in this message? > This has been discussed extensively in Slashdot. Evidently, Microsoft email/web creation programs don't "Do the Right Thing (TM)" in regards to apostrophes, and fail to follow the ASCII standard character codes. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Still can't build kernel_image on slink
On Tue, Jul 27, 1999 at 11:30:45AM +1000, Tadeusz Bak wrote: > > Hi all, > But my attempt to build a new kernel_image still results in the above > error. Is there something wrong with the kernel-package 6.05 or, more > probably, my system is completely disorganized? > Thank you for your attention :-). Try "make-kpkg clean". It should clear out some old cruft in the configuration files. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: ssh client
On Tue, Jul 27, 1999 at 09:32:46AM +0800, Chad A. Adlawan wrote: > hello all, > this is not a debian specific question, but does anyone know of an ssh > client for Win9x/NT that i can use to connect to my ssh ONLY enabled debian > server ? > i need this as i dont have a linux box at home. > TIA, > Chad > I've used SecureCRT, but it is pay-for-ware. Currently, I use TeraTerm PRO, a freeware product with support for SSH through a plugin. I had to go to 3-4 sites to get the various encryption components (damn ITAR), and I don't have the relevant URLs handy. I could send you a ZIP file via email, though. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: gnome-apt for slink?
On Sun, Jul 25, 1999 at 02:05:49PM -0700, Cliff W. Draper wrote: > Does anyone have a .deb for gnome-apt that runs in slink (ie, glibc2.0)? > > thanks, > -Cliff > [EMAIL PROTECTED] > Get the source for a later version of APT and build it yourself. (dpkg-source -x ; cd apt-0.3.11; debian/rules binary). Install this APT, then add a deb-src line to your /etc/apt/sources.list and do a "apt-get -b source gnome-apt". You will need the -dev files for the gnome packages that APT depends on. Rebuilding a package from source is a good experience. With 2500 packages in Debian, you may come across a scenario that requires you to do this because of a package bug. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Using apt with downloaded deb
Copy all of the .debs to /var/cache/apt/archives/. Don't make any subdirectories there, just shove everything in that directory. Then, configure dselect to use the APT method (install the apt package from your archives directory manually with "dpkg -i " if needed). Configure APT to download directly from the mirror that you got your packages from. If a package is already in /var/cache/apt/archives, it won't be downloaded again, but APT will get any packages you don't have that it needs. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: irc.openprojects.net
On Mon, Jul 26, 1999 at 08:50:20AM -0700, Thomas wrote: > Just do dpkg -s ps It's dpkg -S ps, but that doesn't help if the package with ps isn't installed :-). -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Mutt dependency on an MTA
On Mon, Jul 26, 1999 at 08:37:21AM -0400, esoR ocsirF wrote: > I am new to mutt butt I am under the imression that mutt will handle pop > queries directly. If I am wrong tell me, and I will go back to my hole. > Any way the following got my attention; Yes, it can, but piping it through fetchmail and the MTA lets you set up mail filters, and in general gives you a lot more flexibility. > Are the dependencies set correctly for mutt? I mean if mutt can fetch > its own mail from a pop server then shouldn't this be a recommends and > not a depends? It still needs an MTA to send mail :-) -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Change the signatures for -user and -devel!
On Mon, Jul 26, 1999 at 09:27:04AM -0600, David Karlin wrote: > This is would be somewhat akin to the "Unsubscribe..." line which > appeared on mail from the list for some time. (Posting instructions > on how to unsubscribe on such a page might decrease the number of > "unsubscribe" msgs which appear on the list.) > Heh..this has been brought up several times. The consensus was (IIRC), that the signatures for -user and -devel need to be switched, so debian-devel contains the newbie-unfriendly sig that we have right now, and debian-user gets the traditional "mail [EMAIL PROTECTED]" to unsubscribe. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: System woes..unkillable processes and more!
OK, great. Partition Magic hosed my partition table. I'm going to send the drive back to Seagate, 1 MB of bad sectors or not. This incident with this Linux installation is totally opposite all of my other linux experiences. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: Network addressing
> So, my question is: can I simply set the netmask to 255.0.0.0 on both of > them and put appropriate entries into the /etc/hosts file? Is this likely to > break anything? Alternatively, can I use arp to specify the Ethernet address > - IP address correspondence, or something will get confused? Try setting the netmask that way, and see what happens > > And a semi-related question: on a pure Win98 machine, how do you find out > the Ethernet address of the NIC card? I tried looking through the registry, > but didn't find anything that looked right. Start..Run...winipcfg. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org
Re: 3Com 509B Ethernet card problem part 2
On Mon, Jul 26, 1999 at 04:39:21PM +0200, Fredrik Jonsson wrote: > Hi, > > Thanks to all who responded to my first (very incomplete) question! > > I have now disabled PnP on the card and set IRQ to 10 and io to 300. IRQ > 10 is set to ISA (not Pnp) in Bios. Donald Beckers el3-diag report the > card to be OK, se below. > > What I was trying to do was to compile the 3c509 driver hard into the > kernel with "modconf" and was geting only errors. Many suggested to load > the driver as a module insteed so I have put the line "alias eth0 3c509" > in to the file "conf.modules" via update-modules command. But I still get: > > On Boot > eth0 unknown interface > > When I check modprobe > > # This file was generated by: modprobe -c (2.1.121) > [Deletia] > # Aliases > alias eth0 3c509 > alias eth0 off > [Deletia] > > The first line about 3c509 looks like mine but where does the "off" > command come from? Check around /etc/modutils. update-modules looks at all of the files there and combines them into conf.modules. A problem I've had in the past is that your editor will make backup files ending in ~ that inadvertantly get added into conf.modules. > After running insmod "3c509", modprobe "3c509", ifconfig ... and route > ... everything looks fine, se below, but I can't get a connection/ping > other computers on my local network. The settings in ifconfig and route > disappears after a reboot, I presume this is not right. > Au contrair. On boot-time, all ethernet devices are configured by /etc/init.d/network. Change the ifconfig lines there to automatically setup your network card. -- Stephen Pitts [EMAIL PROTECTED] webmaster - http://www.mschess.org