Re: [Cooker] kernel-2.6-test9 config
Andrey Borzenkov [EMAIL PROTECTED] writes: i've uploaded your work. hotplug is maintained by fredl. good he has been silent wrt initscirpts as well. I hope he's not following fpons ... Heu ? I see everything (provided you are talking about me ;-)) Francois.
Re: [Cooker] Re: [CHRPM] perl-URPM-0.94-8mdk
Olivier Thauvin [EMAIL PROTECTED] writes: Found why: @idlist = @{$options{idlist}} 0 ? @{$options{idlist}} : # Here the bad line ($options{start} || 0 .. $options{end} || $#{$urpm-{depslist}}); @idlist or return; I will try to fix this evening if fpons don't. The problem is @{$options{idlist}} which is only valid if idlist is a reference to list (even empty)... It's my error, I should have seen this bad code ;-) A new perl-URPM is uploaded. Francois.
Re: [Cooker] urpmi download file twice
Pascal Terjan [EMAIL PROTECTED] writes: I guess it tries on the first one, fails, tries on the second one, only tells it failed on the last one. Yes, this is exactly the case, in fact, you can check it works by using it on updated mirror and see only one download is done. Francois.
Re: [Cooker] Re: No more fpons: keeping the Drakes happy
Leon Brooks [EMAIL PROTECTED] writes: Is there anything that either Mandrake or we could do to make your job noticeably more survivable? Thanks Leon for your though... Well, I'am not actually completely dead :) There are still some parts of me trying to continue to live :) I'am not working on the dark side, should I have to say it :) Francois.
Re: No more fpons [was: Re: [Cooker] more urpmi issues]
FACORAT Fabrice [EMAIL PROTECTED] writes: bad news ... Oui :-( Good luck François in your new job. Thanks a lot :) Et comme on dit en fraçais : merde ! ;) fran... Thanks a lot again :) Francois.
Re: No more fpons [was: Re: [Cooker] more urpmi issues]
Luca Olivetti [EMAIL PROTECTED] writes: very sad news. I hope the best for him in his new job. Thanks again... IMHO urpmi is a fundamental piece of, so I hope that a (good) decision is taken quickly. Or we'll really have to switch to apt-rpm :-/ Hum... I have some modifications to do on urpmi, so don't switch now, I may need some tester for it ;-) Francois.
Re: [Cooker] Re: [CHRPM] urpmi-4.4-42mdk
Brad Felmey [EMAIL PROTECTED] writes: Eh? What happened to Francois? Nevermind, saw on the list. :( Yes, I left the company... Francois.
Re: [Cooker] 9.1 - 9.2 issues with urpmi
Luca Olivetti [EMAIL PROTECTED] writes: ... Let's go for a summary of the problems: * The urpmi urpmi step worked well. After that I had to stop the urpmi --auto-select because it was scrolling a bazillion of error messages regarding /var/lib/rpm/PubKeys. Had to remove the file and manually rpm --import the keys Problably a problem of specific extension of rpm 4.2 versus 4.0.4 used in 9.1, a missing rpm --rebuilddb could be the cause ? Anyway, a urpmi.update -a could be much nicer after, except if urpmi database is already up-to-date ? * There was some conflict with files with some updated packages. Removed the offending packages and urpmi --auto-select worked well. A commend line --skip pkg1,pkg2,...,pkgN can be used in such case, it is depending on the humor of the moment. * It couldn't update libavifile0.7 libaviplayavcodec (and dependent packages) due to dependencies. I think this is a problem caused by urpmi batching, since when I manually issued an urpmi libavifile0.7 libaviplayavcodec it could upgrade both flawlessly. Maybe this problem will be more noticeable on the other machine since it has many more packages installed. If you could have saved bug report of urpmi for that... in order to see what is the cause. ... gettext-0.11.5-2mdk, kdeutils-3.1-9mdk, screem-0.6.0-1mdk in 9.2: gettext-0.11.5-7mdk, kdeutils-3.1.3-20mdk, screem-0.6.2-4mdk Add to the list automake, but that's because of the naming (I had automake1.6 while 9.2 has automake1.7). Again, it is domage you didn't issue an --bug of urpmi, it is always nice when something seems to be bad. * The last issue is that by rebooting (I didn't upgrade the kernel) I had a dead machine: it couldn't find the initrd so it couldn't mount the root directory (since it's xfs). Hopefully one of the older kernels I had could boot, ran lilo from it and the machine went back to life. I don't understand what happened here, really. The other machine has a much more convoluted lilo setup, so I fear it won't be as easy to recover. lilo was problably updated, and lilo upgrade should call lilo by itself. Well, I'm impressed by how well the procedure worked. Sure, there were some hurdles and the modification in libraries versions wasn't earth shattering (I wonder if everything would work as well with a major glibc update) but still it's much better that what was available before. Oh, btw, I was using the machine (logged in in X) during the whole process. That's incredible. Kudos to mandrake developers and contributors. Yes, but there was still some problems. Thanks, François.
Re: [Cooker] 9.1 - 9.2 issues with urpmi
Luca Olivetti [EMAIL PROTECTED] writes: François Pons wrote: Again, it is domage you didn't issue an --bug of urpmi, it is always nice when something seems to be bad. Since I have another machine to upgrade I will use the option. I suppose urpmi will work normally and just save a log of what it did am I right? Should I use it for every invocation or just for the final urpmi --auto-select that tells me that everything is installed? (also, murphy's law (or heisenberg principle) dictates that when I'll use the --bug option there will be no problem to report ;-) In fact, when you see something bad in the resolution, you cancel the selection and try again. By default, urpmi stop on error (provided you disable transaction split). François.
Re: [Cooker] 9.2 - disaster already?
Rolf Pedersen [EMAIL PROTECTED] writes: # urpmi --bug bug djbsupport installing /contrib/i586/./djbsupport-1.0-5mdk.noarch.rpm Preparing... ## 1:djbsupport ## added medium qmail urpmi database locked error: %post(djbsupport-1.0-5mdk) scriptlet failed, exit status 7 djbsupport is bad, it should not try to modify a database being currently used, if this is *really* needed, a perl script should be used using directly the urpm library and avoiding to explicitely to lock the urpmi db (so that no error will occur), this script is simple but may depend upon newer interface of urpmi (so it need a require on major version of urpmi to be safe). François.
Re: [Cooker] urpmi problem
Per Lindström [EMAIL PROTECTED] writes: When I run urpmi --auto-select --auto --keep urpmi tells me that the rpm packages have faulty signatures. When I run urpmi /var/cache/urpmi/rpms/*.rpm the packages are installed without any problems. The faulty signature is in fact a missing signature (check error message), the problem can occur to check a package when it is supposed to be in an organized area (a medium) or a wild free package around ? It could be better to always check signature package, but just created package are problably always faulty ? François.
Re: [Cooker] OT - RPM question...
Scott Chevalley [EMAIL PROTECTED] writes: I've been updating my box using urpmi --auto-select for at least the past two or three release cycles, constantly running cooker. I'm wondering if there is anyway to use urpmi or rpm to give me a list of files that exist on my system that don't belong to any installed packages, so I can clean things up a little bit. I'm sure it would be much easier to just install from scratch, I just don't feel like tracking down all the contrib and plf packages I've installed over the years again. This could be a good tools to have, allowing unknown files to be seen, msec is already doing this though. It could be nice to rebuild broken rpm database (I suppose this is what you want here ?). François.
Re: [Cooker] urpmi.update broken?
Greg Meyer [EMAIL PROTECTED] writes: On Tuesday 14 October 2003 05:45 am, Stefan van der Eijk wrote: Has anybody else noticed that urpmi.update may be broken? when I run urpmi.update, it doesn't pick up changes. Removing the media and adding it again lets it pick up the changes. # urpmq libltdl3 -r libltdl3-1.4.3-7mdk # urpmi.update -a # urpmq libltdl3 -r libltdl3-1.4.3-7mdk # urpmi.removemedia cooker # urpmi.addmedia cooker file://mirrors/cooker/i586/Mandrake/RPMS with ../base/hdlist.cz # urpmq libltdl3 -r libltdl3-1.4.3-8mdk I cannot confirm ATM, but I have noticed that behavior too. I have had to reload several sources to get them working properly over the past couple of weeks. That not a good news, can you confirm this behaviour, I didn't have any problem like that currently, please report urpmi.update output if you manage to reproduce it including urpmi.cfg file. François.
Re: [Cooker] urpmi.update broken?
Stefan van der Eijk [EMAIL PROTECTED] writes: In order to get urpmi to update my sources I have to almost constantly use urpmi.update -a --no-md5sum. Can you see if this is the problem? looks like it... Ok, I don't know with MD5SUM file are not synced so ... Warly, do you have an idea ? François.
Re: [Cooker] [Bug 6124] [urpmi] urpmi pam destroyed a system!
Charles A Edwards [EMAIL PROTECTED] writes: On Fri, 10 Oct 2003 17:06:25 +0200 Jan Ciger [EMAIL PROTECTED] wrote: 3) the behavior of RPM in this case is strange (it is an RPM not urpmi issue). It is not strange. A looked at the posted syslog provides the answer. At the time the upgrade was done there existed a flux in cooker tree which required that pam-0.77-9mdk be removed. When pam-0.77-9mdk is removed it causes system-auth to to be changed to rpmsave. When then pam-0.77-10mdk is installed it will create system-auth.rpmnew, believing that system-auth still exist. The stated occurrence in an aberration and could only occur under very limited and specialized circumstances. Within the past hr I used --auto-select to install 198mb of updates and experienced no problems in the pam upgrade. But there was a problem with libtool, as libtool-1.4.3-8mdk was present in the hdlist but neither libltdl3-1.4.3-8mdk nor libltdl3-devel-1.4.3-8mdk were. Thanks Charles ! It will help me tracking down this bug of rpm (and not urpmi for sure), there are various way to fix it, workaround in pam itself (make sure a system-auth file is present), workaround in urpmi (in the same way problably), real fix in rpm (which is bar far the best solution but problably the hardest :-(). There is finally a last try, if somebody can try it, a bad invocation of urpmi on the package, this should be available using -v flag which will display what urpmi is putting in transaction that is causing rpm to behave strangely, I will try it monday if nobody can test it. But I suspect nothing is bad on this side, at least I hope for... Francois.
Re: [Cooker] [Mandrake 10] Ideas for RpmDrake [long]
Buchan Milne [EMAIL PROTECTED] writes: /me does 'urpmq -i name' ... Now, if urpmq could display a screenshot ;-) Such screenshot should be available in the package header at least, but it could be a bit too large, anyway an icon is maybe available to visualize instead of screenshot, that's not the best but you will see a picture furthermore problably nice but small :-) François.
Re: [Cooker] [Mandrake 10] Ideas for RpmDrake [long]
Eric Fernandez [EMAIL PROTECTED] writes: I agree with a lot of your comments, and the screenshot is very nice. Since softwares are more and more fragmented into smaller rpms (which is a good thing though), a meta-packaging should be introduced indeed in the next release. That would ease installation, even for urpmi users. But we already talked about this, I remember. For urpmi, it was already an idea to indroduce such behaviour, I think it is time now to do it. François.
Re: [Cooker] Clean install and XFree86
Eskild Hustvedt [EMAIL PROTECTED] writes: I just did a clean install of Cooker using netinstall floppies 7 Oct. It did not want to install X and did not allow me to continue from the Summary screen during the installation to the Updates step because it compained that it could not find SVGA X server. I had to do a reset. When I got the machine booted up and was trying to install X using urpmi (urpmi xfree86-server-4) I got the following message: --- Some package requested cannot be installed: XFree86-server-4.3-23mdk.i586 (due to missing XFree86-xfs-4.3-23mdk.i586) XFree86-xfs-4.3-23mdk.i586 (due to unsatisfied XFree86-libs[== 4.3-23mdk]) chkfontpath-1.9.10-1mdk.i586 (due to missing XFree86-xfs-4.3-23mdk.i586) (Y/n) --- So It didn't install anything. Today (8 oct) I had to download XFree86-server-4.3-23mdk.i586 and XFree86-xfs-4.3-23mdk.i586 and manually install them using: rpm -Uvh --nodeps [package]. I seem to have X running fine now though. Having strict requires on same version and release of XFree86-libs is to be sure it will works, but it can works on some cases. You have been lucky, that's all. You may have got conflicting files or even X11 applications that do not want to start due to linking problems... François.
Re: [Cooker] jpilot plugins not recognized
Bruce F Press [EMAIL PROTECTED] writes: libjpilot_plugin0-0.99.6-1mdk jpilot-0.99.6-1mdk The jpilot plugins installed in /usr/lib/jpilot/plugins are not recognized by jpilot since they end in .0 Problem can be corrected by linking ex. ln -s libkeyring.so.0 libkeyring.so Then jpilot correctly detects the plugin. Install the devel package as a workaround :-( François.
Re: [Cooker] more good news for Mandrake
Adam Williamson [EMAIL PROTECTED] writes: Bizarre...I never got that message. Okay, I'm off to post to the let's stab sympa several thousand times in the back thread...:D I saw it last week, my (perl) filters are working correctly, talk about me or urpmi and I will see it more easily... François.
Re: [Cooker] urpmi and 9.2 updates curious
Vincent Danen [EMAIL PROTECTED] writes: I didn't ever have any errors... but I've been using the root directory as the pointer to hdlist for three years. =) Anyways, I tried 9.1 and point genhdlist to updates/9.1/RPMS instead of updates/9.1 and it seems to work fine. Perfect, so now updates are working again ! François.
Re: [Cooker] urpmi and 9.2 updates curious
Vincent Danen [EMAIL PROTECTED] writes: Updates always worked, except for 9.2. 9.2 was the one that changed so that how it used to be done no longer worked. I don't understand very well, but anyway, it is still perfect that way... Francois.
Re: [Cooker] urpmi and 9.2 updates curious
Vincent Danen [EMAIL PROTECTED] writes: Vincent, how do you fixed it ? For 8.2-9.1 I do: genhdlist /mnt/BIG2/updates/[dist] For 9.2 I have to do: genhdlist /mnt/BIG2/updates/9.2/RPMS So something in 9.2 changed. And not in the installer. Ok, I see the point, I used genhdlist (which I guess has not be modified for months, but I used the one of 9.2) in both case and it produced the good files whatever the directory parameters is given... I'm using the same genhdlist (the host OS is corp 2.1, so 9.0-based). If I need to use a 9.2 genhdlist, you need to tell me this. =) genhdlist of 9.0 is exactly the same as genhdlist of 9.2 (same md5sum) but the underlying libraries have been modified, it could be the cause ? Why don't you make genhdlist /mnt/BIG2/updates/9.1/RPMS ? it should have been done that way no ? because by doing it the way you are doing them now, list file are wrong. If you have an error message, let me known what it is. François.
Re: [Cooker] urpmi and 9.2 updates curious
Vincent Danen [EMAIL PROTECTED] writes: On Thu Oct 02, 2003 at 12:05:16PM +0200, François Pons wrote: Why don't you make genhdlist /mnt/BIG2/updates/9.1/RPMS ? it should have been done that way no ? because by doing it the way you are doing them now, list file are wrong. If you have an error message, let me known what it is. I'll do some tests this afternoon and see if it breaks anything (obviously we can't ask people to remove and re-add their updates media due to this). If it doesn't break anything, then so much the better. If it does well, in my build scripts I had to differentiate 9.2 from the other distribs and I can continue to do that. Not a big deal. But if they all work the same way, it'll be easier. Please, tell me what error message you have with various distrib and what are the organizations of the files, it should be the same as 9.2 in fact, I don't understand why other distro need different organizations (I wonder how it could works). François.
Re: [Cooker] urpmi and 9.2 updates curious
Vincent Danen [EMAIL PROTECTED] writes: Have you tried adding the media and downloading the updates? It should work regardless of what's in the list file. My bad. You're right, it doesn't work. It's fixed now. The way urpmi handles this changed, so now all the old distribs do it one way and 9.2+ do it another way. No, urpmi didn't change on this, but DrakX has changed the way the update media are defined, I check to see if I made something wrong on this... François.
Re: [Cooker] urpmi and 9.2 updates curious
[EMAIL PROTECTED] (François Pons) writes: Vincent Danen [EMAIL PROTECTED] writes: Have you tried adding the media and downloading the updates? It should work regardless of what's in the list file. My bad. You're right, it doesn't work. It's fixed now. The way urpmi handles this changed, so now all the old distribs do it one way and 9.2+ do it another way. No, urpmi didn't change on this, but DrakX has changed the way the update media are defined, I check to see if I made something wrong on this... I don't known how it worked before, the list file has an additional RPMS/ string inside... Vincent, how do you fixed it ? François.
Re: [Cooker] urpmi and 9.2 updates curious (and WINE)
Leon Brooks [EMAIL PROTECTED] writes: No, urpmi didn't change on this, but DrakX has changed the way the update media are defined, I check to see if I made something wrong on this... I didn't use DrakX, if that makes any difference. Yes, it looks like list file are different now... François.
Re: [Cooker] urpmi and 9.2 updates curious
[EMAIL PROTECTED] (François Pons) writes: No, urpmi didn't change on this, but DrakX has changed the way the update media are defined, I check to see if I made something wrong on this... I don't known how it worked before, the list file has an additional RPMS/ string inside... Ok, I see the point, the list file of 9.2 does not contains any RPMS/ strings whereas list file of 9.1 and older does contains it. Why ? Update media should be added relative to the RPMS directory (so the extra RPMS/ in the list file is incorrect), the installer should do this that way, other tools or help says the same. François.
Re: [Cooker] URPMI thoughts
Greg Meyer [EMAIL PROTECTED] writes: Awhile back I had tried to setup urpmi sources with multiple mirrors for each source, thinking that if one mirror was busy, urpmi would move to the next. In this way, I wouldn't have to worry about not being able to login and try again, urpmi would simply move to the next preferred mirror. When I tried this it didn't seem to work and I have not tried again. Is this something that should be possible, or if not, can it be added as a feature? It seems like a useful feature. It should have worked in fact, after having problem to contact a mirror (timeout) it should try another one provided another one is defined, but urpmi will try this for each file, it is not a true dynamic mirror change as it doesn't take care of error code returned by network layer. Can you try to reproduce it and send me a urpmi bug report using --bug dir please ? François.
Re: [Cooker] urpmi.update weirdness
Udo Rader [EMAIL PROTECTED] writes: hi, Hi, as many others I have a local cooker mirror here is updated frequently via troels.rsync. For a couple of days (weeks?) now I find urpmi.update reading all headers of all files in my cooker repository over and over again no matter if they have updated or not, resulting in quite long processing times. So urpmi.update is building the hdlist itself as you do not rsync it ? Not too much of a problem, but I thought previously uprmi.update would only process the changed (timestamp?) files? Or am I missing something here? Strange, yes it should be that way, except if you use --clean somewhere ? François.
Re: [Cooker] urpmi OpenOffice.org weirdness
Leon Brooks [EMAIL PROTECTED] writes: On Mon, 15 Sep 2003 17:04, François Pons wrote: Without a urpmi bug report, I can't do nothing on this topic, please send me a urpmi bug report. Hint: '--bug directory' Yes ;-) I forget to add this... Francois.
Re: [Cooker] urpmi OpenOffice.org weirdness
Levi Ramsey [EMAIL PROTECTED] writes: # urpmi OpenOffice.org One of the following packages is needed: 1- OpenOffice.org-l10n-es-1.1-0.rc4.1mdk.i586 2- OpenOffice.org-l10n-ar-1.1-0.rc4.1mdk.i586 3- OpenOffice.org-l10n-ca-1.1-0.rc4.1mdk.i586 4- OpenOffice.org-l10n-cs-1.1-0.rc4.1mdk.i586 5- OpenOffice.org-l10n-da-1.1-0.rc4.1mdk.i586 6- OpenOffice.org-l10n-de-1.1-0.rc4.1mdk.i586 7- OpenOffice.org-l10n-el-1.1-0.rc4.1mdk.i586 8- OpenOffice.org-l10n-fi-1.1-0.rc4.1mdk.i586 9- OpenOffice.org-l10n-fr-1.1-0.rc4.1mdk.i586 10- OpenOffice.org-l10n-it-1.1-0.rc4.1mdk.i586 11- OpenOffice.org-l10n-ja-1.1-0.rc4.1mdk.i586 12- OpenOffice.org-l10n-ko-1.1-0.rc4.1mdk.i586 13- OpenOffice.org-l10n-nl-1.1-0.rc4.1mdk.i586 14- OpenOffice.org-l10n-pl-1.1-0.rc4.1mdk.i586 15- OpenOffice.org-l10n-pt-1.1-0.rc4.1mdk.i586 16- OpenOffice.org-l10n-pt_BR-1.1-0.rc4.1mdk.i586 17- OpenOffice.org-l10n-ru-1.1-0.rc4.1mdk.i586 18- OpenOffice.org-l10n-sk-1.1-0.rc4.1mdk.i586 19- OpenOffice.org-l10n-sv-1.1-0.rc4.1mdk.i586 20- OpenOffice.org-l10n-tr-1.1-0.rc4.1mdk.i586 21- OpenOffice.org-l10n-zh_CN-1.1-0.rc4.1mdk.i586 22- OpenOffice.org-l10n-zh_TW-1.1-0.rc4.1mdk.i586 What is your choice? (1-22) 9 One of the following packages is needed: 1- myspell-fr_BE-1.0.2-0.20030619.1mdk.noarch 2- myspell-fr_FR-1.0.2-0.20020608.1mdk.noarch What is your choice? (1-2) 2 The following packages have to be removed for others to be upgraded: OpenOffice.org-l10n-en-1.1-0.rc4.1mdk.i586 (due to unsatisfied OpenOffice.org == 1.1) (y/N) Without a urpmi bug report, I can't do nothing on this topic, please send me a urpmi bug report. François.
Re: [Cooker] Re: Missing file in perl-Net_SSLeay?
Buchan Milne [EMAIL PROTECTED] writes: I will try and get a minimal example that fails ... Do you managed to fix it or found an example that failed ? Regards, François.
Re: [Cooker] urpmi / apt4rpm - not for 9.2
Marcel Pol [EMAIL PROTECTED] writes: I quote a reply from Pixel here, from February 24 2002 (Yes, it was discussed before :-) ) quote there was a moment where we had 2 solutions, both time costly: - dump urpmi, and switch to apt-get - enhance urpmi cons for switching to apt-get: - apt-get for rpm still needed some work It should not be the case anymore, at the epoch apt-get was coredumping often. - apt-get for rpm doesn't like file requires (eg: Requires: /usr/bin/perl) I don't know if this is still the case ? - apt-get doesn't handle choices (but rpmdrake doesn't either) No, rpmdrake and urpmi now handles choices, and relatively correctly to propose (problably) the best package by default (locales dependent), but it can be enhanced of course. - we don't know apt-get (which is much bigger than urpmi, currently 11Kloc vs 4Kloc) Still the case ;-) - francois doesn't like C++ anymore :) (did I say I don't either ? oops, sorry! [1]) We both doesn't like to put core files anywhere... - urpmi can be more easily tuned for the Mandrake distribution This is still the case, of course, but strangely urpmi has almost no tuning for Mandrake, and can work for other distro (it has been done for Solaris using rpm). You may now add the following features (I may forget a lot of them) : - https, rsync and ssh protocol supported. - parallel installation (still need works but it can do things now). - urpmi core used by installation (DrakX), urpmi (of course), rpmdrake (was not the case previously), so the code of all these tools is managed by Mandrake team. - NIH syndrom ;-) François.
[Cooker] Re: Missing file in perl-Net_SSLeay?
Buchan Milne [EMAIL PROTECTED] writes: I haven't finished the schema-checking script I wanted to complete for openldap-2.0.x-openldap-2.1.x migration, so I looked at it today, and remembered why not: Hum, after examining current perl-Net_SSLeay, I found that such code should work (current version is 1.25 though... do you still use 1.13 or not ?). François.
[Cooker] Re: Missing file in perl-Net_SSLeay?
Buchan Milne [EMAIL PROTECTED] writes: $ rpm -q perl-Net_SSLeay perl-Net_SSLeay-1.25-1mdk (which seems current) I will try and get a minimal example that fails ... I said that because randomize was *effectively* in the code (and not in *.al files). Regards, François.
Re: [Cooker] Enlarge your bash completion experience
Götz Waschk [EMAIL PROTECTED] writes: Am Mittwoch, 10. September 2003, 02:42:35 Uhr MET, schrieb Guillaume Rousse: The first one provides those mighty improvments for urpmi (and friends) completion: - ligthning-fast package completion using latest urpmi pre-generated package Impressive, it's so fast. It would be nice if this would be included in 9.2. Guillaume, you have done a good job so ;-) François.
Re: [Cooker] Enlarge your bash completion experience
Götz Waschk [EMAIL PROTECTED] writes: Am Mittwoch, 10. September 2003, 10:54:25 Uhr MET, schrieb François Pons: Impressive, it's so fast. It would be nice if this would be included in 9.2. Guillaume, you have done a good job so ;-) François, I'd like to thank you too for the urpmi improvements that made this possible. Yes, but thanks Guillaume first, he told me this improvement for ages and I only have done them recently... François.
Re: [Cooker] urpmi exits with a bad array ref error
Olivier Blin [EMAIL PROTECTED] writes: Hi, Hi Olivier, When upgrading from 9.1 to cooker (after having done urpmi urpmi), I get the following error if I try to urpmi a package : # urpmi libarts1-devel --bug bad_array_ref Can't use string (libm.so) as an ARRAY ref while strict refs in use at /usr/lib/perl5/vendor_perl/5.8.1/urpm.pm line 3208. This is a great improvement that I managed to find this problem which already occured but without bug report. Now it is fixed ;-) Thanks Olivier. François.
Re: [Cooker] Enlarge your bash completion experience
Guillaume Rousse [EMAIL PROTECTED] writes: Ainsi parlait Pascal Terjan : Guillaume Rousse wrote: COMPREPLY=( $( compgen -W '-a -p -P -y -s -q -v -h --help \ - --update --media --excludemedia --sortmedia \ - --synthesis --auto --auto-select --fuzzy --src \ - --install-src --clean --noclean --force \ - --allow-nodeps --allow-force --parallel --wget --curl \ - --limit-rate --proxy --proxy-user --bug --env \ - --X --best-output --verify-rpm --test --excludepath' \ - -- $cur ) ) + --update --media --excludemedia --sortmedia --synthesis \ + --auto --auto-select --no-uninstall --keep --split-level \ + --split-length --fuzzy --src --install-src --clean \ + --noclean --force --allow-nodeps --allow-force --parallel \ + --wget --curl --limit-rate --proxy --proxy-user --bug \ + --env --X --best-output --verify-rpm --test --excludepath \ + --excludedocs ' -- $cur ) ) please add --no-verify-rpm which is quite handy for contribs :) Fill a bugreport against urpmi, it isn't in --help output :-) Guillaume, I saw it, you have been caught againt my mail filter... It is too late to add --no-verify-rpm (furthermore it is already referenced in --help). François.
Re: [Cooker] Problem with urpmi, skip.list and auto-select
Charles A Edwards [EMAIL PROTECTED] writes: rpm -q urpmi perl-URPM urpmi-4.4-33mdk perl-URPM-0.94-4mdk ... Notice the the version and release for the mplayer rpms Are the same as I have installed. What is going on? A bug report will help me to know what happens and will help me to play with. François.
Re: mach 64 problem [was: Re: [Cooker] MDK 9.1 bugs]
Kovács Géza [EMAIL PROTECTED] writes: so your card is correctly detected. your problem is not hardware recognition but configuration i guess. what is your exact problem? When installing MDK, it correctly detects an Ati Utah Mach64. But then, further down the list, there is also a Rage LT Pro entity in the list. You are right, they ought to work both. So I thought, that if it autodetects the card, marks it Mach64 utah and there is a more meaningful ATI Rage LT Pro entry in the list, why not make the detection result match exactly with the card detected? Most users can identify their card by (marketing) name than the included chip version. Mach64-s can differ quite a lot... The description of the card should help on this, we have not one card for each description, maybe adding description more visible would help. François.
Re: [Cooker] odd urpmi problem
Adam Williamson [EMAIL PROTECTED] writes: [EMAIL PROTECTED] radiohead]# urpmi.update plf examining synthesis file [/var/lib/urpmi/synthesis.hdlist.cooker.cz] examining synthesis file [/var/lib/urpmi/synthesis.hdlist.contrib.cz] examining MD5SUM file examining synthesis file [/var/lib/urpmi/synthesis.hdlist.plf.cz] [EMAIL PROTECTED] radiohead]# It could be normal, urpmi.update now check with md5sum of file downloaded and file present. In this case, it check MD5SUM file (distant) with saved md5sum of the file to download to check if they are identical. To force download, use -f --no-md5sum as -f is needed to avoid timestamps and size checking. François.
Re: [Cooker] odd urpmi problem
Adam Williamson [EMAIL PROTECTED] writes: Okay, just tried with --no-md5sum and it decided plf was current, so sounds like that's it. However, it seems the md5sum stage is responsible for the slowdown - the whole thing is a LOT faster with -a --no-md5sum than with just -a. Any idea why it takes so long? Check a /var/lib/urpmi/MD5SUM file is present ? and contains something ? Maybe urpmi compute each time md5sum ? which is not so normal... François.
Re: [Cooker] urpmi: please remove option --update
Guillaume Rousse [EMAIL PROTECTED] writes: A media is marked update only if the user declares it using --update when adding this media through urpmi.addmedia. Or is it some way for a media to declare itself as update ? Using rpmdrake, there is a check box to declare media as update (if my memory has no real problem). You can add update flag by hand in urpmi.cfg (documented in man page, now ;-)) François.
Re: [Cooker] urpmi --auto-select asks too many questions
Liam Quin [EMAIL PROTECTED] writes: I'd like it to prompt yes/always/No/reject all I can't do that for 9.2, but I can make yes act as always and no act as reject all. It could be fixed more cleanly after 9.2 François.
Re: [Cooker] urpmi --auto-select just freezes
Kim Schulz [EMAIL PROTECTED] writes: hi After upgrading urpmi and perl-URPM (why isn't there a version requirement between these two?) yesterday I now cant do urpmi --auto-select It just stands still at the prompt until I ^C out of it (I had it stading like this for 5 hours). I can see in the process list that it runs and that it in the beginning uses up alot of cputime. anyone that can confirm this? Send me bug report using urpmi --bug without actually doing an --auto-select, this will help me tesing it and fixes it. François.
Re: [Cooker] what upgrade method is supported and _tested_ ?
Tibor Pittich [EMAIL PROTECTED] writes: before releasing mdk 9.2 i have one principial question: what method of upgrade are precise tested? what method is fully supported and works well? (urpmi, cd upgrade, handy rpm -Uvh, other?) is cd upgrade method fully tested with qa team? is this method fully tested and supported with 9.1, 9.0, 8.2? especially focus on 8.2, because this version ends support from mandrake at 30.sep, then many users can make _painless_ upgrade from this version to 9.2. btw. i can't find any related documentation about upgrade procedure at mandrakelinux.com web site (maybe my fault..) I will try to make an urpmi update for a lot of distribution, including current urpmi and current perl-URPM, compiled for their respective targets (at least 9.0 and 9.1, I will try to add 8.2). Eventually, security updates will be done to add the support so that urpmi could more easily used to upgrade the system (normally I will try to add them in unsupported first). Anyway, upgrade using installer (booting the installer in upgrade mode to be precise) is the way to upgrade the distribution. Using urpmi should be another method : * upgrade urpmi if needed (updates or direct update of target urpmi (complicated)). * urpmi.addmedia --distrib removable://mnt/cdrom will import gpg keys and 9.2 media to be used. * urpmi --auto-select (will need a lot of space in /var unless network media are used), kernel may need to be installed (depending if a require on some of the package needs explicitely a newer kernel, highly probable with older version). * reboot (still better not to use old copy of almost all libraries and current libraries). François.
Re: [Cooker] urpmi.uppdate pb? holds solution.
guran [EMAIL PROTECTED] writes: OBSERVE that this shit is written in my home direcrory, = working directory, and not to /var/cashe/urpmi Which version of urpmi did you used ? I made systematic chdir into the cache for download (to avoid writing in current directory for download program used (especially wget)) since urpmi-4.4-27mdk, but it appears you are more up-to-date according to kernel used (unless you upgraded only the kernel ?). I managed to get back an error using sunet style mirror, exit code 8 of curl meaning server has sent data curl didn't understood. François.
Re: [Cooker] urpmi.uppdate pb? holds solution.
guran [EMAIL PROTECTED] writes: I am using the same urpmi now, and the problem has disappeared, might be because I have sent mails to both uninett and sunet. [EMAIL PROTECTED] guran]$ rpm -qa urpmi* urpmi-4.4-29mdk (this was reported in my first mail on this thread, http://archives.mandrakelinux.com/cooker/2003-09/msg00721.php You may want to configure one medium with hdlist and severall medium with synthesis so that if one mirror is failing, the other are continuing working (or at least may be continuing). François.
Re: [Cooker] urpmi --auto-select just freezes
s [EMAIL PROTECTED] writes: I can confirm that urpmi seems broken. mine does nothing but sit there when calling up urpmi.update or urpmi. As far as bug report, it doesn't say much/anything, if I'm looking in the right place. in /var/log/urpmi.log all I get is: Thu Sep 4 09:05:53 2003 urpmi called with --bug Thu Sep 4 09:07:49 2003 urpmi called with --bug --auto-select Please type the following 2 lines in a root shell window : urpmi --bug bugreport tar cvf - bugreport | bzip2 -9 bugreport.tar.bz2 and then please send me the file bugreport.tar.bz2 directly to me. François.
Re: [Cooker] what upgrade method is supported and _tested_ ?
Thierry Vignaud [EMAIL PROTECTED] writes: [EMAIL PROTECTED] (François Pons) writes: I will try to make an urpmi update for a lot of distribution, including current urpmi and current perl-URPM, compiled for their respective targets (at least 9.0 and 9.1, I will try to add 8.2). we still have to fix a few issues. eg: rpm.8 man page was in man-pages-fr in 9.1 but will be in rpm package in 9.2. I don't see the point with urpmi ? François.
Re: [Cooker] what upgrade method is supported and _tested_ ?
Thierry Vignaud [EMAIL PROTECTED] writes: you'll have troubles because of such bugs (this very one was catched yesterday by qa team on update because since there was no tag informing perl-URPM that man-pages-fr had to be ordered before latest rpm, update failled). i've planned to fix this one but there might be more (eg: the doc has moved some files between mandrake_doc{,-drakxtools}, ...) You means mandrake_doc and man-pages-fr does not upgrade between 9.1 and 9.2 ? This need to be fixed so. Just add conflicts in one of the package, this should have been done ? I already reported this to maintainers. François.
Re: [Cooker] Re: [CHRPM] urpmi-4.4-26mdk
Luca Berra [EMAIL PROTECTED] writes: On Tue, Sep 02, 2003 at 03:52:29PM -0700, Ben Reser wrote: Ugh... -z may be a very bad idea. If you have a fast connection to the mirror (e.g. a local mirror, and I would presume those using rsync have local mirrors since not many mirrors provide rsync access and those that do usually severally limit the number of clients) then compression will usually make things slower. besides it is not nice on the CPU of the mirrors, sunet.se i.e. asks people not to set -z MOTD: We *don't* suggest use of '-z' option, as most of the files here are MOTD: already compressed, and running compression is CPU intensive thing MOTD: anyway. Just *wait*, and the transfer will eventually complete. Ok, I should add an option for using -z so, as some users experienced increased download speed. François.
Re: [Cooker] urpmi.uppdate pb?
guran [EMAIL PROTECTED] writes: The examined line indicates to me that urpmi has left out the 'direction' to the file it has controlled and found OK, so the update can't continue. In the working directory I found the following:(taken from mc) Can you try another mirror ? François.
Re: [Cooker] urpmi: please remove option --update
Guillaume Rousse [EMAIL PROTECTED] writes: --update was introduced many times ago to allow a primitive way of excluding some media when using urpmi. It is primitive in the sense that you have to declare those media at configuration time, when using urpmi.addmedia. It is confusing because many people think of it as an equivalent of --auto-select, or as selecting only the official update section. Currently we have the --media option that allows to do exactly the same without those disadvantages. So why not just drop --update ? Currently , all stable media are marked as update (to avoid unresolved dependencies). Removing --update will break compatibilty (urpmi is still (almost) compatible with earlier version). François.
Re: [Cooker] urpmi problem
Bellegarde Cedric [EMAIL PROTECTED] writes: When i run an urpmi --auto-select, urpmi don't tell me anyting if there is noting to upgrade. fixed in urpmi-4.4-29mdk. François.
Re: [Cooker] urpmi keep ignoring some sources - right after adding them
Kim Schulz [EMAIL PROTECTED] writes: # urpmi.update -a unable to access hdlist file of main1, medium ignored unable to access hdlist file of main3, medium ignored unable to access hdlist file of contrib1, medium ignored unable to access hdlist file of contrib3, medium ignored Can you send me your urpmi.cfg file ? François.
Re: [Cooker] [REQUEST] option to tell urpmi skip signature check
FACORAT Fabrice [EMAIL PROTECTED] writes: is there a way to tell urpmi or rpm to skip signature check ? when installing package from main, no pb, but when coming to contrib you have to say yes each time. This is time consuming You have --no-verify-rpm for that. Note I may add a All along with Yes/No maybe ? Or use Yes for all signature checking ? François.
Re: [Cooker] weird problem with missing packages
Kim Schulz [EMAIL PROTECTED] writes: Can anyone tell me how to fix this? Change of mirror ? François.
Re: [Cooker] Problems using MakeCD to build ISO still persists
Warly [EMAIL PROTECTED] writes: Hum, install uploads and perl-URPM versions are becoming too fast for me to follow. You kidding ? It looks like You need vacations dear Warly ;-) François.
Re: [Cooker] Problem with conditional build
Buchan Milne [EMAIL PROTECTED] writes: it should be written: %define build_foo 0 %{?_with_foo: %{expand: %%global build_foo 1}} Dump the %expand, it's not necessary: %{?_with_foo: %global build_foo 1} FYI, I used the following to make it work in adiusbadsl : %{expand:%%define build_modules %(if [[ `/bin/uname -r|sed 's/-.*$//'` 2.4.21 ]]; then echo 1; else echo 0; fi)} #- how to use the following: rpm --without modules -ba SPECS/... %{?_with_modules: %{expand: %%define build_modules 1}} %{?_without_modules: %{expand: %%define build_modules 0}} But I'am not expert, this could be simplfied using %global ? François.
Re: [Cooker] Re: Strange urpmi perl error
David Walser [EMAIL PROTECTED] writes: fpons, maybe urpmi could do a bug log everytime it's used, but just overwrite it everytime (so it doesn't fill your disk). This could be a great idea... Even a rotation of bug report directly in /var/cache/urpmi/bug ? François.
Re: [Cooker] urpmi wierdness
Brian Tyndall [EMAIL PROTECTED] writes: Well after the problem occurred, check rpm -q --whatprovides libexpat.so.0 just to see ? I can't really do that... slbd is running, installing packages into the chroot, rebuilding the src.rpm and then de-installin the installed packages. Well, in fact this problem was (unless it still occurs) problably related to a bug in rpm which has been fixed since, now we have changed rpm behaviour to be more 4.0 compliant with provides/obsoletes (in fact now obsoleting a virtual provides does no more obsoletes package containing it because this caused too many problems with us). Perhaps try using a bare chroot environment and then installing kdelibs-devel into it, as it is quite a large install. Ok, I try to check if it has effectively been removed. Best Regards, François. PS: is --root functionality planned for urpme ? Re PS : It is already added, since end of last year, but it is still not added in --help output, I fix this.
[Cooker] Re: Problems in drakxtools regarding ltmodem
Buchan Milne [EMAIL PROTECTED] writes: I would like to see ltmodem support out-the-box for boxed sets, and have working packages, that with drakconnect (on beta2) do most things right (drakconnect will install them when a ltmodem is found if they are accessible by urpmi). The problem is still that drakconnect for some reason insists on making a link for /dev/modem (at present it asks you which of 8 serial ports to use, or if /usr/lib/libDrakx/network/modem.pm is patched, whatever device is listed there for ltmodem, in 9.1 this was ttyS14), and this breaks auto-loading of the lt_serial module (which works via 'alias /dev/modem lt_serial' in modules.conf, and registering the symlink via devfs). At the epoch it has been made necessary to do that else it will not work. But now, if it works out of the box correctly (it could be nice if it works even if devfs is not used too) we can remove this. Best Regards, François.
Re: [Cooker] weird curl/urpmi problem
Brian Tyndall [EMAIL PROTECTED] writes: retrieving source hdlist (or synthesis) of noo1... curl: option - is unknown curl: try 'curl --help' or 'curl --manual' for more information This was a problem in urpmi using curl with proxy, it should be fixed since latest urpmi version, can you upgrade them first ? François.
Re: [Cooker] Re: Strange urpmi perl error
FACORAT Fabrice [EMAIL PROTECTED] writes: Le lun 25/08/2003 à 12:40, François Pons a écrit : David Walser [EMAIL PROTECTED] writes: fpons, maybe urpmi could do a bug log everytime it's used, but just overwrite it everytime (so it doesn't fill your disk). This could be a great idea... Even a rotation of bug report directly in /var/cache/urpmi/bug ? better use standard tools /var/log/urpmi/bug and configure logrotate I don't know if logrotate can perform urpmi bug rotation ? François.
Re: [Cooker] Re: Strange urpmi perl error
FACORAT Fabrice [EMAIL PROTECTED] writes: it can perform rotate of whatever text file u give it and thus depending on criteria ( age, size, ... ) But urpmi does not produce any text file for bug report, only directory of archive, unless they gets uuencoded but I do not see any interrest here ;-) François.
Re: [Cooker] urpmi(.update) problem ?
Fabien ILLIDE [EMAIL PROTECTED] writes: In case of : [EMAIL PROTECTED] fabien]# ll /var/cache/urpmi/partial/ -rw-r--r--1 root root 356352 aoû 22 14:31 hdlist.cz -rw-r--r--1 root root 98 aoû 20 11:01 MD5SUM For sure an hdlist is not 356352 bytes longs, you have got an error to read hdlist which has been wrongly reported as list being unable to be built. Try again updating and check it still works, do you have used symlink in /var/cache/urpmi or /var/lib/urpmi to perform some tasks ? François.
Re: [Cooker] Strange urpmi perl error
Charles A Edwards [EMAIL PROTECTED] writes: Running urpmi --auto-select this morning gives this error: [EMAIL PROTECTED] charles]# urpmi --auto-select Can't use string (libkaboodlepart.so) as an ARRAY ref while strict refs in u se at /usr/lib/perl5/vendor_perl/5.8.1/urpm.pm line 3073. If kdemultimedia is removed --auto-select performs without error Can you install back kdemultimedia and perform a simple urpmi --bug autoselectwrong without the --auto-select so that I can reproduce it ? Thanks, François.
Re: [Cooker] Invalid Key ID
Levi Ramsey [EMAIL PROTECTED] writes: Never mind... manually upgrading to urpmi-4.4-22mdk got me the keys added, though there are a couple of problems: 1) PLF key (caba22ae) is assigned for my remote Cooker source (which I use as a backup in the event that I need a package I normally exclude from my local mirror). 2) local mirror has no key assigned, but attempts to install some packages (not all, vim-common (but not any other vim packages) and postfix did not result in this behavior) failed with signature issues. This should now be fixed in urpmi-4.4-23mdk. François.
Re: [Cooker] something wrong in URPMI gpg keys management
Guillaume Rousse [EMAIL PROTECTED] writes: For the second time, URPMI inserted another key id in my PLF section configuration. The first time it was MDK key, i corrected it manually, now it is JPackage key: plf file://mnt/nfs/bakounine/ftp/pub/plf/cooker { ... key-ids: c431416d } Looks like urpmi is gathering keys all around the globe to fed up plf media ;-) Ok, I look at it of course. François.
Re: [Cooker] Today's urpmi woes
Frank Griffin [EMAIL PROTECTED] writes: Autoconf 2.13-18 isn't being recognized as 2.13: The following packages have to be removed for others to be upgraded: kdevelop-2.1.5-10mdk.i586 (due to unsatisfied autoconf == 2.13) libkdevelop2-2.1.5-10mdk.i586 (due to unsatisfied kdevelop == 3:2.1.5-10mdk) (y/N) but... [EMAIL PROTECTED] ftg]$ cd /data/cooker/i586/Mandrake/RPMS [EMAIL PROTECTED] RPMS]$ ls autoconf* autoconf-2.13-18mdk.noarch.rpm autoconf2.5-2.57-5mdk.noarch.rpm and [EMAIL PROTECTED] man]# rpm -q -f /usr/bin/autoconf autoconf-2.13-17mdk Apparently I have -17 installed, but -18 isn't seen as satisfying KDevelop's requirement. Normal, the epoch is missing in requires of kdevelop, try with command line directly using rpm -Uvh --test, you should get the same result (or you can blame me a lot). François.
Re: [Cooker] problems with initscript package
Todd Lyons [EMAIL PROTECTED] writes: Kim Schulz wanted us to know: the package comes when I do urpmi --auto-select on my cooker box (using only mirrors of the cooker packages as source) You have a 9.1 update source defined it appears. You verified that there is not one in urpmi.conf? It could happens if you use old media and want to install specific packages on these media (not by using --auto-select normally). Can you try to make a bug report using --bug urpmibug option of urpmi and send me the compressed tar of the directory urpmibug please ? François.
Re: [Cooker] Invalid GPG signatures on Mandrake RPMs
Guillaume Rousse [EMAIL PROTECTED] writes: Check your urpmi configuration /etc/urpmi/urpmi.cfg, it often confuses keys. Arg... François.
Re: [Cooker] Re: rpmdrake and proxy
David Walser [EMAIL PROTECTED] writes: --- François_Pons [EMAIL PROTECTED] wrote: David Walser [EMAIL PROTECTED] writes: fpons - any comment on this? David Walser wrote: --- David Walser [EMAIL PROTECTED] wrote: --- Guillaume Cottenceau [EMAIL PROTECTED] wrote: David Walser [EMAIL PROTECTED] writes: I'm writing, because if the location [rpmdrake is] downloading from is the same as the proxy, it should not use the proxy, as that sometimes plain doesn't work. that's curl's job I suppose.. No. Plus you'd have to do it upstream in wget too. Anyway, Unix tools in general let you do stupid things if you really want, including downloading from a host through a proxy on that host, even if that doesn't work. With rpmdrake though, it's kind of doing a stupid thing for you w/out you explicitly asking it to, as it's picking up that http_proxy variable. rpmdrake should check for this. Ok, we're both wrong. It's not curl or rpmdrake's job to check for this, it's urpmi's. Imagine this scenario, you've got a proxy configured, which is also the location of one of your urpmi sources, but you have sources from other servers too. You run a urpmi command to install some packages, some which will come from each place. If you don't do http_proxy= first, the ones from that server won't work, but if you do, the ones from the other servers won't use the proxy. So if urpmi is installing something from the same host as the proxy, it shouldn't use the proxy for that package. Proxy by media is the best... yes I agree. François. Well just given the current situation, global proxy (either configured in urpmi or used from the environment variables), shouldn't it just be fixed for now that if proxy != host don't use the proxy? Maybe it could be fixed for that, provided the proxy host match the url used. François.
Re: [Cooker] Re: rpmdrake and proxy
David Walser [EMAIL PROTECTED] writes: fpons - any comment on this? David Walser wrote: --- David Walser [EMAIL PROTECTED] wrote: --- Guillaume Cottenceau [EMAIL PROTECTED] wrote: David Walser [EMAIL PROTECTED] writes: I'm writing, because if the location [rpmdrake is] downloading from is the same as the proxy, it should not use the proxy, as that sometimes plain doesn't work. that's curl's job I suppose.. No. Plus you'd have to do it upstream in wget too. Anyway, Unix tools in general let you do stupid things if you really want, including downloading from a host through a proxy on that host, even if that doesn't work. With rpmdrake though, it's kind of doing a stupid thing for you w/out you explicitly asking it to, as it's picking up that http_proxy variable. rpmdrake should check for this. Ok, we're both wrong. It's not curl or rpmdrake's job to check for this, it's urpmi's. Imagine this scenario, you've got a proxy configured, which is also the location of one of your urpmi sources, but you have sources from other servers too. You run a urpmi command to install some packages, some which will come from each place. If you don't do http_proxy= first, the ones from that server won't work, but if you do, the ones from the other servers won't use the proxy. So if urpmi is installing something from the same host as the proxy, it shouldn't use the proxy for that package. Proxy by media is the best... yes I agree. François.
Re: [Cooker] Re: hdlist.cz asnd synthesis
David Walser [EMAIL PROTECTED] writes: Why aren't the ones in /base symlinks to the ones in the actual RPMS directories? This is not possible when making CD. François.
Re: [Cooker] Re: [CHRPM] urpmi-4.4-21mdk
Olivier Blin [EMAIL PROTECTED] writes: - added -a flag for urpmq (so that urpmq -a -y -r will do what is requested more or less). Thanks a lot Francois. Can this be made the default behaviour ? Or a global urpmi option ? (named for example urpmq-select-all) I may make it the default if -y is given ? Any comments ? François.
Re: [Cooker] Urpmi.update funky
w9ya [EMAIL PROTECTED] writes: This is self-explanatory. [EMAIL PROTECTED] bfinch]# urpmi.update -f the entry to update is missing (one of Installation CD 1 (x86) (cdrom1), Installation CD 2 (x86) (cdrom2), International CD (x86) (cdrom3), update_source, plf, main, contrib) I do not see any problem here. François.
Re: [Cooker] more urpmi weirdness
Olivier Blin [EMAIL PROTECTED] writes: This is a known problem (bug 4637) I will fix today. This wasn't reported when I wrote my previous e-mail ;-) Should we fill bug reports for such urpmi problems or is it enough to report them on ML ? Maybe, but I was on vacations for the last 3 days, so I read everything at once. Thanks for the fix. Normal. François.
Re: [Cooker] Begging, pleading, for the love of God can we get mod_perl rebuilt?
Brad Felmey [EMAIL PROTECTED] writes: Please? Pretty Please? Pretty please with sugar on top? done. Fraçois.
Re: [Cooker] more urpmi weirdness
Olivier Blin [EMAIL PROTECTED] writes: urpme tries to remove 37 MB but doesn't list the packages that will be removed (not even msec). ... François, do you need the --bug output ? This is a known problem (bug 4637) I will fix today. François.
Re: [Cooker] urpmi --auto-select - how to get around it?
Frank Griffin [EMAIL PROTECTED] writes: Kim Schulz wrote: just tried this and it wanted to install around 300 rpm's but ended up with this: Installation failed: arts = 3001:1.1.3-1mdk is needed by libarts-devel-1.1.3-1mdk arts = 3001:1.1.3-1mdk is needed by libarts-1.1.3-1mdk libxml2.so is needed by (installed) libcroco0-devel-0.1.0-2mdk kdelibs = 3000:3.1.2-20mdk is needed by (installed) kdelibs-devel-3.1.2-20mdk libstdc++.so is needed by (installed) octave-2.1.50-1mdk kdegraphics = 1:3.1.2-6mdk is needed by (installed) kdegraphics-devel-3.1.2-6mdk libgcc_s.so is needed by (installed) octave-2.1.50-1mdk libreadline.so.4.2 is needed by (installed) ghc-5.00.2-2mdk libm.so is needed by (installed) libcroco0-devel-0.1.0-2mdk libm.so is needed by (installed) libgtop1-devel-1.0.13-5mdk libm.so is needed by (installed) octave-2.1.50-1mdk libglib-1.2.so is needed by (installed) libgtop1-devel-1.0.13-5mdk libz.so is needed by (installed) libcroco0-devel-0.1.0-2mdk perl-base = 5.8.0 is needed by (installed) gnome-db-0.2.96-6mdk perl-base = 2:5.8.0 is needed by (installed) gaim-0.64-3mdk perl-base = 2:5.8.0 is needed by (installed) gaim-0.64-3mdk Try installation without checking dependencies? (y/N) what now I'm not sure what you're upgrading from. I did it from base 9.1, and I'd say that it involved at least 300 packages. And, like I said, you get some scary prompts. I don't recall the components you list being in them, but the feeling I got from the list was that the packages were OK, it was just the dependencies that were incorrect. So I just answered YES to everything. You may first try urpmi urpmi in order to avoid bug inside urpmi or rpm, so that it could be much better. If the problem persist, try using --keep (as urpmi has been updated). François.
Re: [Cooker] rpmdrake vs. urpmi weirdness
Frank Griffin [EMAIL PROTECTED] writes: Charles A Edwards wrote: Had you not used --allow-force urpmi would have given you the same error as did rpmdrake. That's not what I would expect. From the man page: --allow-force Allow urpmi to ask user to continue installation using no depen- dencies checking or forced installation due to error. By default urpmi exit immediately in such case. Previously, I've had it display the errors and then prompt me as to whether I want to continue. I have to change the man page, now --allow-force authorize the resolution dependencies algorithm to ignore unsatisfied dependencies in order to avoid removing packages, this is a hack of course. Why it has been added to this option ? Because --allow-force allo package to be installed with broken dependencies, so we continue on the same road... Francois.
Re: [Cooker] Re: [CHRPM] perl-URPM-0.93-1mdk
Levi Ramsey [EMAIL PROTECTED] writes: On Wed Aug 06 20:00 +0200, François Pons wrote: * Wed Aug 06 2003 Fran?ois Pons [EMAIL PROTECTED] 0.93-1mdk - added URPM::Signature for handling armored gpg file and internal rpm pubkey. Does this mean that urpmi won't prompt on PLF packages? It could, urpmi will now search for armored file (if available) and use it to import automatically the key in rpmdb and add automatically too the id in urpmi.cfg for the given media, this means no other package (without the given key) will be allowed, of course this still need some testing but the core code is here. If it does, then I worship the ground on which you walk... It could be so. Francois.
Re: [Cooker] Status Check
w9ya [EMAIL PROTECTED] writes: Hello gang; 1 - For me, I have been getting these results for a few days ro so, and with some packages well over a week. I feel uncomfortable defeating the deps and/or manually uninstalling such a large list of packages. So, I thought I would at least report these pic-a-dillos. 2 - This update script is old, and recently (the past week or so) it has required several invocations to get all the packages that can be upgraded to upgrade. Is there a better script someone can suggest using ? Thanks in advance for your time. You have a better error report now in urpmi, look at man page, there are 4 or 5 exit code just for installation failure report (missing files and/or failed transaction). You may use these error code to exit correctly from your script. You may try the following option too --keep which avoid updating in order not to remove existing package (only upgrade are allowed, this avoid using --no-uninstall). Another method is to use urpmi --auto-select directly as there is now an integrated split mechanism to avoid such a loop you use, furthermore the split is done in a way the package are installed in right order to minimize transaction size (which is not the case when using an urpmq --auto-select output). Best Regards, François.
Re: [Cooker] rpmdrake vs. urpmi weirdness
Frank Griffin [EMAIL PROTECTED] writes: I just tried to install the MandrakeClub Commercial FlashPlayer package on today's cooker with rpmdrake, and got the error: FlashPlayer-6.0-3mdk.i586 (due to unsatisfied rpmlib(PayloadFilesHavePrefix) = 4.0-1 This is strange, such error should not happens in fact because there are simply ignored (normally). I then went to command line and tried: urpmi --no-verify-rpm --allow-force FlashPlayer and it installed cleanly without a single complaint. Beware that adding --allow-force change way the resolution is done (especially for unsatisfied dependencies...). So try again without this flag. Francois.
Re: [Cooker] urpmi --auto-select - how to get around it?
Kim Schulz [EMAIL PROTECTED] writes: Im trying to get my cooker to update (it's been almost a month due to missing internet con.), but I get this when I do an auto select: # urpmi --auto-select One of the following packages is needed: 1- ksymoops-2.4.8-2mdk.i586 2- kernel-utils-0.1-1mdk.i586 What is your choice? (1-2) 2 Some package requested cannot be installed: glibc_lsb-2.3.1-1mdk.i586 gnome-db-0.2.96-8mdk.i586 (due to unsatisfied perl-base[== 2:5.8.0]) kdegraphics-3.1-3tex.i586 (trying to promote libkpovmodelerpart.so by selection of URPM::Package=SCALAR(0x8470080)) kdelibs-3.1-7tex.i586 (trying to promote libmad.so by selection of URPM::Package=SCALAR(0x8509954)) kdelibs-devel-3.1-7tex.i586 (trying to promote libqt3-devel by selection of URPM::Package=SCALAR(0x85056fc)) latex2html-2002-4mdk.noarch (due to unsatisfied perl(Win32)) lsb-1.3-7mdk.noarch (due to unsatisfied glibc_lsb) mozilla-1.4-2mdk.i586 swfdec-mozilla-0.2.2-2mdk.i586 (due to unsatisfied mozilla[= 1.4]) (Y/n) The following packages have to be removed for others to be upgraded: am-utils-6.0.9-3mdk.i586 (due to unsatisfied libamu2 == 6.0.9-3mdk) apache2-mod_xslt-2.0.46_1.0.5a-1mdk.i586 (due to unsatisfied apache2 == 2.0.46) gaim-0.64-3mdk.i586 (due to unsatisfied perl-base == 2:5.8.0) gcc2.96-2.96-0.82mdk.i586 (due to missing glibc-devel) gcc2.96-c++-2.96-0.82mdk.i586 (due to unsatisfied gcc2.96 == 2.96-0.82mdk) gdome2-devel-0.7.4-1.i386 (due to unsatisfied libxml2-devel = 2.4.26) ghc-5.00.2-2mdk.i586 (due to missing libreadline.so.4.2) ghc-prof-5.00.2-2mdk.i586 (due to unsatisfied ghc == 5.00.2) glibc_lsb-2.3.1-1mdk.i586 (due to unsatisfied lsb = 1.3-5mdk) gnome-pim-conduits-1.4.9-1mdk.i586 (due to unsatisfied gnome-pim == 1.4.9) gtk+licq-0.51-11mdk.i586 (due to unsatisfied licq = 1.1.0) imgSeek-0.7-1.i586 (due to missing libqt-mt.so.3) kmplayer-0.7.4c-1mdk.i586 (due to missing mplayer) libcroco0-devel-0.1.0-2mdk.i586 (due to missing libm.so, due to missing libxml2.so, due to missing libz.so) libgstreamer0.4.2-devel-0.4.2-3mdk.i586 (due to missing libxml2-devel) libgtop1-devel-1.0.13-5mdk.i586 (due to missing libm.so, due to missing libglib-1.2.so) libgtop2.0_0-devel-2.0.1-1mdk.i586 (due to unsatisfied libglib2.0-devel = 2.0.0) licq-1.2.6-3mdk.i586 (due to missing libqt-mt.so.3) minidom-devel-0.1.4-0.i586 (due to missing libxml2-devel) mozilla-1.4-2mdk.i586 (due to unsatisfied libnss3 == 1.4-2mdk, due to missing libprldap50.so, due to missing libldap50.so) mozilla-dom-inspector-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mozilla-enigmail-1.4-2mdk.i586 (due to unsatisfied mozilla-enigmime == 1.4-2mdk) mozilla-enigmime-1.4-2mdk.i586 (due to unsatisfied mozilla-mail == 1.4-2mdk) mozilla-irc-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mozilla-js-debugger-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mozilla-mail-1.4-2mdk.i586 (due to missing libmsgbaseutil.so) mozilla-spellchecker-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mplayerplugin-0.80-1mdk.i586 (due to missing mplayer-gui) octave-2.1.50-1mdk.i586 (due to missing libm.so, due to missing libgcc_s.so, due to missing liboctinterp.so, due to missing libstdc++.so, due to missing liboctave.so, due to missing libcruft.so) psi-0.8.7-3mdk.i586 (due to unsatisfied libqt3 = 3.0.5) umbrello-1.1.1-3mdk.i586 (due to missing libkparts.so.2) (y/N) how do I get around this? Use --keep François.
Re: [Cooker] Re: rpmdrake and proxy (and bwlimit!)
Olivier Blin [EMAIL PROTECTED] writes: Try grep limit_rate /usr/lib/perl5/vendor_perl/5.8.0/urpm.pm Bandwith limit is almost here, it would be quite esay to add one or two more lines of code in urpm.pm to handle a limit_rate: xxx option in urpmi.cfg damned, it's already here, but there's a typo, line 430 : elsif (($k, $v) = /^(limit-rate|excludepath|key_ids|split-(?:level|length))\s*:\s*(.*)$/) { unless (exists($urpm-{options}{$k})) { $v =~ /^'([^']*)'$/ and $v = $1; $v =~ /^([^]*)$/ and $v = $1; $urpm-{options}{$k} = $v; } limit-rate should be limit_rate ... Otherwise it won't work :) Strange, it is effectively a bit bad written here as limit-rate is public option and limit_rate is internal name (to avoid quoting the string ... in order to save my fingers...). key_ids is an exception, I shouldn't have named this one as this. François.
Re: [Cooker] urpmi --auto-select - how to get around it?
Kim Schulz [EMAIL PROTECTED] writes: [snip] Use --keep not an option for my version of urpmi. and when I try to upgrade urpmi, then I get alot of other deps (especially alot of kde and mozilla stuff). I may do an unsupported update of urpmi for 9.1, but not yet. François.
Re: [Cooker] urpmi --auto-select
Adam Williamson [EMAIL PROTECTED] writes: On Fri, 2003-08-01 at 20:42, Dave Cotton wrote: What is going on? urpmi --auto-select suddenly in the middle of everything there's this installing /mnt/Cooker/libfreetype6-2.1.4-5mdk.i586.rpm removing libgdk-pixbuf2-devel-0.22.0-2mdk.i586 XFree86-static-libs-4.3-13mdk.i586 libfontconfig1-devel-2.2.1-2mdk.i586 lsb-1.3-7mdk.noarch freetype2-devel-2.1.4-4mdk.i586 glibc_lsb-2.3.1-1mdk.i586 libgtk+2.0_0-devel-2.2.2-5mdk.i586 libgnome32-devel-1.4.2-7mdk.i586 libkdecore4-devel-3.1.3-2mdk.i586 libpango1.0_0-devel-1.2.3-2mdk.i586 libnas2-devel-1.6-9mdk.i586 libkdenetwork2-devel-3.1.3-2mdk.i586 libimlib1-devel-1.9.14-8mdk.i586 libkdeutils1-devel-3.1.3-1mdk.i586 libarts-devel-1.1.3-1mdk.i586 libkarchiver3-3.0.4-1mdk.i586 libqt3-devel-3.1.2-11mdk.i586 libgdk_pixbuf2.0_0-devel-2.2.2-5mdk.i586 XFree86-devel-4.3-13mdk.i586 libMesaGLU1-devel-5.0.1-5mdk.i586 libgtk+1.2-devel-1.2.10-37mdk.i586 libungif4-devel-4.1.0-23mdk.i586 Something similar here. Best guess, all those packages depended on the old freetype package, this one seems to be a new name. Why did urpmi not say it was going to do this first, though? It is a bug in the resolver, so, I add a sanity check... François.
Re: [Cooker] Re: urpmi error message
Mike Dewey [EMAIL PROTECTED] writes: Hi All: I did this morning urpmi.update -a and main cooker and contrib hdlist where downloaded. Then when I did urpmi --auto-select I got this error: urpmi --auto-select Can't use string (kdebase) as an ARRAY ref while strict refs in use at /usr/lib/perl5/vendor_perl/5.8.0/urpm.pm line 2989. Can you test with latest urpmi, as for sure, the line 2989 goes in wrong place currently. François.
Re: [Cooker] More on the obsoletes stuff
Charles A Edwards [EMAIL PROTECTED] writes: On 01 Aug 2003 17:39:09 +0200 [EMAIL PROTECTED] (François Pons) wrote: The problem is the semantics we affect, a package provides a property and another obsoletes it, does we need to keep it or ignore it ? Does not using Provides: foo = %version Obsoletes: foo %version elicit the desired behaviour A new rpm (15mdk) has been uploaded, and such behaviour has no more effects in fact. Now obsoletes only affect *real* name of package and no more provides, this means there will be no more strange package being removed (arts or kdebase) without notice. A new perl-URPM is being done to handle this new obsoletes behaviour (as perl-URPM was trying to follow rpm bugs, which of course was relatively hard to follow correctly). François.
Re: [Cooker] arts obsoletes current kdebase
Olivier Thauvin [EMAIL PROTECTED] writes: Le Jeudi 31 Juillet 2003 22:29, Olivier Blin a écrit : $ rpm -q arts --obsoletes | grep kdebase3 kdebase3 kdebase3-nsplugins kdebase3-devel $ rpm -q --whatprovides kdebase3 kdebase-3.1.3-1mdk arts obsoletes current kdebase, how can I upgrade to kde 3.1.3 ? Hue, I have doubt this will be understanded by any rpm which require a kdebase, this one can safely be removed, some bytes will be saved... I complain because arts spec is not light and not clear, but kdeteam is not so stupid !! kdebase-3.1.3-1mdk means kdebase = 3.1.3-1mdk, the obsolete apply to kdebase3, and kdebase != kdebase3. And it seems rpm is a bit buggy, I send a bug report about some strange things I found to F. Lepied and F. Pons, but I am still waiting a response. We are waiting from response from RH, F. Lepied forwarded to them. But vacations may not be helpful for us, I may dig into rpm again else. François.
Re: [Cooker] Re: [CHRPM] perl-5.8.0-28mdk gaim
Luca Berra [EMAIL PROTECTED] writes: On Tue, Jul 29, 2003 at 06:53:48PM +0200, Luca Berra wrote: Levi Ramsey wrote: # rpm -q --provides perl-base ... perl-base = 2:5.8.0-28mdk Looks like there's an Epoch problem in rpm/urpmi... nope, it is a problem in gaim: perl-base 2:5.8.0-28mdk is GREATER than perl-base 2.8.0 gaim should require perl-base = 5.8.0 now we have [EMAIL PROTECTED] cache/urpmi/rpms # urpmi --test perl-base-5.8.1-0.RC3.1mdk The following packages have to be removed for others to be upgraded: gaim-0.64-3mdk.i586 (due to unsatisfied perl-base == 2:5.8.0) (y/N) since perl-base = 2:5.8.1-0.RC3.1mdk is GREATER than perl-base = 2:5.8.0 what about using the = symbol i proposed early A better approach could be a provides in perl-base (and perl) about which version are correctly supported, ie the following provides for perl-base : perl-base = 2:5.8.0 perl-base = 2:5.6.1 perl-base = 2:5.6.0 So that gaim will not be removed again, the same is problably needed for perl too. I let Pixel decide if he wants this, maybe I missed something about that. François.
Re: [Cooker] urpmi (wget) instability - annoyance?
Tomasz Chmielewski [EMAIL PROTECTED] writes: Hi, I have a dsl connection, dynamic IP, and my provider changes my address every few hours. Whenever it happens, and I use urpmi in a meantime - urpmi just freezes. Maybe it's more of wget issue than of urpmi, but it would have been nice r if urpmi checks from time to time if something is really being downloaded, and if not - restart wget, ask what to do etc. Below - the state in which wget and urpmi are for an hour or so. 32233 pts/2S 0:09 /usr/bin/perl /usr/sbin/urpmi --no-verify-rpm --auto-select 32235 pts/2S 0:00 /usr/bin/perl /usr/sbin/urpmi --no-verify-rpm --auto-select 32236 pts/2S 0:00 /usr/bin/perl /usr/sbin/urpmi --no-verify-rpm --auto-select 32607 pts/2S 0:00 /usr/bin/wget --progress=bar:force -o - --retr-symlinks -NP /var/cache/urpmi/partial ftp://fr2.rpmfind.net/linux/MandrakeCooker/cooker/i586/Mandrake/RPMS/libbon... Default timeout for wget is 15 minutes, did you wait enough to see if timeout worked or not ? I may lower it, and retry again automatically... François.
Re: [Cooker] weird curl/urpmi problem
Kim Schulz [EMAIL PROTECTED] writes: Hi Hi, I wanted to use curl insted of wget for the updating of my urpmi hdlists, but curl gives me alot of errors: urpmi.update -a --proxy wwwproxy.nork.auc.dk:3128 --curl retrieving description file of noo1... ...retrieving failed: curl failed: exited with 22 or signal 0 This sounds normal, but this annoying message was removed in current urpmi. retrieving source hdlist (or synthesis) of noo1... curl: option - is unknown curl: try 'curl --help' or 'curl --manual' for more information This is not normal, what version do you have of curl, urpmi ? ftp://sunsite.uio.no/linux/Mandrake/Mandrake-devel/cooker/i586/Mandrake/base/hdlist.cz 57% of 18.2M completed, ETA = 0:00:03, speed = 2484k (after 100%) ...retrieving done examining synthesis file [/var/lib/urpmi/synthesis.hdlist.noo1.cz] retrieving description file of noo2... ...retrieving failed: curl failed: exited with 22 or signal 0 Still normal, but now removed. Best Regards, François.
Re: [Cooker] urpmi - segmentation fault (long mail)
John van Spaandonk [EMAIL PROTECTED] writes: While experimenting with urpmi I tried urpmi --auto-select --bug --media cooker and I noticed that the --bug supresses my explicit media selection, so in my case packages from cooker, plf and contrib were installed... urpmi --auto-select --bug urpmibug --media cooker will work better. Is this intentional behavior of urpmi? Is is strange cooker has been resolved to something (It has been interpreted as a package name, subname or provides). François.
Re: [Cooker] urpmi urpmi gives weird errors
Kim Schulz [EMAIL PROTECTED] writes: Im trying to upgrade my urpmi to latest version (to get rid of the curl error messages) but I keep getting this: # urpmi urpmi Some package requested cannot be installed: kdebase-3.1.3-4mdk.i586 (due to missing kdebase-kdm-config-file-3.1.3-3mdk.i586) kdebase-kdm-config-file-3.1.3-3mdk.i586 (due to unsatisfied kdebase[== 3.1.3-3mdk]) (Y/n) The following packages have to be removed for others to be upgraded: ghc-5.00.2-2mdk.i586 (due to missing libreadline.so.4.2) ghc-prof-5.00.2-2mdk.i586 (due to unsatisfied ghc == 5.00.2) gimp-perl-1.2.5-2mdk.i586 (due to unsatisfied gimp == 1.2.5-2mdk) kdegraphics-devel-3.1.2-6mdk.i586 (due to unsatisfied kdegraphics == 1:3.1.2-6mdk) kdelibs-devel-3.1.2-20mdk.i586 (due to unsatisfied kdelibs == 3000:3.1.2-20mdk) libcroco0-devel-0.1.0-2mdk.i586 (due to missing libm.so, due to missing libxml2.so, due to missing libz.so) libgtop1-devel-1.0.13-5mdk.i586 (due to missing libm.so, due to missing libglib-1.2.so) mozilla-1.4-2mdk.i586 (due to missing libprldap50.so) mozilla-dom-inspector-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mozilla-enigmail-1.4-2mdk.i586 (due to unsatisfied mozilla-enigmime == 1.4-2mdk) mozilla-enigmime-1.4-2mdk.i586 (due to unsatisfied mozilla-mail == 1.4-2mdk) mozilla-irc-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mozilla-js-debugger-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) mozilla-mail-1.4-2mdk.i586 (due to missing libmsgbaseutil.so) mozilla-spellchecker-1.4-2mdk.i586 (due to unsatisfied mozilla == 1.4-2mdk) octave-2.1.50-1mdk.i586 (due to missing libm.so) swfdec-mozilla-0.2.2-1mdk.i586 (due to unsatisfied mozilla = 1.4-0.beta.1mdk) (y/N) What do I need to upgrade to get this to work? Try with urpmi urpmi perl-URPM --keep François.