Re: [Cooker] we've been right all along
Le mar 01/04/2003 à 13:05, James Sparenberg a écrit : Or how about one where MDK drops urpmi for up2date and joins united linux. *grin* I agree, this could effectively be a great improvement of using up2date instead of urpmi, I will think about this. Again, joining united linux will remove some mandrake specifics too. François.
Re: [Cooker] 9.0 - 9.1 with urpmi - experience
Le dim 23/03/2003 à 05:14, Adam Williamson a écrit : Overall, I can broadly say...it worked. Good marks for that :). But, there were some gotchas along the way. First of all there was rather an annoying problem with openSSL. When I just did an --auto-select, it failed due to missing openssl-devel. I *think* the problem is the 0.96 (9.0) and 0.97 (9.1) packages don't work well together, and the --auto-select was just planning on removing the 0.96 packages (or at least the devel package) without updating to the 0.97 packages. My workaround was to download and install the 0.97 packages manually. This Since urpmi is experiencing big improvement since 9.0, it was necessary to upgrade at least before urpmi (using updates or unsupported) before doing the auto-select itself. For the 9.1 this is still the case as lot of fixes have been done for newer urpmi (so small glitches using old urpmi may appear on some case which is problably the case using openssl-devel). Secondly, urpmi is still not quite good enough for doing a between-versions upgrade entirely automatically, because of the problem of adding new packages. It installed several config files which expected the Galaxy theme packages, but of course it didn't install these, I had to do it manually. I also manually installed the zcip and tmdns packages, because it seems these are necessary for 9.1 machines to work as simple network clients - if you try and set up a system with drakconnect without these packages installed, it tries to install them (which obviously doesn't work if you only have network sources :). If using urpmi to update is to become a recommended path, as I believe some people have discussed, I propose it needs some kind of way of taking into account highly recommended new packages in the new version like this. All of this is config files handling (which is not done by urpmi) and media management (for avoiding network usage by drakconnect) which need doing a little to help it. Overall, though, I was very impressed - after I worked around these niggles, it happily updated the entire distribution (over 600 packages). I tweaked the surprisingly few config files that installed as .rpmnew, installed the new Cooker kernel, rebooted and there was 9.1, in all its glory! Good stuff, it's nice that this pretty much works now :). It will be effectively better when it upgrades smootly correctly, there will be new features for next urpmi which will be smaller transaction, allowing /var not to become full for upgrading entire distribution. Doing updates of urpmi will also help this process to be taken into account, else it is generally necessary to just urpmi urpmi in order to update urpmi first before updating everything else. François.
Re: [Cooker] Ideas 4 future (urpmi incremental, cookerdrake)
Le jeu 20/03/2003 à 06:25, James Sparenberg a écrit : Correct me if I'm wrong here but couldn't urpmi.update -a just use rsync to grab the hdlist. I'm thinking that if you have it once and since the name never changes the second++ time you grab it it would always be an incremental change... not a whole download. This is exactly the case, unless you use -f which avoid this feature. François.
Re: [Cooker] Ideas 4 future (urpmi incremental, cookerdrake)
Le jeu 20/03/2003 à 16:41, Leon Brooks a écrit : On Thursday 20 March 2003 01:25 pm, James Sparenberg wrote: Correct me if I'm wrong here but couldn't urpmi.update -a just use rsync to grab the hdlist. Deem yourself corrected. The hdlist is re-sorted and re-compressed from scratch each time it changes. This does not make for a happy rsync experience. )-: The sorting should be kept with incremental version of package (always ldconfig first, ... according to most used package (perl-URPM counts reference to each package globaly before building hdlist)). Compression of hdlist is done by section, not the whole file is compressed but only section of 40 bytes (generally) of concatenated rpm headers (some headers have size of 1.2Mb). This means rsync can behave correctly. François.
Re: [Cooker] urpmi features
Le mar 18/03/2003 à 04:09, Robert L Martin a écrit : Priority for sources. So that if two (or more) totaly same packages exist in two (or more) sources you can set the priority so that the package gets installed from the media with higher priority. For example some prefere local sorces to have higher priority and some prefer some FTP server for packages. This is currently linked to media priority, which sound enough for me. François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 19:55, Wesley J. Landaker a écrit : urpmi then installs them all, supposedly in a nice correct order (I don't doubt it's correct--I just haven't ever looked at how it does it currently No problem, this problem is solved by rpmlib which looks at PreReqs and Requires. Instead, what urpmi could do is the following (and maybe it already does some of this internally--if so, hey! less work!) Currently, urpmi use a non recursive algorithms without global optimization on requires to be fetched (to be fast enough, do not forget resolution algorithms is pure perl wchich is fast but interpreted). Yeah, it's not going to get the perfectly smallest subsets when there are cycles. But it in real practical use for RPMs it's *going* to generally find lots of small subsets, even if they're not optimal, with very little processing overhead. This is what will be done, I didn't looked yet about a specific algorithm but I is effectively not very hard to solve. François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 19:35, Pascal Terjan a écrit : * possibility to use only one or several media, but less restricted. There is an actual option to use only one source, --media, but what I would like FYI you can use more than one source in fact, separated by comma (,). In fact the best improvment for urpmi would be documentation anad advertising :-) FYI --media is documented this way using comma separated list, I agree there is lack of documentation, but user documentation is minimal and only devel documentation is lacking (especially for perl-URPM and urpm libraries). It will be taken as to be strongly needed, as perl-URPM reaches 1.0 release (only 0.81 currently, not far from now). François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 20:08, Pascal Terjan a écrit : I think it could at least be on urpmi.org (which is an independant website which goal is to gather documentation about urpmi so it can be more advertized). Just type urpmi in google and gather all documentation available and links, I tried it yesterday to gather opinions and idea about improvement. François.
Re: [Cooker] urpmi features
Le mar 18/03/2003 à 00:08, Steffen Barszus a écrit : - rsync for updates now that rsync is supported by urpmi (as well as https protocol), we can propose other mirror capabilities. - if a download was aborted use the part that is allready there and download the rest then This can automatically activated according to protocol used, typically rsync allow that, whereas http or ftp protocol should be tried differently, maybe by allowing continue download and download again on error (only the beginning) if rpm is still inconsistent ? - if a server is to slow / not reachable, take the next Auto tune server before downloading ? - ability to configure a path there packages go to if --noclean ability to use a different configuration file, which may authorize defining all other file to use. - urpmi.importmedia (mirror content to hd, build hdlist, add media) Yes, media management can be improved to create automatically environment for updating media, spliting and others. This looks like mkcd or gendistrib features, maybe factorisation ? Don't know if it is meant by p2p magic-hdlist : a click on a link to add media/ urpmi://...hdlist.cz would maybe be fine too (Easy Urpmi thought one step further) I was thinking on synthesis containing all needed informations as an extensions. One other idea is nothing that can do urpmi: meta-packages for special purposes (like the groups during installation) Yes, I was thinking about it for 9.1, just need to use compssUsers file, no added for 9.1 but great effectively. François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 21:55, Buchan Milne a écrit : Can we have per-source 'skip.list's ? This has just shown up, since we run samba-server-ldap (requires samba-common-ldap), which obsoletes and provides samba-server (and vice versa) to allow users to move from one to the other. All goes well however, until you get a security update for samba, and it installs samba-server over samba-server-ldap and we have downtime while I scp samba-server-ldap packages to the server and add samba-server to the skip.list but now that I have newer ldap-enable builds (will be on ftp.samba.org soon), I would have to take samba-server out of the skip list again. This sounds me a problem of using obsoletes whereas a conflicts may be better ? François.
Re: [Cooker] Re: urpmi features
hatLe mar 18/03/2003 à 11:29, Buchan Milne a écrit : -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Wesley J Landaker wrote: On Monday 17 March 2003 5:35 pm, Pascal wrote: # urpmsave --dir ~/rpmsavedir myrpmname (before updating glibc and crashing all the system :) ) creates a tgz of all files installed by the rpm in the specify directory so that u can restore them in case of emergency This would be handy if it would have options for restoring backed up rpms made in this fashion and handled doing smart things with the urpm database. A simple implementation isn't too hard, though: #!/bin/sh rpm -ql myrpmname | xargs tar cvfj ~/rpmsavedir/myrpmname.tar.bz2 No need. 'rpm --repackage' works on upgrade, erase and (apparently) intall. Having urpmi support this would be cool ... Great feature I wasn't aware of :-) Yes, it can be done that way. François.
[Cooker] urpmi features
Possible features of urpmi for next release : * virtual medium (no need to update explicitely, only with synthesis ?). * delay before accepting using a package from a medium (cooker) * always ask confirmation if fuzzy search is used (even for 1 package) * on the fly sorting of media (according to regex like file,rsync,ftp,http) * urpm centralized tools, as well as perl-URPM managing media. * p2p urpmi database (export database as magic synthesis) * -h by default for urpmi.addmedia * do install of package by groups which are shorter as possible (apt-get like) * allow file conflicts error to be handled by recovering errors and try again. * conflicts, provides and requires tag added for global rpm behaviour in order to allow broken dependencies to be not resolved or to avoid removing important package (generalize basesystem) Any other idea are wellcome. François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 18:37, Eric Fernandez a écrit : * possibility to use only one or several media, but less restricted. There is an actual option to use only one source, --media, but what I would like FYI you can use more than one source in fact, separated by comma (,). François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 18:39, Levi Ramsey a écrit : * do install of package by groups which are shorter as possible (apt-get like) Couldn't this be accomplished just as effectively through virtual packages? This has nothing to do with virtual package in fact, this is just organization of grouping package during installation with separate transaction (in order to avoid blocking if available memory disk space is tight, or some other error appears, like conflicts with file). * conflicts, provides and requires tag added for global rpm behaviour in order to allow broken dependencies to be not resolved or to avoid removing important package (generalize basesystem) I'm going to working on a an automatic rebuild capability (somewhat Gentoo-ish)... I've been thinking of doing this for a while, and the time for action draws near... ;o) This features is more an rpm features than a urpmi one. François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 19:12, Pascal Terjan a écrit : François Pons wrote: Possible features of urpmi for next release : * virtual medium (no need to update explicitely, only with synthesis ?). * delay before accepting using a package from a medium (cooker) I'm not sure that's usefull. Or maybe for some important packages, but I everyone delays testing, that will delay fixing... This can be seen as second quality testing, its purpose is to avoid using completely broken package uploaded to cooker. This has been requested by some user of cooker wich use this almost on production machine. - Explicit priority for sources (Or at least document that currently it's based on the order :) This has been requested for changing order on the fly, the idea is to use the same approach as --media, this can be very discriminant or global to match all local based media. - Acces to information like package description from hdlist with urpmq This one seems to be requested too. François.
Re: [Cooker] urpmi features
Le lun 17/03/2003 à 19:17, Eric Fernandez a écrit : Having urpmi.setup integrated, which would do the same than http://plf.zarb.org/~nanardon. It's here : http://www.urpmi.org/en/urpmi.setup/index.html And I don't think it would be a problem of legality with PLF since they would not be configured in advance, would they ? This is already the case, but using a console tools approach (not a gtk based application), look at --distrib-XXX option and all the other tunning options used. François.
Re: [Cooker] BUG: cyrus-imapd-2.1.12-4mdk - Install error
Le jeu 13/03/2003 à 15:46, Florin a écrit : rpm -qa|grep imap ? rpm -q --whatprovides imap ? $ urpmf --provides --conflicts :imap courier-imap:provides:imap courier-imap:provides:imap-server cyrus-imapd:provides:imap cyrus-imapd:conflicts:imap bincimap:provides:imap bincimap:provides:imap-server imap:provides:imap[== 1:2002a-2mdk] imap-devel:provides:imap-devel[== 1:2002a-2mdk] php-imap:provides:imap.so imap:provides:imap[== 1:2000-6mdk] imap-devel:provides:imap-devel[== 1:2000-6mdk] php-imap:provides:imap.so François.
Re: [Cooker] BUG: cyrus-imapd-2.1.12-4mdk - Install error
Le jeu 13/03/2003 à 16:19, magic a écrit : Same thing I get (except the order is different): # urpmf --provides --conflicts :imap imap:provides:imap[== 1:2002a-2mdk] imap-devel:provides:imap-devel[== 1:2002a-2mdk] php-imap:provides:imap.so courier-imap:provides:imap courier-imap:provides:imap-server cyrus-imapd:provides:imap cyrus-imapd:conflicts:imap bincimap:provides:imap bincimap:provides:imap-server - So, what does that tell me? (No expert on urpmi.) It tells you I'am smoking the moket here ;) and it appears providing and conflicting isn't valid whereas I suspected it was supported. So package will be remade. François.
[Cooker] Re: noauto: tag must be removed in Mandrake/base/hdlists (onmirrors) for gendistrib to work
Le sam 08/03/2003 à 13:54, Warly a écrit : Franco Silvestro [EMAIL PROTECTED] writes: inside Mandrake/base/hdlists must be removed noauto: tag for gendistrib to work cuand good work...;o) François, you have not included noauto support in gendistrib? No, strange it wasn't, but now it is fixed in 9mdk. François.
Re: [Cooker] URPMI in the spotlight, was: question on ftp sites
Le lun 10/03/2003 à 12:16, Michael Scherer a écrit : Le Lundi 10 Mars 2003 12:04, El Gringo ( aka Guillaume Rousse ) a écrit : Ainsi parlait Michael Scherer : Very effective, everybody who read Linux Magazine France know that apt-get is so cool http://lis.snv.jussieu.fr/~rousse/linux/urpmi.html for instance ? Guillaume, it should be nice to add extra functionnalities added since one year ? Does someone is interrested ? For example, * ssh:// and rsync:// supported protocol. * distributed installation support. * cache management improved. * urpmf logical expression management. * mirrors management support compatible with urpmi.setup. * allow some global options to be defined in urpmi.cfg file. * a lot of newer options since for urpmq/urpmi/urpmf. Yes, exactly what I was thinking :-) Very good intro, that is what make me discover urpmi. Guillaume is really smart to explain things... Another thing to do is that we should stop to say that rpm can be find on rpmfind.net. Almost each time people go on rpmfind, they download a rpm for another distro, and, when it don't work, they say that rpm is bad. I wholeheartly agree. Users see rpms as just another archive format, and completly miss the integration work behind. What about using a different extension ? Something as mrpm ? No, it is a rpm, urpmi should work on other distros, and urpmi can work if you build media of package not initially done for Mandrake. There is no specific extension in rpm for urpmi, only standard flag are used. François.
Re: [Cooker] URPMI in the spotlight, was: question on ftp sites
Le lun 10/03/2003 à 14:22, Buchan Milne a écrit : While we are here, any chance to have gui tools for 1)Selecting how often to run automatic updates This could be great indeed, any idea how to improve urpmi/rpmdrake is welcome as next version will be designed soon now. 2)Adding entries to /etc/urpmi/skip.list I had a bad experience when testing the above script the first time, since I had samba-server-ldap running in production on a server, and urpmi --auto-select decided the samba-server-2.2.7-2mdk from updates should replace samba-server-ldap-2.2.7a-3mdk running on the server ... resulting in about 2 minutes of downtime wrt. samba while I scp'ed ldap-enabled RPMS to it etc). You may want to use --bug $LOG/`date '+%Y%m%d%H%M%S'` to the urpmi command line, so that you have a per session based update log (so that if something fail, you have a bug report to send me too). Just create a compressed archive maybe on the fly to avoid keeping too much place if necessary ? François.
Re: [Cooker] URPMI in the spotlight, was: question on ftp sites
Le lun 10/03/2003 à 16:17, Buchan Milne a écrit : This could be great indeed, any idea how to improve urpmi/rpmdrake is welcome as next version will be designed soon now. A request from the [EMAIL PROTECTED] list was to have a delay available, so that you could run updates daily, but never get an update which was less than X days old. This would allow you to run with delay of 0 on a test server, and a greater delay on production machines (10-15 days seems to be about optimal according to some studies), allowing you to have peace-of-mind that you will not unknowingly install a broken update on a production server, assuming you keep an eye on the test server or the security lists. Ok, noted. The problem was AFAIK not due to, or fixable by urpmi, since we have samba-server obsolete samba-server-ldap, and vice versa, to allow people to switch between them easily. This is really a hack due to samba only having compile-time configurable account backends in 2.2.x, solved in 3.0 with run-time configurable (including multi) account backends. This was not obsoletes but conflicts which was needed, urpmi ask confirmation to remove the conflicting package. François.
Re: [Cooker] fonts in UTF8 sv-us
Le mer 05/03/2003 à 17:27, guran a écrit : Hi Hi, Hd.img inst UTF8 sv-us version: Mandrake Linux Cooker-i586 20030304 21:21 /ChangeLog/1.801/Tue Mar 4 13:54:59 2003// This is the way it is. [EMAIL PROTECTED] root]# urpmi --auto-select --auto allt är redan installerat ^^^ this should be ä. Can you check with 28mdk ? Can you check with urpmi --auto-select --auto | cat - too François.
Re: [Cooker] URPM broken or Cooker hdlist.cz?
Le dim 02/03/2003 à 16:09, Brian J. Murrell a écrit : I am trying to add my local Cooker repository here as a urpmi source but it's failing to parse the hdlist: # urpmi.addmedia Local\ Cooker http://linux.ilinx/mandrake/devel/cooker-tree/Mandrake/RPMS/ with ../base/hdlist.cz added medium Local Cooker examining synthesis file [/var/lib/urpmi/synthesis.hdlist.contrib.cz] examining synthesis file [/var/lib/urpmi/synthesis.hdlist.Texstar.cz] examining synthesis file [/var/lib/urpmi/synthesis.hdlist.plf.cz] retrieving description file of Local Cooker... retrieving source hdlist (or synthesis) of Local Cooker... http://linux.ilinx/mandrake/devel/cooker-tree/Mandrake/base/hdlist.cz ...retrieving done examining hdlist file [/var/cache/urpmi/partial/hdlist.Local Cooker.cz] examining synthesis file [/var/cache/urpmi/partial/hdlist.Local Cooker.cz] unable to parse hdlist file of Local Cooker examining synthesis file [/var/lib/urpmi/synthesis.hdlist.Local Cooker.cz] problem reading synthesis file of medium Local Cooker unable to update medium Local Cooker I have the latest URPM package: $ rpm -q perl-URPM perl-URPM-0.81-10mdk The hdlist does look suspect however as it's only 9190174 bytes big as opposed to the 14MB or so 9.0 is. I can successfully add my 9.0 repository as a source on this machine so URPM/urpmi seems to be generally functional. There seems to be problem sometimes on hdlist, if the problem persist use synthesis instead. Mirror could have problem temporary ? François.
Re: [Cooker] Removing basesystem?
Le mar 25/02/2003 à 17:55, Jack Coates a écrit : On Tue, 2003-02-25 at 07:47, Alexander Skwar wrote: Hi! After quite some time, I now wanted to do a urpmi --auto-select again, to get my system to the current Cooker state. urpmi shows me: Die folgenden Pakete müssen entfernt werden, um die Aktualisierungen durchführen zu können: basesystem-9.0-3mdk gstreamer-audiofile-0.4.2-4mdk libgnomeprint2.0_0-1.116.0-2mdk libmysql10-devel-3.23.53-5mdk libpgperl-7.2.3-2mdk libpgsqlodbc0-7.2.3-2mdk mozilla-xmlterm-1.1-10mdk postgresql-tk-7.2.3-2mdk Why does urpmi want to remove basesystem? no idea, but it does sometimes act a bit flaky. Most likely it wants to replace it with something else, probably basesystem-9.1-rc1-minor-version-mdk. Personally, I would download some of the other packages and urpmi them one-by-one, then see if the problem went away or added some explanatory symptoms. First, according to message displayed you are not using latest urpmi because for some big time now, urpmi give a reason why a package will be removed. So please upgrade urpmi and perl-URPM first, and then try again to see why. François.
Re: [Cooker] urpmi.listmedia (or the equivalent)
Le lun 24/02/2003 à 03:09, Jon Gabrielson a écrit :
Thanks, i fixed a small bug where the url wasn't being
updated correctly with multiple media sources. I also made
it so that it outputs the message as a sample syntax.
Attached is the new version.
Don't you prefer a shorted one ?
perl -Murpm -e '$a=new urpm;$a-read_config; print map {
$_-{name}\t$_-{url}\n } @{$a-{media}}'
François.
Re: [Cooker] Proposal: new option for urpmi
Le sam 22/02/2003 à 06:23, Ben Reser a écrit : Here's a script that I wrote for this purpose. Though it doesn't use the changelogs in hdlist. Since gc mentioned they are in there maybe I should go back and figure out how to get them from that file and rewrite the script. Yes, you will gain a lot of time by reading directly hdlist of urpmi media by using urpm library. I have take a quick look of script, you will save many seconds of time. François.
Re: [Cooker] urpm.po
Le jeu 20/02/2003 à 21:41, Steffen Barszus a écrit : 1) Anyone can say, in which context this will be displayed ? callback is :\n %s\n This is for urpmf (debug mode, or if you prefer verbose mode) in order to see perl code generated by the expression given on command line. 2) --from - use specified url for list of mirrors, the default is\n %s\n is this meant as use specified url to fetch a list of mirrors, the default is\n ? Yes, my english is as bad as my french, so do not expect a huge improvement on these topics. François.
Re: [Cooker] Question.
Le mer 19/02/2003 à 22:02, Pedro Soria-Rodriguez a écrit : So my question is: Is my train of thought wrong? Am I missing something? Problably not, but cooker is experimental and your problem may be caused by bug wich appears on your computer due to hardware or other installed software. François.
Re: [Cooker] short list of fixes to make rc1 livable
Le jeu 20/02/2003 à 16:53, tarvid a écrit : urpmi often produces broken installations, e.g. urpmi xchat fails, rpm -Uvh on the downloaded file works You are using urpmi on RC1 directly on network ? What is your mirror as we do not have produced such a live tree (only iso) ? François.
Re: [Cooker] urpmi feature request
Le jeu 13/02/2003 à 23:01, Tibor Pittich a écrit : hello hello, i have a idea for feature which will be imho useful. if it is possible now, sorry, but i don't know about it now. there is many possible urpmi sources, which isn't pure mandrake source. i mean for example plf, mandrake club packages, and other. imagine this situation: i want daily update my machine from cooker, and i'm sure, that i want use only cooker packages. if i have define all sources as equals (which is imho now only possibility) then urpmi only check versions at all sources and download install never. this behavior i can change with --source switch, but this is venomous for daily usage. my question is: is it possible create one or more source as non important and this source will be used only if i requested his usage with --source ? normally urpmi (--auto-select etc.) wasn't used this source. In fact, there is exactly the reverse, there is an --update flag allowing to select stable media, but there is no flag for selecting unstable media. François.
Re: [Cooker] rpmdrake-2.1-4mdk, -5mdk CPU usage problem (akagrpmi zombie)
Le ven 14/02/2003 à 14:33, Buchan Milne a écrit :
# echo initscripts /etc/urpmi/skip.list
# urpmi.update -a
# urpmi --auto-select --auto --no-verify-rpm
(the last argument I would like to drop, but is necessary if you use
contrib :-()
No, see changelog, you add the following line into /etc/urpmi/urpmi.cfg
{
no-verify-rpm
}
François.
Re: [Cooker] [urpmi] no longer work with club authenticated HTTPdownloads when used with curl
Le jeu 13/02/2003 à 18:25, Lea Gris a écrit : Using the --wget option for urpmi is a workaround for this bug that appear to be linked to the urpmi way of using curl And curl alone gives what ? François.
Re: [Cooker] another drakx report
Le jeu 06/02/2003 à 02:09, Pixel a écrit : Bad: network install doesn't properly configure urpmi database Detail: this has been a problem for many versions. I don't see why this can't work. I just installed from an ftp site, urpmi should be setup to use that in the future. The problem is that I logged in with username: anonymous passwork: [EMAIL PROTECTED] and uprmi gets setup for ftp://anonymous:[EMAIL PROTECTED]@mandrake.redbox.cz/.. I know it's nice to hide the password, but it's also stupid to have to urpmi.remove that, and then urpmi.addmedia it right back. i think the password is there in /var/lib/urpmi/list.*, but francois is more authoritative on this Yes, it is a known limitation, please protect the @ of the password, there is a %40 to use instead of @ in the password. François.
Re: [Cooker] urpmf bug
Le mar 11/02/2003 à 17:12, Guillaume Cottenceau a écrit : humm i reported this humm bug to libstdc++ but that's a feature to let you use perl regexp in file searching. Try to but backslash before the +. Probably, seeing two '+' next to each other could be a hint to disable perl regexp searching, francois WDYT? I was thinking exactly at this solution, this improve my volonté to code it so. François.
[Cooker] Re: wrong dependency on gd2
Le dim 02/02/2003 à 02:07, David Walser a écrit : gd2 should require for build libpng-devel and not libpng3-devel, since it works with libpng2 (and libpng3-devel provides libpng-devel). Also the regular requires should be just libpng, and not libpng2 (!) likewise. Fixed. Also (Ben Reser), when I put gd2 in rpmmon it didn't work, even though that's the name of the SRPM. I had to put libgd2. This is because only binary package are searched in database I suppose. Warly may answer better than me on this topic. François.
Re: [Cooker] urpmi --auto-select broken?
Le lun 03/02/2003 à 12:19, Adam Williamson a écrit : Trying to do a standard-issue: urpmi.update -a urpmi --auto-select -v I get this: [root@aw280 adamw]# urpmi --auto-select -v examining synthesis file [/var/lib/urpmi/synthesis.hdlist.plf.cz] examining synthesis file [/var/lib/urpmi/synthesis.hdlist.cooker.cz] examining synthesis file [/var/lib/urpmi/synthesis.hdlist.contrib.cz] One of the following packages is needed: 1- libgstreamer0.6-0.6.0-2mdk.i586 2- libgstreamer0.6_3-0.6.0-1mdk.i586 What is your choice? (1-2) 1 The following packages have to be removed for others to be upgraded: libGConf2_4-devel-2.2.0-1mdk (due to missing libORBit2-devel) libIDL2_0-devel-0.8.0-1mdk (due to missing libglib2.0-devel) libORBit2_0-devel-2.6.0-1mdk (due to unsatisfied libIDL2-devel = 0.7.4) libatk1.0_0-devel-1.2.2-1mdk (due to unsatisfied libglib2.0-devel = 2.0.0) libbonobo-activation4-devel-2.2.0-1mdk (due to unsatisfied liblinc-devel = 0.5.1) libbonobo2_0-devel-2.2.0-1mdk (due to unsatisfied libbonobo-activation-devel = 1.0.0) libbonoboui2_0-devel-2.2.0-1mdk (due to unsatisfied libgnomecanvas2-devel = 1.116.0) libgail16-devel-0.16-1mdk (due to missing libgnomecanvas2_0-devel) libgdk_pixbuf2.0_0-devel-2.2.0-3mdk (due to unsatisfied libglib2.0-devel = 2.2.0) libglade2.0_0-devel-2.0.1-3mdk (due to unsatisfied libgtk+2.0-devel = 2.0.0) libglib2.0_0-devel-2.2.0-1mdk (due to unsatisfied libglib2.0_0 == 2.2.0) libgnome-desktop-2_2-devel-2.2.0-1mdk (due to missing libgnomeui2_0-devel) libgnome-vfs2_0-devel-2.2.0-1mdk (due to unsatisfied libbonobo2_0-devel = 2.0.0) libgnome2_0-devel-2.2.0.1-1mdk (due to unsatisfied libbonobo2_0-devel = 2.0.0) libgnomecanvas2_0-devel-2.2.0.1-1mdk (due to unsatisfied libgtk+2-devel = 2.0.3) libgnomeprint2-2_0-devel-2.2.1-1mdk (due to missing libbonobo2_0-devel) libgnomeprintui2-2_0-devel-2.2.1-1mdk (due to unsatisfied libgnomeprint2-2_0-devel = 2.2.1) libgnomeui2_0-devel-2.2.0.1-1mdk (due to unsatisfied libgnomecanvas2_0-devel = 2.0.0) libgstreamer0.4.2-devel-0.4.2-3mdk (due to missing libglib2-devel) libgstreamer0.5_3-devel-0.5.2-1mdk (due to missing libglib2-devel) libgtk+-linuxfb-2.0_0-devel-2.2.0-3mdk (due to unsatisfied libpango1.0-devel = 1.2.0-3mdk) libgtk+-x11-2.0_0-devel-2.2.0-3mdk (due to unsatisfied libpango1.0-devel = 1.2.0-3mdk) libgtkhtml2_0-devel-2.2.0-1mdk (due to missing libgail-devel) liblinc1-devel-1.0.1-1mdk (due to unsatisfied libglib2.0-devel = 2.0.0) libpanel-applet-2_0-devel-2.2.0-1mdk (due to unsatisfied gnome-desktop-devel = 2.1.4) libpango1.0_0-devel-1.2.0-4mdk (due to unsatisfied libglib2.0-devel = 2.2.0) libwnck-1_4-devel-2.2.0-1mdk (due to missing libgtk+2.0-devel) do you agree ? (Y/n) N It looks like some of these are basically bogus; e.g. I don't have libglib2.0-devel, because it's libglib-2.0_0-devel... What's going on? I don't see any problem, what is wrong ? beware that libglib2.O-devel may be a virtual provide (ie not a package name), furthermore, if you encounter a bug, please send me a bug report. François.
[Cooker] Re: wrong dependency on gd2
Le dim 02/02/2003 à 02:07, David Walser a écrit : gd2 should require for build libpng-devel and not libpng3-devel, since it works with libpng2 (and libpng3-devel provides libpng-devel). Also the regular requires should be just libpng, and not libpng2 (!) likewise. Fixed. Also (Ben Reser), when I put gd2 in rpmmon it didn't work, even though that's the name of the SRPM. I had to put libgd2. That problably normal as only binary package are searched ? Warly ? François.
Re: [Cooker] urpmi --auto-select broken?
Le lun 03/02/2003 à 14:27, Adam Williamson a écrit : On Mon, 2003-02-03 at 12:02, François Pons wrote: What's going on? I don't see any problem, what is wrong ? beware that libglib2.O-devel may be a virtual provide (ie not a package name), furthermore, if you encounter a bug, please send me a bug report. François. The problem is that urpmi has never complained about these missing dependencies before, and I don't think they're actually missing. I always update my system with urpmi --auto-select, so there certainly shouldn't be any unsatisfied dependencies. The problem doesn't come with unsatisfied dependencies actually present on your system, but about unsatisfied dependencies due to remove a package in order to upgrade some other, and broken dependencies are used to find additional package to remove. François.
Re: [Cooker] MySpell Bulgarian dependancies
Le jeu 30/01/2003 à 14:52, Robert Fox a écrit : Somehow MySpell Bulgarian got installed and when I try to remove it - it says: [root@foxbase rfox]# urpme myspell-bg_BG-1.0.1-0.20020716.5mdk To satisfy dependencies, the following packages are going to be removed (42 MB): OpenOffice.org-1.0.1-10mdk OpenOffice.org-l10n-en-1.0.1-10mdk mozilla-spellchecker-1.3-0.alpha.3mdk myspell-bg_BG-1.0.1-0.20020716.5mdk Is this OK? (Y/n) Huh? Problably because you got only this myspell-dictionary and removing it cause all requiring package to be removed too, install another myspell-dictionary and you will be allowed to remove it safely ? François.
Re: [Cooker] Re: [CHRPM] urpmi-4.2-14mdk
Le mar 28/01/2003 à 20:30, Tibor Pittich a écrit :
sorry, if this question is stupid, but i would ask to format of these
new options, especially limit-rate, if i want use it in global
urpmi.cfg. unfortunately documentation (man page) isn't updated with
this new features, but it is useful (therefore i'm requested it;)
) maybe not only for me.
i'm trying change urpmi.cfg like this:
{
limit-rate 10k or limit-rate : 10k
}
or add these variations into source definition, but always i get syntax
error.
if i try put arbitrary of these lines which i see in changelog:
{
verify-rpm : on|yes
verify-rpm
no-verify-rpm
}
there is no syntax error message.
is this problem in implementation (bug?) or only my configuration
mistake/ignorance ?
This is a mistake from me, an error message is displayed *BUT* the
option is taken into account by urpmi (and not by urpmi.update).
I fix the error message and the support in urpmi.update and
urpmi.addmedia.
François.
Re: [Cooker] urpmi is broken in latex cooker
Le ven 24/01/2003 à 23:05, Charles A Edwards a écrit : On Fri, 24 Jan 2003 22:43:50 +0100 Kim Schulz [EMAIL PROTECTED] wrote: What is wrong? Same here. urpmi-4.2-11 caused it. You can revert to -10 or wait for -12 and manually install it. Ok, this is a stupid typo from me when using curl, 12mdk fixes it. François.
Re: [Cooker] urpmi doesn't see what I have installed.
Le sam 25/01/2003 à 12:48, Han Boetes a écrit : ~% sudo urpmi freetype2-2.1.3-5mdk The following packages have to be removed for others to be upgraded: XFree86-devel-4.2.99.5-0.20030122.2mdk (due to unsatisfied fontconfig-devel = 2.1-4mdk) XFree86-static-libs-4.2.99.5-0.20030122.2mdk (due to unsatisfied XFree86-devel == 4.2.99.5-0.20030122.2mdk) freetype2-2.1.3-5plf (in order to install freetype2) freetype2-devel-2.1.3-5plf (due to unsatisfied freetype2 == 2.1.3-5plf) freetype2-static-devel-2.1.3-5plf (due to unsatisfied freetype2-devel == 2.1.3-5plf) libfontconfig1-devel-2.1-5mdk (due to unsatisfied freetype2-devel = 2.1.3-4mdk) libgdk-pixbuf2-devel-0.22.0-1mdk (due to missing gtk+-devel) libglade2.0_0-devel-2.0.1-3mdk (due to unsatisfied libgtk+2.0-devel = 2.0.0) libgtk+-x11-2.0_0-devel-2.2.0-3mdk (due to missing XFree86-devel) libgtk+1.2-devel-1.2.10-30mdk (due to missing XFree86-devel) libimlib1-devel-1.9.14-5mdk (due to missing libgtk+-devel) libpango1.0_0-devel-1.2.0-4mdk (due to unsatisfied freetype2-devel = 2.1.3-4mdk) do you agree ? (Y/n) n ~% rpm -q libfontconfig1-devel-2.1-5mdk libfontconfig1-devel-2.1-5mdk There must be a mistake here. I understand your problem, the message is badly formulated problably, if you look at the message more carefully, you will notice the following : freetype2-2.1.3-5plf (in order to install freetype2) which is caused to allow installation, after that it is only propagation of unsatisfied requires to the above recursively. So libfontconfig1-devel-2.1-5mdk will be removed, but you didn't agree so nothing has been removed. François.
Re: [Cooker] Re[2]: urpmi is still slow to check for valid medumon supermounted CD
Le ven 24/01/2003 à 07:04, Andrey Borzenkov a écrit : At this level, urpmi has workaround for supermount to avoid too much slowness, it first copy rpm in the cache and perform operation as usually directly from the cache in order to increase speed. You miss the point. I am aware of this problem and workarounf and _this_ problem is fixed in current kernel (in short, bad interaction between supermount and the rest of kernel caused long delays on file close - 4-5 seconds. So scanning large number of files was slow). I mean different problem (there was a bug report in Bugzilla about it but I forgot the number). Urpmi needs much time to check for correct media, _before_ it even started to copy or install files. I do not know what happens but I see that supermounted media is being accessed several times before urpmi finally asks to insert correct media. As usual, nothing fatal but bery annoying. There is a possible problem on this topic which will be very hard to fix, urpmi when a medium is available first check for CD to be correct by accessing all files before, this should be very fast normally but supermount may re-access the drive all the time. A delay may allow it to be faster on this. François.
[Cooker] Re: urpmi is still slow to check for valid medum on supermounted CD
Le jeu 23/01/2003 à 12:28, Andrey Borzenkov a écrit : Even with all current patches and with my patch @home that reduces number of failed mount attepmts to 1 (instead of current 6 :) for supermount it is still very slow when it checks for correct media in supermounted drive. Even for a single RPM it attempts to access CD several times that results in long delay. At this stage it is very unlikely to be supermount problem, besides it has never happened in urpmi 3.x so I do believe it is new urpmi logic. I tried to get a look, but my Perl knowledge is not up to Perl Obfuscated Programming Contest winner :)) Do you mean urpmi is obfuscated :-( At this level, urpmi has workaround for supermount to avoid too much slowness, it first copy rpm in the cache and perform operation as usually directly from the cache in order to increase speed. François.
Re: [Cooker] Prepare for the onslaught.
Le lun 20/01/2003 à 18:35, Guillaume Cottenceau a écrit : Ben Reser [EMAIL PROTECTED] writes: On Mon, Jan 13, 2003 at 03:50:46PM +0100, Daouda LO wrote: I thought we have a rpmdrake feature that diffs the rpmnew file with the current and asks for a merge? no? Yes but DrakX does not have this feature. And as far as I know rpmdrake is not supported for upgrading between versions. I think gc Well rpmdrake supports mostly what urpmi supports... and francois said that urpmi is now capable of upgrading between versions, isn't it? Yes, but not with too old version provided package are well done (no conflicts on file, not supported yet but will be added in a future release, maybe for 9.1). said urpmi would work but when asked if it was supported, no answer was given... And urpmi does not have such a feature. Hum.. For merging .rpmnew files ? François.
Re: [Cooker] Feature Request Dumping of installation mediacontents to HDD
Le mar 14/01/2003 à 02:19, Leon Brooks a écrit : Another feature I'd like to see is offering to copy the install CDs to disk (maybe in /var/install?), set them up as (a) URPMI source(s) in place of the removable media, and go hunting for a nearby update mirror (perhaps on first reboot). Users then don't need to fiddle to get updates working, or find CDs when they want to add software from the distro. You may want to do the following : mkdir /var/install mount /mnt/cdrom cp -a /mnt/cdrom/* /var/install umount /mnt/cdrom # for each additional cdrom... mount /mnt/cdrom cp -a /mnt/cdrom/Mandrake/RPMS* /var/install/Mandrake umount /mnt/cdrom # remove old entries urpmi.removemedia cdrom # add new ones urpmi.addmedia --distrib local file://var/install A reputation for staying up to date would help Mandrake. We are not up do date ? generally we are too much up to date ? François.
Re: [Cooker] urpmi --auto-select problem
Le ven 10/01/2003 à 13:33, marco a écrit : I can't find fileset mm 1.2.1. It just a problem of rebuilding package against mm 1.2.2. François.
Re: [Cooker] urpmi --auto-select problem
marco wrote: François Pons wrote: Le ven 10/01/2003 à 13:33, marco a écrit : I can't find fileset mm 1.2.1. It just a problem of rebuilding package against mm 1.2.2. What can I do now??? I could not find fileset mm 1.2.2 too. urpmq -f mm is your friend.
Re: [Cooker] urpme urpmi :(
Le mer 08/01/2003 à 04:27, Olivier Thauvin a écrit : Ok, this is stupid, but some beginners will do it one days: [olivier@andromede olivier]$ urpme urpmi Pour satisfaire les dépendances, les paquetages suivants vont être désinstallés (30 Mo): [snip] Est-ce correct ? (O/n) o [olivier@andromede olivier]$ urpmi sudo: /usr/sbin/urpmi: command not found [olivier@andromede olivier]$ Maybe it is better, if urpmi is on the list of packages to remove, ask somethings like: Are you sure you want to remove me? y/N And this should failed with --auto, this can be usefull on cooker with rebuild script which auto install or uninstall packages, to not see urpmi removing itself. Now, I'll rpm -Uvh urpmi ;) Well what do you think about rpm -rf /, do you want additional confirmation ? urpme urpmi will say urpmi will be deleted, the same for another urpme XXX wich will include urpmi on the list of package to remove. But anyway, urpmi will be listed to be removed, if someone don't read or don't understand what he is reading, he should not even have tried urpmi but should be using rpmdrake. Does anybody else have though on this ? François.
Re: [Cooker] More LDAP support in draktools?
Le lun 06/01/2003 à 18:23, Buchan Milne a écrit : First I must just say that over the holidays I read (or skimmed) most of the book Mastering Windows 2000, and learnt quite a bit, mostly that Windows 2000's support for Active Directory makes quite a few important features available, such as automatic software installation (based on membership of an LDAP OU and possible of a group also) for users (ie log in, you should have visio2002, and it is installed, someone else logs in who shouldn't have it, and it is removed) and for machines (on next user login, machines will install any new software assigned to their OU) and many other features (mostly collectively called Group Policy Objects). (Note that in AD, all computers joined to a domain have an account, with samba domain controllers this is also usually the case, except with the new optional nua - No User Account - sam backends in samba3). I was wondering if there were some applications that would apply to Mandrake. Some examples: 1)urpmi support for ldap, so that on every boot (and via cron?) machines would check which software they: a)Must have b)should not have This is the generic problem of distributed urpmi but with a lesser extend. It means if almost everyone login at the same time, there will be a lot of traffic downloading files ? This could be another tools (of urpmi suite) allowing such behaviour, for me it looks like better in that way. But I think adding or removing a software is very hard for the user, it removes a lot of freedom ? and automatically install/remove the software. This 2)urpmi support for configuring urpmi sources in ldap See above, assume you have a new application you want to roll out to all desktops, create a new urpmi source which as the app, add the package to list of required packages for the OU containing the machines, and go home. (yes, there is overlap with urpmi --parallel). Of course it overlaps, see above. It remove all the benefit of --parallel on bandwith. 3)msec support for ldap, so that security policies can be implemented per OU (including inheritance etc). Adding this to urpmi will imply adding this to msec as well ? And sorry for answering lately, François.
Re: [Cooker] urpme urpmi :(
Le mer 08/01/2003 à 12:51, Buchan Milne a écrit : What do this command, rpm -fr / ;))) I think fp meant rm -fr / That exactly the case, thanks. Related to my ldap mail regarding urpmi (require or conflict certain packages for ldap machine accounts), how about just doing this with a file, so that if any urpmi operation in --auto tries to remove a required rpm or install a conflicted rpm, that package will be dropped. /etc/urpmi/deps.list: Requires: urpmi Conflicts:apt (for instance) This sound like a very good idea, I will add Provides as well (so simulating more or less the skip.list differently). Would hopefully ensure that you will never lose urpmi, and never be able to install apt, except by confirming it when interactive. LDAP-enabling this in urpmi would be even cooler ... ;-) François.
Re: [Cooker] Please don't make urpmi stupid
Le mar 07/01/2003 à 05:22, Steve Fox a écrit : I would much prefer to see the tool remain smart. Really, why should it even care if unrelated dependencies are unresolved? It's kind of like getting into other people's business even though they didn't ask you to. (ok, maybe that's not the best analogy, but it's all I can think of) Ok, I hope I will not hurt sensibility but urpmi doesn't care if unrelated dependencies are unresolved. *BUT* urpmi when resolving related dependencies (and related should be understanded in the very large part) it may add unresolved dependencies. François.
Re: [Cooker] Very strange prob with xine-oss
Le lun 30/12/2002 à 23:13, Pierre a écrit : # urpmi xine-oss Pour satisfaire les dépendances, les paquetages suivants vont être installés (0 Mo): ldconfig-2.3.1-6mdk.i586 Est-ce correct ? (O/n) installation de /var/cache/urpmi/rpms/ldconfig-2.3.1-6mdk.i586.rpm Preparing... ## package ldconfig-2.3.1-6mdk is already installed L'installation a échoué If the problem is still present, please send me a bug report (use --bug of urpmi and send me a compressed archive of the directory created). Thanks, François.
Re: [Cooker] Urpmi fails since last update (was rpm + popt)
Le mar 24/12/2002 à 14:26, Pascal Terjan a écrit : Can't call method traverse on an undefined value at /usr/lib/perl5/vendor_perl/5.8.0/i386-linux-thread-multi/URPM/Resolve.pm line 714. It should be a problem of rpm as this error occurs if no database handle can have been opened. Anyway an more explicit report may be more usefull. François.
Re: [Cooker] Kernel change 'à chaud'
Le jeu 19/12/2002 à 17:14, Stéphane Teletchéa a écrit : You could imagine, and this could be a great improvement in cooker'developpment that you just switch from one kernel to another one. Yes, it is definitively funny. I iomagine this could be accomplished by freezing ALL the kernel features, letting only the old kernel control all the protocol. As an example, your kernel handles the etx3fs, freezes all operations on it, transfers all the features to the new kernel, and unfreezes it. And so on for all the needed features. Yes, it could be done for high availability update ? I think of this because of urpmi : With it, you can update your whole system, but the critical point is the kernel. With the procedure, all could be done automatically ;-)) No, you are wrong, the critical is not only the kernel, all running program are still in memory as well as all used libraries. Furthermore there could exist some problem with living application because they lost essential files which where not opened, this is still a somewhat critical environment which need a reboot (of course when updating only some packages this is a no do, but for upgrading from a whole distribution to another one, it is recommended to reboot even if the system is still useable). If you start new programs or still existing applications, newer applications and libraries are loaded in memory but the old are kept. Merry Xstmas !!! I hope you will still have a good Xstmas anyway ;-) François. PS : i'm not bugged nor drunk. PS : mi2
Re: [Cooker] urpmi bug.. package removal.
Le mar 10/12/2002 à 22:27, Jesper Krogh a écrit : I really think this is a bug: [root@luke root]# urpme XFree86 To satisfy dependencies, the following packages are going to be removed (43 MB): XFree86-ISO8859-2-Type1-fonts-1.0-16mdk drakconf-9.0-6.1mdk X11R6-contrib-4.2.1- 3mdk drakcronat-0.1.2-9mdk XFree86-100dpi-fonts-4.2.1-3mdk XFree86-4.2.1-3mdk xt erm-166-1mdk drakxtools-1.1.9-53.1mdk XFree86-75dpi-fonts-4.2.1-3mdk xinitrc-2.4 .4-69mdk harddrake-ui-1.1.9-53.1mdk XFree86-Xvfb-4.2.1-3mdk Is this OK? (Y/n) Y [root@luke root]# urpme XFree86 To satisfy dependencies, the following packages are going to be removed (43 MB): XFree86-ISO8859-2-Type1-fonts-1.0-16mdk drakconf-9.0-6.1mdk X11R6-contrib-4.2.1- 3mdk drakcronat-0.1.2-9mdk XFree86-100dpi-fonts-4.2.1-3mdk XFree86-4.2.1-3mdk xt erm-166-1mdk drakxtools-1.1.9-53.1mdk XFree86-75dpi-fonts-4.2.1-3mdk xinitrc-2.4 .4-69mdk harddrake-ui-1.1.9-53.1mdk XFree86-Xvfb-4.2.1-3mdk Is this OK? (Y/n) [root@luke root]# If it really means that it cannot remove the packages, then it should tell me instead of letting me know which other packages it need to remove. Or what's wrong since it cannot remove the packages. I really think that it is total leagal to remove XFree86 :-) Yes it is legal, of course. There should be a problem removing package and this is not visible by default (just found the bug and it will be fixed soon). François.
Re: [Cooker] urpmi bug.. package removal.
Le mer 11/12/2002 à 14:07, Jesper Krogh a écrit : Is it not possible to somewhere in this output insert a newline after each packagename so it becomes human-readable? Even better sorted by alphabet too This sounds like not to be a bad idea, so look at 13mdk of urpmi. François.
Re: [Cooker] urpmi bug.. package removal.
Le jeu 12/12/2002 à 02:55, Ben Reser a écrit : On Wed, Dec 11, 2002 at 02:07:45PM +0100, Jesper Krogh wrote: Is it not possible to somewhere in this output insert a newline after each packagename so it becomes human-readable? Even better sorted by alphabet too I've always wished for that too... Sorry, too late, this is done now. François.
Re: [Cooker] URPMI suggestion. ( /etc/urpmi/urpmi.cfg )
Le ven 06/12/2002 à 21:02, Jesper Krogh a écrit : If i remove: /var/lib/urpmi/hdlist.secure.cz And run: urpmi.update secure I get: unable to access hdlist file of secure, medium ignored But thats not true... it could just get it from the Ftp it knows. Yes, first when it reads the configuration file, it see no hdlist so the medium is ignored but urpmi.update ignored_medium_name allow to try to update an ignored medium (and drop the ignored flag). And a second suggestion. The possibility to have more than one server defining the same sources. So I wont miss if my local mirror reboots when my cronjob runs urpmi.update. It will be done, I don't known exactly now how to code this but it will be available (maybe even soon ?) But urpmi is really shaping up and looks very good. Thanks ! François.
Re: [Cooker] man urpmi (french)
Le dim 08/12/2002 à 13:04, pascal a écrit : The french men of urpmi of mdk9.0, need to be update Exemple: $ man urpmi .. DESCRIPTION .. Pour installer toutes les entrées de menu de tous les paquetages connus, utilisez autoirpm-icons. Thanks for pointing me this out, But do not worry too much, generally man pages are updated just before releasing a new distribution (well it could be earlier). François.
Re: [Cooker] urpmi rpm-cache
Le lun 09/12/2002 à 13:34, Steffen Barszus a écrit : Is it possible somehow to make the rpm-cache persistent ? I have a ISDN-Dialup and so I don't want to download things twice. Or can I change urpmi in that way ? If so can you guide me how to do so ? use it with --noclean (or --no-post-clean) François.
Re: [Cooker] R-base missing dependency
Le lun 09/12/2002 à 17:53, Ryan T. Sammartino a écrit : $ sudo /usr/sbin/urpmi --no-verify-rpm R-base installing /var/cache/urpmi/rpms/R-base-1.5.1-4mdk.i586.rpm Installation failed: libblas.so.3 is needed by R-base-1.5.1-4mdk $ sudo /usr/sbin/urpmi --no-verify-rpm libblas no package named libblas Assuming libblas.so.3 is provided by libblas package is very reductive, at least try with libblas.so.3 in order not to lose information. So where can I get libblas.so.3? $ urpmq -r libblas.so.3 libblas3-3.0-1mdk|liblapack3-3.0-7mdk but urpmi libblas.so.3 will do it the same way. François.
Re: [Cooker] R-base missing dependency
Le lun 09/12/2002 à 18:33, François Pons a écrit : Le lun 09/12/2002 à 17:53, Ryan T. Sammartino a écrit : $ sudo /usr/sbin/urpmi --no-verify-rpm R-base installing /var/cache/urpmi/rpms/R-base-1.5.1-4mdk.i586.rpm Installation failed: libblas.so.3 is needed by R-base-1.5.1-4mdk $ sudo /usr/sbin/urpmi --no-verify-rpm libblas no package named libblas Assuming libblas.so.3 is provided by libblas package is very reductive, at least try with libblas.so.3 in order not to lose information. So where can I get libblas.so.3? $ urpmq -r libblas.so.3 libblas3-3.0-1mdk|liblapack3-3.0-7mdk but urpmi libblas.so.3 will do it the same way. But of course, I'am stupid on this specific topic, install contrib package and it will do it (or update it). Package has been badly build (using contrib instead of only main) François.
Re: [Cooker] URPMI suggestion. ( /etc/urpmi/urpmi.cfg )
Le ven 06/12/2002 à 18:11, Jesper Krogh a écrit :
Hi. I have a suggestion.
The current layout of urpmi.cfg file is this way:
secure {
hdlist: hdlist.secure.cz
with_hdlist: ../base/hdlist.cz
list: list.secure
update
}
entities.
It would be wery nice if all relevant information was stored in the urpmi.cfg
file, so you could copy it to another system and do a:
urpmi.update -a
and it would retrieve the hdlists if they were accessible.
It is the case (url) now, except if a user/password is visible inside
it.
Due to the current layout of configuration files for urpmi, is it quite
impossible to view the configuration and verify that it is as I would expect
it to be. And even more impossible to copy a complete configuration to
another system ( if I don't have the exact urpmi.addmedia commands ).
Not enough simple ?
François.
Re: [Cooker] urpmi won't install mozilla
Le jeu 05/12/2002 à 19:46, Ryan T. Sammartino a écrit : Check out this transcript: $ sudo /usr/sbin/urpmi --no-verify-rpm mozilla To satisfy dependencies, the following packages are going to be installed (54 MB): galeon-1.3.0-2mdk.i586 libnspr4-1.2.1-1mdk.i586 libnspr4-devel-1.2.1-1mdk.i586 libnss3-1.2.1-1mdk.i586 mozilla-1.2.1-1mdk.i586 mozilla-devel-1.2.1-1mdk.i586 mozilla-irc-1.2.1-1mdk.i586 mozilla-mail-1.2.1-1mdk.i586 Is this OK? (Y/n) y % Total% Received % Xferd Average Speed Time Curr. Dload Upload TotalCurrent LeftSpeed 100 5341k 100 5341k0 0 90939 0 0:01:00 0:01:00 0:00:00 120k 100 1654k 100 1654k0 0 78562 0 0:00:21 0:00:21 0:00:00 75342 100 122k 100 122k0 0 31754 0 0:00:03 0:00:03 0:00:00 69162 100 10.1M 100 10.1M0 0 70210 0 0:02:30 0:02:30 0:00:00 69088 100 130k 100 130k0 0 33408 0 0:00:03 0:00:03 0:00:00 63150 100 187k 100 187k0 0 43123 0 0:00:04 0:00:04 0:00:00 75765 100 581k 100 581k0 0 64104 0 0:00:09 0:00:09 0:00:00 88226 everything already installed $ rpm -q mozilla mozilla-1.2-3mdk and urpmi ? perl-URPM ? Did you tested with last cooker ? François.
Re: [Cooker] urpmi gives an error
Le mer 04/12/2002 à 14:25, Kim Schulz a écrit : How do I solve this one? examining hdlist file [/var/cache/urpmi/partial/hdlist.norge1.cz] writing list file for medium norge1 examining synthesis file [/var/lib/urpmi/synthesis.hdlist.contrib.cz] unable to uncompress synthesis file at /usr/lib/perl5/vendor_perl/5.8.0/urpm.pm line 888. my version is: urpmi-4.1-8mdk C mal barré This is an error about reading synthesis file (which is gzip compressed), so try using gzip -dc on it and see what it is outputing. François.
Re: [Cooker] URPMI is Beautiful
Le ven 29/11/2002 à 18:29, nDiScReEt a écrit : I love that the download status now stays on one line instead of scrolling down showing each and every line status. I fear this is not a urpmi feature but wget problably... But anyway, urpmi is beautiful :-) François.
Re: [Cooker] urpmi error
Le dim 01/12/2002 à 22:24, gabor a écrit :
i have 2 sources defined in urpmi.cfg ( one for cooker, one for contrib)
called 'cz1' and 'cz2'
when doing urpmi --auto-select,
urpmi's output begins with:
medium cz2 trying to use an already used list, medium ignored..
?
my urpmi.cfg:
-
cz1 ftp://mandrake.redbox.cz/Mandrake-devel/cooker/cooker/Mandrake/RPMS
{
hdlist: hdlist.cz1.cz
with_hdlist: ../base/synthesis.hdlist.cz
synthesis
}
cz2 ftp://mandrake.redbox.cz/Mandrake-devel/contrib/i586 {
hdlist: hdlist.cz2.cz
with_hdlist: ../../cooker/cooker/Mandrake/base/synthesis.hdlist2.cz
synthesis
}
-
and there are other problems as well...
like:
urpmi.update -a -f begins with the same line:
medium cz2 trying to use an already used list, medium ignored
urpmi --media cz2 --auto-select
medium cz2 trying to use an already used list, medium ignored
Can't call method arch on an undefined value at
/usr/lib/perl5/vendor_perl/5.8.0/i386-linux-thread-multi/URPM/Resolve.pm
line 14.
any ideas what went wrong?
For all problems related to this, I made a mistaken (sorry, another one)
about list file usage (which has been removed for most of medium).
I fix it now, so next version will have more than 2 media useable.
Sorry again,
François.
Re: [Cooker] urpmi : bad install ordering for Scratch installation
Le lun 02/12/2002 à 13:03, BERNARD Sebastien a écrit : I'm trying to install a UML install using urpmi - no really différent than a chrooted installation. I wrote a small perl script which is doing a chrooted install of all packages. Doing something like : Rpm --root root path --initdb Then basicaly Urpmi --root root path the list of the basic packages I want My problem is that urpmi got the dependancy list wrong. The postinstall scripts are trying to use packages which are not installed yet. These are : sh-utils, diffutils, time, grep, gnupg, findutils, binutils, make, SysVInit, libsasl (libtermcap.so.2 not found ). Console-tools (cannot find add-help in rpm-helper installed after) Vim-minimal, gcc, gcc-cpp (/usr/bin/perl bad interpreter, in update-alternatives, perl is not yes installed). It is mandatory to include basesystem in the list of package, check you have done that (normaly requires should be good but... base package are always installed and could lead to missing dependencies ?) Maybe the dependency list is not right for certain packages. What gives urpmq --list | grep missing package to check repository is right. For example: perl-URPM does not need perl ??? It requires perl-base and not perl, that's normal (maybe a bad example). The last problem I got is that the kernel (which I do not need but I have to put it there to make the dependancy happy) is trying to execute lilo which fails miserably (of course). How can I disable this post-install script ? install kernel manually using --noscripts or create a false lilo package with a /sbin/lilo being /bin/true ? François.
Re: [Cooker] URPMI Weirdness
Le sam 30/11/2002 à 10:34, Sebastian Dransfeld a écrit : On Sat, 2002-11-30 at 06:09, Olivier Thauvin wrote: This is a know problem with your version, supposed to be fixed in -4mdk. I have same problem, but: [olivier@andromede urpmi.setup]$ rpm -q urpmi urpmi-4.1-5mdk I am up to date. :( Force an update 'urpmi.update -a -f' The forced update should work, however there could be case where it doesn't :-( If the bug persist, 6mdk is here for that (one of the fixes) François.
Re: [Cooker] Problems with urpmi / hdlist
Le dim 01/12/2002 à 11:10, Gilles Mocellin a écrit : Le sam 30/11/2002 à 12:42, Gilles Mocellin a écrit : I can't install new packages with urpmi. It seems that the package name in hdlist is the full path to the rpm. Directly with a rpm command, I can upgrade. It seems that urpmi.cfg format has changed. No, but some freedom are now tolerated ;-) I removed and re-added all my sources. But now, it can't/don't build the list.* files for main and contrib sources. It always mark some sources has ignore and can't find new packages. With --synthesis, it finds new packages but can't download them (missing list.* ?) [root@guitare urpmi]# urpmi --auto-select -v --synthesis examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.main.cz] examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.contrib.cz] examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.plf.cz] examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.main2.cz] examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.contrib2.cz] examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.plf2.cz] examen de la liste de synthèse [/var/lib/urpmi/synthesis.hdlist.jpackage.free.cz] examen de la liste de synthèse If you still have empty line for synthesis examination, please told me (that's absolutely not normal). [/var/lib/urpmi/synthesis.hdlist.jpackage.nonfree.cz] Pour satisfaire les dépendances, les paquetages suivants vont être installés (3 Mo): fluxbox-0.1.13-7mdk.i586 perl-DBI-1.31-1mdk.i586 wmcoincoin-2.4.2-0.a.1mdk.i586 Est-ce correct ? (O/n) paquetage wmcoincoin-2.4.2-0.a.1mdk.i586 non trouvé paquetage perl-DBI-1.31-1mdk.i586 non trouvé paquetage fluxbox-0.1.13-7mdk.i586 non trouvé impossible de récupérer les paquetages sources, abandon [root@guitare urpmi]# This one should be fixed as well as previous case (for --auto). François.
Re: [Cooker] Re: Proposition for rpmdrake
Le ven 29/11/2002 à 11:03, daniel beck a écrit : oh, and an other nice feature for urpmi that would be (rember the libpng2 to libpng3 transition) . Every time i install a new rpm, all the files which are overwritten are stored in /var/cache/urpmi/temp or something like that. so I could do so : there is already A2.0 urpmi --safe A2.1.rpm urpmi overwrites some files from A2.0 but store them in /var/cache/urpmi/temp the programm A2.1 brings everyting so I want to have my old configuration again. so I do : urpmi --recover and everything is just like befor the update of A2.0. That would give really a safe impression to the user ! You can do urpmi A2.0 (if it still exists of course) and you get back the package... François.
Re: [Cooker] Problem with rsync urpmi method
Le ven 29/11/2002 à 08:51, Chris Picton a écrit : I've just tried the rsync method with urpmi and got the following results: Ok, I check here and try to fix. François.
Re: [Cooker] Propositions for rpmdrake
Le mer 27/11/2002 à 08:57, Jesper Krogh a écrit : I miss a small feature too: If I'd like to install a rpm from say: http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm I can do this: rpm -ivh http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm And rpm will download it and install it, but if it has some dependencises it wil fail and i'll have to do this: wget http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm urpmi ./filename.mdk.rpm The smart soloution would be to be able to do this: urpmi http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm I will do this one and it is not complicated and very usefull. François.
Re: [Cooker] Re: Propositions for rpmdrake
Le ven 29/11/2002 à 11:45, Tarmo Hyvärinen a écrit : Can we also have a '--update' flag for urpmi.update ? So then: # urpmi.update --update; urpmi --auto-select --auto --update would only update hdlists for updates? urpmi.update update_source;urpmi --auto-select --auto does the trick for updates, which automatically takes care of them. I have this timed too, very useful, always up-to-date system The problem occurs when you have a lot of updates (wich will the case now as you can register multiple updates media and urpmi will fecth packages potentially from all of them in order to avoid problem from down site or not up-to-date mirror. So features is in urpmi-4.1-3mdk in cooker since yesterday along with mirrors management. François.
Re: [Cooker] Re: [CHRPM] urpmi-4.1-3mdk
Le ven 29/11/2002 à 13:27, Charles A Edwards a écrit : [root@localhost charles]# urpmi --curl --auto-select The following packages have to be removed for others to be upgraded: galeon-1.2.6-2mdk do you agree ? (Y/n) y To satisfy dependencies, the following packages are going to be installed (23 MB): bonobo-activation-2.1.1-1mdk.i586 eog-1.1.2-2mdk.i586 gnome-games-2.1.3-1mdk.i586 gnome-system-monitor-2.0.3-1mdk.i586 gnome-vfs2-2.1.3.1-1mdk.i586 libbonobo-activation4-2.1.1-1mdk.i586 libbonobo-activation4-devel-2.1.1-1mdk.i586 libbonoboui-2.1.1-1mdk.i586 libbonoboui2_0-2.1.1-1mdk.i586 libbonoboui2_0-devel-2.1.1-1mdk.i586 libdvdread2-0.9.3-3mdk.i586 libdvdread2-devel-0.9.3-3mdk.i586 libgdkcardimage0-2.1.3-1mdk.i586 libgnome-vfs2_0-2.1.3.1-1mdk.i586 libgnome-vfs2_0-devel-2.1.3.1-1mdk.i586 libgnomeui2-2.1.3-1mdk.i586 libgnomeui2_0-2.1.3-1mdk.i586 libgnomeui2_0-devel-2.1.3-1mdk.i586 Is this OK? (Y/n) y package gnome-vfs2-2.1.3.1-1mdk.i586 is not found. package libdvdread2-devel-0.9.3-3mdk.i586 is not found. package libgnomeui2_0-2.1.3-1mdk.i586 is not found. package libgnome-vfs2_0-devel-2.1.3.1-1mdk.i586 is not found. package libdvdread2-0.9.3-3mdk.i586 is not found. package libgnomeui2_0-devel-2.1.3-1mdk.i586 is not found. package gnome-system-monitor-2.0.3-1mdk.i586 is not found. package libgnomeui2-2.1.3-1mdk.i586 is not found. package libgnome-vfs2_0-2.1.3.1-1mdk.i586 is not found. package eog-1.1.2-2mdk.i586 is not found. package libbonoboui-2.1.1-1mdk.i586 is not found. package libgdkcardimage0-2.1.3-1mdk.i586 is not found. package gnome-games-2.1.3-1mdk.i586 is not found. unable to get source packages, aborting rpm -qa |grep urpmi urpmi-4.1-3mdk gurpmi-4.1-3mdk Sorry, stupid typo from me. fixed in 4mdk (along with current features added). François.
Re: [Cooker] Propositions for rpmdrake
Le mer 27/11/2002 à 18:08, Gilles Mocellin a écrit : On Wed, Nov 27, 2002 at 08:57:29AM +0100, Jesper Krogh wrote: On Wednesday 27 November 2002 04:17, daniel beck wrote: Nice, featurelist above. I miss a small feature too: If I'd like to install a rpm from say: http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm I can do this: rpm -ivh http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm And rpm will download it and install it, but if it has some dependencises it wil fail and i'll have to do this: wget http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm urpmi ./filename.mdk.rpm The smart soloution would be to be able to do this: urpmi http://testserver.com/path/to/smart-mandrake-stuff/filename.mdk.rpm And it would install the file solving the dependencies as usual. Done in 4mdk And my 2c : It would be nice if urpmi supports multiple sources for the same packages, and tries each in case of network failure, package not here... transfert rate too low... Parallelization of package chunks (like prozilla)... Already done in 3mdk (which is buggy so use 4mdk) François.
Re: [Cooker] Rpmdrake - Everything Allready Installed - Error
Le ven 29/11/2002 à 15:46, Simon Ree a écrit : Installed urmpi/gurpmi-4.1-4mdk and now rpmdrake says 'Everything allready installed(is this supposed to happen at all?)' when I try to install the rest of the available updates. Tried urmpi from the command line to see what it says: [root@localhost simon]# urpmi --auto everything already installed [root@localhost simon]# urpmi --auto-select The following packages have to be removed for others to be upgraded: galeon-1.2.6-2mdk do you agree ? (Y/n) y To satisfy dependencies, the following packages are going to be installed (42 MB): akpi-0.3.3-2mdk.i586 eog-1.1.2-2mdk.i586 gnome-desktop-2.1.3-1mdk.i586 gnome-icon-theme-0.1.3-1mdk.noarch gnome-system-monitor-2.0.3-1mdk.i586 gnome-vfs2-2.1.3.1-1mdk.i586 ispell-3.2.06-8mdk.i586 ispell-en-3.2.06-8mdk.i586 libbonoboui-2.1.1-1mdk.i586 libdvdread2-0.9.3-3mdk.i586 libgnome-desktop-2_2-2.1.3-1mdk.i586 libgnome-vfs2_0-2.1.3.1-1mdk.i586 libgnomeui2-2.1.3-1mdk.i586 libgnomeui2_0-2.1.3-1mdk.i586 libgsf-1_1-1.5.0-1mdk.i586 libmrproject-0.7-1mdk.i586 libmrproject0-0.7-1mdk.i586 libnspr4-1.2-2mdk.i586 libnss3-1.2-2mdk.i586 makedev-4.0-2mdk.noarch mozilla-1.2-2mdk.i586 mozilla-irc-1.2-2mdk.i586 mozilla-mail-1.2-2mdk.i586 shorewall-1.3.11-1mdk.noarch Is this OK? (Y/n) y package gnome-vfs2-2.1.3.1-1mdk.i586 is not found. package ispell-en-3.2.06-8mdk.i586 is not found. package mozilla-mail-1.2-2mdk.i586 is not found. package libmrproject0-0.7-1mdk.i586 is not found. package libmrproject-0.7-1mdk.i586 is not found. package akpi-0.3.3-2mdk.i586 is not found. package libgnomeui2_0-2.1.3-1mdk.i586 is not found. package gnome-icon-theme-0.1.3-1mdk.noarch is not found. package gnome-desktop-2.1.3-1mdk.i586 is not found. package libgsf-1_1-1.5.0-1mdk.i586 is not found. package libdvdread2-0.9.3-3mdk.i586 is not found. package ispell-3.2.06-8mdk.i586 is not found. package gnome-system-monitor-2.0.3-1mdk.i586 is not found. package libgnomeui2-2.1.3-1mdk.i586 is not found. package libgnome-vfs2_0-2.1.3.1-1mdk.i586 is not found. package libgnome-desktop-2_2-2.1.3-1mdk.i586 is not found. package eog-1.1.2-2mdk.i586 is not found. package mozilla-1.2-2mdk.i586 is not found. package libbonoboui-2.1.1-1mdk.i586 is not found. package mozilla-irc-1.2-2mdk.i586 is not found. package libnspr4-1.2-2mdk.i586 is not found. package libnss3-1.2-2mdk.i586 is not found. package shorewall-1.3.11-1mdk.noarch is not found. unable to get source packages, aborting [root@localhost simon]# Sorry 4.1-3mdk introduced a bug, to fix it using 4mdk, please urpmi.update -a before... Frnaçois.
Re: [Cooker] Re: Propositions for rpmdrake
Le jeu 28/11/2002 à 12:22, Lea Gris a écrit : Whth urpmi you have to update entire filelists everytime you want to upgrade/download from an internet source because otherwise you get pointers to unavailables packages. Yes, maybe it can be automated but urpmi.update -a urpmi --auto-select --auto is not quite hard to write if any you write it. with apt-get update you don't have to. Packages are up to date everytime. I don't know precisely how it work but you don't have to update entire filelists everytime. Filelists represent like 25MB of data on a cooker + contribs this is quite huge to download whan you just like to update one or two packages or just want one new installed. $ wc -c /var/lib/urpmi/synthesis.hdlist.cooker.cz 142518 /var/lib/urpmi/synthesis.hdlist.cooker.cz Is it too large ? It is less than 25MB anyway. Of course if you want to use urpmf, it is no more usable but it is enough for installing packages (even with --auto-select) maybe Mandrakesoft can improove urpmi in some way in that direction. Currently the last improvement are use another mirror for downloading file (if some site gets unreachable), allow not creating list file (more an internal improvement) and some bug fixes. François.
Re: [Cooker] Re: Propositions for rpmdrake
Le jeu 28/11/2002 à 14:40, Buchan Milne a écrit : Can we also have a '--update' flag for urpmi.update ? So then: # urpmi.update --update; urpmi --auto-select --auto --update would only update hdlists for updates? Why not, wait some minutes. François.
Re: Urpmi improvment request WAS : Re: [Cooker] Re: Propositionsfor rpmdrake
Le jeu 28/11/2002 à 13:53, Lea Gris a écrit : With urpmi you *have* to download entire filelists everytime you want to download/upgrade package from constantly upgrading mirrors because serverside new or updated packages don't magically go to your local filelist. You have to get an updated one everytime. I may not have been clear enough, you may want to download synthesis file instead of hdlist one, but you will lack urpmf support. But synthesis file size is approximately 140 KB instead of 22.5 MB to be precise on hdlist size you have. A lot of other cooker contributors have already posted these remarks yet. François.
Re: [Cooker] Re: Propositions for rpmdrake
Le jeu 28/11/2002 à 14:10, Lea Gris a écrit : Adam Williamson wrote: Um, that's only the complete hdlist, which you only need to download if you want package descriptions. Otherwise you can use the synthesis files, synthesis.hdlist.cz (for Cooker) and synthesis.hdlist2.cz (for contribs), which are 137KB and 105KB respectively at the moment - hardly anything. Outch, sorry ! Din't know I could use these ones :) I used to do this : urpmi.addmedia --distrib ftp://mynearestmirror.com/pub/linux/distributions/mandrake-devel/cooker/cooker/ And that one reference the huge hdlist.cz files Yes, sorry from me for that part, but cdrom or local mirror (or copy) of cdrom use a lot of hdlist (so a lot of media) which is complicated by hand for each of them, and using a local or cdrom device, it is problably easier to get the full hdlist support (for urpmf). Adding cooker directly using network interface was at least the same of adding cooker or contrib by hand, and some people don't want contrib but only cooker so --distrib is *always* defaulting to hdlist instead of synthesis, whereas -h flag is prefering the synthesis all the time instead of hdlist (since -h may be more usefull with network access ?). A shortest way to add entire distrib download source with synthesis.hdlist files would help by the way. So I will add a flag to prefer synthesis or hdlist... François.
Re: [Cooker] Re: Re: Propositions for rpmdrake
Le jeu 28/11/2002 à 16:45, Stephen Pickering a écrit : I'd like rpmdrake urpmi to have an option to list updates that are over say 'n' days old, a command line switch for urpmi, eg: urpmi --available_for 5 I've found that sometimes an rpm is released and it screws the system, a few hours or days later a new rpm is released which cures the problem. Yeah, I would like this. I would use it for production releases as well, not just keeping a bit of distance from bleeding edge Cooker. The real reson for my post, ignore the rest of my email, if this feature is added I'm happy :-) Hum, it could catch titi sucking, yes it is a great feature ;-) François.
Re: [Cooker] Re: Urpmi improvment request
Le jeu 28/11/2002 à 22:09, Steffen Barszus a écrit : A full ACK on that the hdlist.cz should be updated trough diffs, however this is realized This sounds to me like a reasonable solution. For hdlist.cz file, rsync (or diff ?) could do it, but for synthesis it sounds diffcult to be efficient, the file is small and fully compressed whereas hdlist have only compression by parts. Anyway there should already be a rsync protocol for urpmi but never tested, if someone want to play with it. François.
Re: [Cooker] Re: Urpmi improvment request
Le jeu 28/11/2002 à 22:52, Gilles Mocellin a écrit : What french mirror supports rsync ? For instance to play with it I suppose you have to create an rsync mirror, then create your medium using rsync://... François.
Re: [Cooker] amavis: problem with loading text extractor :perl-Convert-UUlib
Le jeu 21/11/2002 à 13:07, Pascal Cavy a écrit : this path /usr/lib/perl5/vendor_perl/5.8.0/i386-linux/ is not in the standard path. [...] So what package is to be fixed ?? perl-Convert-UUlib which has been compiled with an older perl without multi-thread support. just compiling it again will problably fixes it. François.
[Cooker] Re: rpmdrake - handling of errors during installation
Borsenkow Andrej [EMAIL PROTECTED] writes: One of annoying features of rpmdrake is that it completely ignores errors during RPM installation phase - i.e. such things like bad RPM or missing depndencies and removes all packages from list assuming they have been installed. Unfortunately it turned out not to be easy to fix. grpmi calls rpmRunTransactions and relies on it to return a list of problems during install. Now rpm does not always returns something meaningful. It does it in such cases like out of space or missing dependencies but does not if you have corrupted RPM or in case of *any* other error during actual installation. To test just truncate any RPM and see - you get just overall installation failed status without any more detailed information which RPM failed. What could be done is to examine what has been installed, and to update according to what should have been installed. François.
Re: [Cooker] Re: [CHRPM] urpmi-3.2-7mdk
Charles A Edwards [EMAIL PROTECTED] writes: After updating, with supermount enabled mkbootdisk is again giving error 256- mount: /dev/fd0 already mounted or /tmp/mkbootdisk busy For upgrading mkbootdisk using urpmi or using mkbootdisk ? François.
Re: [Cooker] feature request for rpmdrake
[EMAIL PROTECTED] writes: I have a couple of feature requests for rpmdrake: 1) if it errors out, tell me why (i.e. max. anon ftp reached, let me know instead of just due to some error) 2) depends-on and is-depended-on tabs. I would like to know what package depends on a particular package and vice versa. I know there are command line tools for this already. I would like to see it in rpmdrake. Yes, such features will be nice and may be coded but for a next release. François.
Re: [Cooker] lilo Fatal: First boot sector is version 21.7. Expecting version 22.1.
Gregoire Favre [EMAIL PROTECTED] writes: if that is the case, re-install lilo cleanly (eg: rpm -i lilo-22*.rpm --force) and it should work. No: rpm -i lilo-22.1-4mdk.i586.rpm --force lilo Fatal: First boot sector is version 21.7. Expecting version 22.1. rpm -e lilo and then rpm -i gives same result??? Look at /boot/lilo, where does it points ? Do you have modified it by hand ? (possible values are currently lilo-text, lilo-graphic, lilo-menu, lilo-bmp). It should point to a directory containing the real boot.b for the right stage2 lilo, after that /boot/boot.b should point to lilo/boot.b (always the case, previously the choice was made changing /boot/boot.b (to /boot/boot-menu.b for example) but now to use different message file suitable for the used stage2 lilo, only /boot/lilo is a symlink suggested to change). François.
Re: [Cooker] lilo Fatal: First boot sector is version 21.7. Expecting version 22.1.
Neuromancer [EMAIL PROTECTED] writes: Thus spake François Pons about Life, the Universe, and Everything: | Look at /boot/lilo, where does it points ? Do you have modified it by hand ? | (possible values are currently lilo-text, lilo-graphic, lilo-menu, lilo-bmp). Out of curiosity, what is lilo-bmp? A new extension (official) from lilo authors to support 640x480 bmp file in 16 colors. Look at README.bitmaps in lilo documentation for more details. But really, don't you think 640x480 (or even 800x600) in 256 colors is not better ? François.
Re: [Cooker] perl mysql conflicts
[EMAIL PROTECTED] writes: while trying to install Msql-Mysql-modules-1.2216-1mdk.i586.rpm: This means Msql-Mysql-modules in contrib should be removed, as main has a better version. Thanks for pointing us that, François.
[Cooker] Re: urpmi loops on RPM not found - please, fix it finally.
Borsenkow Andrej [EMAIL PROTECTED] writes:
Now, when urpmi is almost perfect, it is even more annoying.
What the problem it tries to access your hard disk all the time (I suppose this
is a zip or jaz (since you mount it on /mnt/hd)).
{pts/1}% LANGUAGE=en ll /mnt/hd/cooker/i586/Mandrake
ls: cannot read symbolic link /mnt/hd/cooker/i586/Mandrake/RPMS.cooker:
Invalid
argument
ls: cannot read symbolic link /mnt/hd/cooker/i586/Mandrake/RPMS2:
Invalid argument
total 108
drwxr-xr-x2 500 ntools 4096 ñÎ× 29 09:20 base/
drwxr-xr-x5 500 ntools 4096 ñÎ× 28 23:19 mdkinst/
drwxr-xr-x2 500 ntools 94208 ñÎ× 29 09:18 RPMS/
lrwxrwxrwx1 500 ntools 22 ñÎ× 29 17:07 RPMS2
lrwxrwxrwx1 500 ntools 4 óÅÎ 8 14:28 RPMS.cooker
drwxr-xr-x3 500 ntools 4096 ñÎ× 18 14:06 share/
Or you have a problem on your drive ? Do you use supermount ?
François.
If RPM is not there, it is not there.
Yes, but why do you have error using ls ?
François.
Re: [Cooker] RE: urpmi loops on RPM not found - please, fix it finally.
Borsenkow Andrej [EMAIL PROTECTED] writes: Urpmi checks for correct media by looking at directory. It sees RPMS2 and assumes it is second CD. From now on it blindly tries to fetch non-existent RPM. Francois, it is irrelevant why ls returns an error. There may be thousands reasons why your media gets corrupted. But uprmi must not loop indefinitely. If iy cannot find required RPM it must tell me so and either quit or go on. This is what happens if you have file:// media and does not work with removable:// media. I check this point and try to make it clever. François.
Re: [Cooker] RE: urpmi loops on RPM not found - please, fix it finally.
[EMAIL PROTECTED] (François Pons) writes: Borsenkow Andrej [EMAIL PROTECTED] writes: Urpmi checks for correct media by looking at directory. It sees RPMS2 and assumes it is second CD. From now on it blindly tries to fetch non-existent RPM. Francois, it is irrelevant why ls returns an error. There may be thousands reasons why your media gets corrupted. But uprmi must not loop indefinitely. If iy cannot find required RPM it must tell me so and either quit or go on. This is what happens if you have file:// media and does not work with removable:// media. I check this point and try to make it clever. Does urpmi ask you to change disk or not ? or does it loop forever ? François.
Re: [Cooker] RE: urpmi loops on RPM not found - please, fix it finally.
Borsenkow Andrej [EMAIL PROTECTED] writes: Does urpmi ask you to change disk or not ? No. Ok, I see the problem, fixes will be on next urpmi release. or does it loop forever ? Yes. This is clear that way. François.
Re: [Cooker] people about new boot screen (lilo?)
Borsenkow Andrej [EMAIL PROTECTED] writes: And reply to this is even more amusing :-) === Yea, we don't want no slick and shiny boot screens. After all this is the OS for true men (and women), people who have witnessed the rise of the transistor. Or maybe this is a nasty trick from the marketing department to get at big company buyers, reminding them of those happy drunken nights in front of a 64K machine when one had time to kill a sixpack while the soft loaded from tapes. Have you noticed that mdk is so fond of that screen that they extended default boot timeout to 10s? Our graphist will appreciate :-) If you have very complex pictures to display, you can use latest lilo which contains a quite small pieces of SuSe patch to extend message file to 512Kb. Enjoy, François.
Re: [Cooker] Re: [CHRPM] Mesa-4.0.1-3mdk
Charles A Edwards [EMAIL PROTECTED] writes: It now appears the poor performance of Radeon vid card was caused by old Mesa and not by XFree. Am happy to report that after install of the Mesa-4.0.1 rpms performance of Radeon 32MB DDR shot from 100fbs to 700fbs. Sorry, since i810 is a bit broken in 4.2, I didn't have checked Radeon performance, but Mesa contains a tips that if libGL.so.1.3 exist, it assume it is a new XFree style libGL.so.1 and use it for rendering instead of libGL.so.1.2, so poor performance. Another way to set GL_WRAPPER_XF4_GLX to change default behaviour. Francois be praised. François :-)
Re: [Cooker] Re: [CHRPM] urpmi-3.2-6mdk
Charles A Edwards [EMAIL PROTECTED] writes: Though it may or may not have been intentional this update also resolved the issue of installing rpms from a removable source in rpmdrake as well as rectifing the supermount mkbookdisk 256 error. :-) again. François.
