December 2006 - Pacman-dev - lists.archlinux.org

[pacman-dev] pacman-optimize updates
by Dan McGee 28 Dec '06

28 Dec '06

pacman-optimize updates. Bug #5009 is out there, I'll think about it and try to make a patch hopefully without screwing up my own database. -Dan Index: pacman-optimize =================================================================== RCS file: /home/cvs-pacman/pacman-lib/scripts/pacman-optimize,v retrieving revision 1.3 diff -u -r1.3 pacman-optimize --- pacman-optimize 21 Dec 2006 01:53:41 -0000 1.3 +++ pacman-optimize 22 Dec 2006 19:37:41 -0000 @@ -20,7 +20,7 @@ # USA. # -myver='2.9.8' +myver='3.0.0' usage() { echo "pacman-optimize $myver" @@ -59,10 +59,6 @@ dbroot=$1 fi -if [ "`id -u`" != 0 ]; then - die "You must be root to optimize the database" -fi - # make sure pacman isn't running if [ -f /tmp/pacman.lck ]; then die "Pacman lockfile was found. Cannot run while pacman is running." @@ -72,6 +68,11 @@ die "$dbroot does not exist or is not a directory" fi +#if [ "$EUID" != 0 ]; then +if [ ! -w "$dbroot" ]; then + die "You must have correct permissions to optimize the database" +fi + # don't let pacman run while we do this touch /tmp/pacman.lck

2 2

[pacman-dev] New mirrors file
by Dan McGee 28 Dec '06

28 Dec '06

Saw this (http://bugs.archlinux.org/task/6083) bug today on the bugtracker and also knew this was a problem anyway, so here is my go at it. I tried to make a more up to date mirror list from both the Wiki and the download page. I ran the rankmirrors script on it a few times to hopefully weed out the old/bad mirrors. If someone that is skilled with automake/autoconf wants to finish the job, I'd appreciate it. Instead of having four more or less duplicate files in our repository, we could generate all of them from this one file. -Dan /etc/pacman.d/mirrorlist.in: # # @REPO@: Arch Linux @REPO@ repository # # United States Server = ftp://ftp.archlinux.org/@REPO@/os/@CARCH@ Server = ftp://ftp.nethat.com/pub/linux/archlinux/@REPO@/os/@CARCH@ Server = ftp://ftp-linux.cc.gatech.edu/pub/linux/distributions/archlinux/@REPO@/os/@… Server = ftp://mirror.cs.vt.edu/pub/ArchLinux/@REPO@/os/@CARCH@ Server = ftp://ftp.ibiblio.org/pub/linux/distributions/archlinux/@REPO@/os/@CARCH@ Server = http://www2.cddc.vt.edu/linux/distributions/archlinux/@REPO@/os/@CARCH@ # Europe # - Austria Server = ftp://gd.tuwien.ac.at/opsys/linux/archlinux/@REPO@/os/@CARCH@ # - Belgium Server = ftp://ftp.belnet.be/packages/archlinux/@REPO@/os/@CARCH@ # - Czech Republic Server = ftp://ftp.sh.cvut.cz/MIRRORS/arch/@REPO@/os/@CARCH@ # - Estonia Server = ftp://ftp.estpak.ee/pub/archlinux/@REPO@/os/@CARCH@ # - France Server = ftp://ftp.rez-gif.supelec.fr/pub/Linux/distrib/archlinux/@REPO@/os/@CARCH@ Server = ftp://mir1.archlinuxfr.org/archlinux/@REPO@/os/@CARCH@ Server = ftp://mir2.archlinuxfr.org/archlinux/@REPO@/os/@CARCH@ Server = ftp://distrib-coffee.ipsl.jussieu.fr/pub/linux/archlinux/@REPO@/os/@CARCH@ #Server = http://mirrors.jakimowicz.com/archlinux/@REPO@/os/@CARCH@ # - Germany #Server = ftp://ftp.archlinux.de/pub/archlinux/@REPO@/os/@CARCH@ Server = ftp://ftp.tu-chemnitz.de/pub/linux/sunsite.unc-mirror/distributions/archlinux/@REPO@/os/@CARCH@ Server = ftp://ftp.hosteurope.de/mirror/ftp.archlinux.org/@REPO@/os/@CARCH@ # - Great Britain Server = http://ftp.parrswood.manchester.sch.uk/Mirrors/ftp.archlinux.org/@REPO@/os/… # - Greece Server = ftp://ftp.ntua.gr/pub/linux/archlinux/@REPO@/os/@CARCH@ # - Ireland Server = ftp://ftp.heanet.ie/mirrors/ftp.archlinux.org/@REPO@/os/@CARCH@ # - Italy Server = ftp://mi.mirror.garr.it/mirrors/archlinux/@REPO@/os/@CARCH@ # - Lithuania Server = ftp://saule.mintis.lt/pub/linux/@REPO@/os/@CARCH@ # - Netherlands Server = ftp://ftp.nluug.nl/pub/metalab/distributions/archlinux/@REPO@/os/@CARCH@ Server = ftp://ftp.surfnet.nl/pub/os/Linux/distr/archlinux/@REPO@/os/@CARCH@ # - Norway Server = ftp://mirror.djzradio.net/archlinux/@REPO@/os/@CARCH@ # - Poland Server = ftp://sunsite.icm.edu.pl/site/linux-sunsite/distributions/archlinux/@REPO@/os/@CARCH@ Server = ftp://mirror.icis.pcz.pl/archlinux/@REPO@/os/@CARCH@ # - Portugal Server = ftp://gul.est.ips.pt/repos/archlinux/@REPO@/os/@CARCH@ Server = ftp://cesium.di.uminho.pt/pub/archlinux/@REPO@/os/@CARCH@ Server = ftp://darkstar.ist.utl.pt/pub/archlinux/@REPO@/os/@CARCH@ # - Romania Server = ftp://ftp.iasi.roedu.net/mirrors/archlinux.org/@REPO@/os/@CARCH@ # - Sweden Server = ftp://ftp.ds.hj.se/pub/os/linux/archlinux/@REPO@/os/@CARCH@ Server = ftp://ftp.gigabit.nu/@REPO@/os/@CARCH@ # - Switzerland Server = ftp://archlinux.puzzle.ch/@REPO@/os/@CARCH@ # Australia Server = ftp://mirror.pacific.net.au/linux/archlinux/@REPO@/os/@CARCH@ # Asia # - China Server = http://mirror.vxmatrix.net/ArchLinux/@REPO@/os/@CARCH@ Server = http://dxmirror.vxmatrix.net/ArchLinux/@REPO@/os/@CARCH@ Server = http://cncmirror.vxmatrix.net/ArchLinux/@REPO@/os/@CARCH@

4 4

[pacman-dev] [PATCH] conditional build of pacman.static
by Jürgen Hötzel 28 Dec '06

28 Dec '06

Hi, #1 from TODO.autoconf Jürgen

2 1

[pacman-dev] My $0.02 on the Frugalware v.s. Arch bit...
by Alex Smith 27 Dec '06

27 Dec '06

Hi all, I hope you all don't mind me writing this - I'm going to try to write it in a way that seems neutral (i.e. not on the Arch side or the Frugalware side), even though I'm from the Frugal side of things. And just in case you're wondering why I'm writing this here instead of telling you, VMiklos, I think this implies to both sides, therefore belongs here. Anyway, I'm going to start by saying bluntly what I think: This is just pathetic bickering. For goodness' sake, it's a package manager, not the world's most sacred thing. Why do I say this? Let's start with the MaxTries patch. Aaron said that it was "not valid". I've searched the archives of this list, and I can't find any reason for this. Yet, I've found reason for this here: http://bbs.archlinux.org/viewtopic.php?p=216399#216399 It would have been far more appropriate to say that here, then the problem with this patch could have been rectified. However, here you just said "not valid". Now, about our "patch queue". This wasn't a seperate codebase, it was simply a repo for us to develop our patches in and find/fix bugs before submitting to upstream. It was never a fork - we simply never had upstream commit access, so we had to do this. Personally, I don't think we wanted to fork, however it's hard to work with someone who says stuff is not valid without giving reason. But really, wouldn't it be so much easier if this was the case: 1. VMiklos creates the MaxTries patch, discusses it here. 2. Other people point out any problems/flaws with it, in a relaxed manner, rather than having a shouting match with 20 foot flames all around. 3. Problems get sorted after discussion, it gets committed. The main point to that is the relaxed manner. I'll repeat what I said at the beginning: this is a package manager, a small application, not something so important that the world will blow up if something goes wrong. I really think that we should merge back into 1 project, rather than leading 2 seperate paths with 2 different APIs, and co-operate better, have fun coding, instead of this bickering. Having 2 APIs would just suck - someone could come along and create a really great libalpm frontend that works on Frugalware, and then the Arch users will be disappointed when it won't even compile for them. We've cooperated up until now in relaxed way, why change it. But anyway, as I said in the topic, this is just my $0.02, so, choose to ignore it if you wish, but I seriously think that a relaxed attitude is the key here. Thanks, Alex ------ Alex Smith, Frugalware Linux developer - http://www.frugalware.org

2 1

[pacman-dev] IDEA: checkdb script
by Roman Kyrylych 24 Dec '06

24 Dec '06

Hi! After yet another local database corruption I've got an idea for a script that will do the following: go through files in /var/lib/pacman/local/ and check dependencies of them with those, listed in /var/lib/pacman/{current,extra,testing,unstable,community} and try to repair any found problems. For example both pacman and pacman3 write wrong dependency info into local database after some actions with some packages (there are few bugs about this already, but some may be unknown) - this script could fix this by rebuilding dependency tree from repos (it's highly recommended that all packages in current are up-to-date for this, but not strictly necessary, because depends are changed not so frequently). Of course there is the problem with this - what if this script will build wrong dependency tree or have other bugs. :p But I think there shouldn't be big problem to ensure it will work like swiss watch, and it can be used to find bugs in pacman when doing -S / -R intensively. What do you think about this? -- Roman Kyrylych (Роман Кирилич)

3 2

[pacman-dev] anoter "user vs. sudo" issue
by Roman Kyrylych 22 Dec '06

22 Dec '06

I'm thinking about why abs should be run as root currently? I don't see the point for this. :-/ Why not make it behave like makepkg - to be run as ordinary user and store everything in his home dir (or another dir, configurable)? What do you think? -- Roman Kyrylych (Роман Кирилич)

5 5

[pacman-dev] abs script changes
by Dan McGee 22 Dec '06

22 Dec '06

I don't think I can look at a whole lot without wanting to change/fix the code. Some changes on the abs script: 1. Remove old update function, and patch new one as needed. 2. Remove extra $ABSROOT directory check 3. Remove duplicate (and I believe unnecessary?) block of code to check for passive mode flag. This double code is actually in the current abs code, is there a reason? Also added a to-do note, let me know what you think. -Dan Index: abs =================================================================== RCS file: /home/cvs-pacman/pacman-lib/scripts/abs,v retrieving revision 1.2 diff -u -r1.2 abs --- abs 21 Dec 2006 01:53:41 -0000 1.2 +++ abs 21 Dec 2006 22:32:59 -0000 @@ -18,27 +18,17 @@ for sup in "${SUPFILES[@]}"; do if [ "$sup" != "testing" ]; then if [ "$sup" = "${sup#!}" ]; then - cvsup -L 1 -r 0 -g -b $ABSROOT -c .sup /etc/abs/supfile.$sup + $CVSUP -L 1 -r 0 -g -b $ABSROOT -P $CONNMODE -c .sup /etc/abs/supfile.$sup fi elif [ "$sup" = "testing" ]; then if [ ! -d /var/abs/testing ]; then mkdir /var/abs/testing; fi cd $ABSROOT/testing - cvsup -L 1 -r 0 -g -b $ABSROOT/testing -c .sup /etc/abs/supfile.testing + $CVSUP -L 1 -r 0 -g -b $ABSROOT/testing -P $CONNMODE -c .sup /etc/abs/supfile.testing cd $ABSROOT fi done -} - - -update() { - cd $ABSROOT - for sup in "${SUPFILES[@]}"; do - if [ "$sup" = "${sup#!}" ]; then - $CVSUP -L 1 -r 0 -g -b $ABSROOT -P $CONNMODE -c .sup /etc/abs/supfile.$sup - fi - done } if [ "$1" = "-h" ] || [ "$1" = "--help" ]; then @@ -46,11 +36,12 @@ exit 0 fi -if [ ! -d $ABSROOT ]; then +if [ ! -d "$ABSROOT" ]; then echo "abs: $ABSROOT does not exist (or is not a directory)" exit 1 fi -if [ ! -w $ABSROOT ]; then + +if [ ! -w "$ABSROOT" ]; then echo "abs: no write permissions in $ABSROOT" exit 1 fi @@ -64,24 +55,12 @@ exit 1 fi -if [ ! -d "$ABSROOT" ]; then - echo "abs: directory $ABSROOT does not exist" - exit 1 -fi - -if [ "$1" = "-p" ] || [ "$1" = "--passive" ]; then - CONNMODE="-" - shift -else - CONNMODE="m" - shift -fi - if [ "$1" = "-p" ] || [ "$1" = "--passive" ]; then CONNMODE="-" shift else CONNMODE="m" + #TODO: do we want to shift if this argument was not present? shift fi

3 7

[pacman-dev] [PATCH] configure.ac: remove unnecessary swig-checks + quote possible unset variables in test
by Jürgen Hötzel 22 Dec '06

22 Dec '06

Hi, configure checked 3 times for swig. Only one check is required for all swig-based bindings. path enclosed. Jürgen

3 2

[pacman-dev] Organization: pacman todo and timeline
by Aaron Griffin 22 Dec '06

22 Dec '06

Ok. So I think it's time i get this written up. Right now, I plan on getting pacman3 into [testing] ASAP. There were some NoUpgrade issues (previous thread) and they have been "handled" for the time being. This means that stability is the single most important thing. So, I need any and all available people to test this out: http://archlinux.org/~aaron/pacman/ I will periodically push new versions here based on changes which will most likely be discussed on this list. Now, beyond that, I have been keeping a "TODO.aaron" file separate from the other stuff. There is alot of reorganization I'd like to do before I get moving on any newer (complex) features. First and foremost, having functions which are over 100-200 lines tend to imply that something needs to be split out. Alot of these functions can be split out into logical units. For instance, _alpm_add_commit (lib/libalpm/add.c) contains a large chunk of lines for determining / checking md5sums. This would make alot of sense as a separate function which would, say, determine the ACTUAL output file. This is just an example. If anyone needs something to work on on the C level, I would HIGHLY appreciate some refactoring help here. Feel free to post any and all suggestions - nothing is stupid. So as for a timeless timeline, the plans are as follows: * release pacman3 to the masses * accept bug reports and fix what people complain about a lot * refactor the libalpm / pacman internals and remove some of the overlap * feature time (I have about 3 or 4 really good ideas) Unordered, but also important: * new backends (selectable at runtime?) * make the public libalpm interface easier to use

1 0

[pacman-dev] rankmirrors script
by James Rosten 22 Dec '06

22 Dec '06

Because rankmirrors just tests a whole file, I made a variation that takes a mirror's url as an argument and tests just that mirror. Sorry if the license information is wrong at the top (I got it from GNUs website), I noticed differences between it and the one that is in all the source files in pacman-lib's cvs. I think the only difference was only the address of FSF. http://jrosten.googlecode.com/svn/testmirror is where the python script can be found (pasting made it look funky, I am starting to not like copy and paste). Criticize away. ~ Jamie R.

1 0