October 2007 - Pacman-dev - lists.archlinux.org

[pacman-dev] bsd port?
by Xavier 24 Oct '07

24 Oct '07

On Tue, Oct 23, 2007 at 04:03:34PM -0500, Aaron Griffin wrote: > The above applies here as well, BUT, I wanted to know where this was > being discussed. If pacman was ported to BSD, I've never seen, nor > heard of, patches for it. This would be valuable information for us. > Not only would it save us work, but Dan is doing this _right now_. > I just read the above on arch-dev-public. A while ago, someone came on IRC while no one was around to report the status of getting pacman running on freebsd : "got pacman to compile on libfetch and libdownload that are in FreeBSD 6.2 1. cut out with that damn bashism and linux non-posix crap 2. check the makefiles, they're full of bugs but all this du -cb and cp -a crap... and then the "source" syntax instead of "." and there's also chown root.root, which should be root:root, or even better 0:0" I just noted it because I thought it could be interesting, but I don't know anything about this stuff. And I thought no one else cared. But if Dan is interested in doing this, I just thought he should know about the last 3 lines of this report. If that doesn't help at all, sorry, just ignore this mail :)

5 7

[pacman-dev] Submitting Patches
by Aaron Griffin 24 Oct '07

24 Oct '07

I want to bring this up a little bit. There's lots of ideas thrown around on this list, but more often than not, no code, or cryptic patches. This does nothing. All it does is waste both mine and Dan's time. On 10/24/07, Xavier <shiningxc(a)gmail.com> wrote: > But if your patch is rejected, then you should probably also provide one > separate patch with only the small bug fixes. This should ALWAYS be the case. Submitting a patch with 5 different changes is NOT acceptable. "One feature per patch" is the way this works. This is not just pacman either. Most other open source projects do it this way. Think of it this way: if you submit a patch with 5 changes, and one change is rejected, then you need to resubmit the whole thing. If you submit 5 patches, and 4 get merged, you only need to resubmit one small patch AND you get listed in the SCM history 4 times. Yay. Secondly, if a patch is submitted, and it's broken, resubmit the patch. Don't say "oh change this". If you'd like a shining example of how this process works, take a look at Nathan's recent patches. He submitted them, in git format, they were commented on, he changed them, resubmitted, and they were merged. It's really that simple. Dan and I do not need all this extra work. What Nathan did is EXACTLY how this process should work (PS Thanks a lot Nathan) Going forward, I think we're going to have to do the following: If you want a change made, a patch MUST be sent in git format. Otherwise it will simply be rejected, not for lack of effort but for the simple fact that it's a waste of our time to have to deal with.

3 6

[pacman-dev] feature request: -Sug
by Nagy Gabor 22 Oct '07

22 Oct '07

Hi! We discussed here, that after "pacman -S group" you have to answer tons of annoying questions (even if your 'foo' package is up-to-date, you are asked if you want to install it or not) or install needless packages. Usually I want to _upgrade_ my whole _installed_ group (when my net is slow, and -Su would download hundreds of MBs); then I must do pacman -S `pacman -Qg group|awk ...` magic. Or in more general: we may add a new option to -S: upgrade-only -- not add. Bye, ngaba ---------------------------------------------------- SZTE Egyetemi Könyvtár - http://www.bibl.u-szeged.hu This mail sent through IMP: http://horde.org/imp/

3 3

[pacman-dev] [GIT] The official pacman repository branch, master now at v3.0.0-473-gf5b77eb
by Dan McGee 19 Oct '07

19 Oct '07

Hello, This is an automated email from the git hooks/update script, it was generated because a ref change was pushed to the repository. Updating branch, master, via f5b77eb9894ea92a08fac586ddf03e9ba70c3de8 (commit) via e472e80c084c5e94fdc15d2f6565c50e65854372 (commit) via 520db578daf4dfe4863f9c32026460ada9131b78 (commit) via 0c2cc108d332837de4a33ca9cabe717bd414c407 (commit) via 1118e00fe91125e464e8c3487f58f6000e0036b2 (commit) via b8dba7a6faac9c44a1eda5a38b6452852e3c79be (commit) via 12e134625286be8d86f824ce2f73d1a96ac0b547 (commit) via 0ff02995f169d460d60a9512d7fa8257c74ed8f4 (commit) via 619bf56e6613333f319fb1add22fa989060e208d (commit) from 128b288ca7f396d3cbb6c9620807075274bdbe8e (commit) - Log ----------------------------------------------------------------- commit f5b77eb9894ea92a08fac586ddf03e9ba70c3de8 Author: Nathan Jones <nathanj(a)insightbb.com> Date: Fri Oct 19 13:17:54 2007 -0400 Add download size to target list. This displays the download size, taking into account delta files and cached files. This closes FS#4182. Signed-off-by: Nathan Jones <nathanj(a)insightbb.com> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit e472e80c084c5e94fdc15d2f6565c50e65854372 Author: Nathan Jones <nathanj(a)insightbb.com> Date: Fri Oct 19 13:17:53 2007 -0400 Download delta files if UseDelta is set. Delta files will be used if the size is smaller than a percent (MAX_DELTA_RATIO) of the package size. Signed-off-by: Nathan Jones <nathanj(a)insightbb.com> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 520db578daf4dfe4863f9c32026460ada9131b78 Author: Nathan Jones <nathanj(a)insightbb.com> Date: Fri Oct 19 13:17:52 2007 -0400 Refactor md5sum checking. This will allow deltas and packages to share the md5sum checking code. Signed-off-by: Nathan Jones <nathanj(a)insightbb.com> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 0c2cc108d332837de4a33ca9cabe717bd414c407 Author: Nathan Jones <nathanj(a)insightbb.com> Date: Fri Oct 19 13:17:51 2007 -0400 Add pmdelta_t structure and functions to libalpm. Signed-off-by: Nathan Jones <nathanj(a)insightbb.com> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 1118e00fe91125e464e8c3487f58f6000e0036b2 Author: Nathan Jones <nathanj(a)insightbb.com> Date: Fri Oct 19 13:17:50 2007 -0400 Add delta information to the repository database. (Dan: cleaned up one if statement) Signed-off-by: Nathan Jones <nathanj(a)insightbb.com> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit b8dba7a6faac9c44a1eda5a38b6452852e3c79be Author: Dan McGee <dan(a)archlinux.org> Date: Thu Oct 18 19:32:30 2007 -0500 Add Hungarian mirror Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 12e134625286be8d86f824ce2f73d1a96ac0b547 Author: Andrew Fyfe <andrew(a)neptune-one.net> Date: Fri Oct 19 01:18:01 2007 +0100 scripts/makepkg.sh.in: Strip bins/libs in all {bin, sbin, lib} directories. In one of the original clean up patches[1] I changed the search path for stripping binaries and libraries. This resulted in only usr/{bin,sbin,lib} being searched. This patch reverts that change. [1] 721ceee1e2c9b18425d84cf39f6541b2f04072b3 Signed-off-by: Andrew Fyfe <andrew(a)neptune-one.net> commit 0ff02995f169d460d60a9512d7fa8257c74ed8f4 Author: Chantry Xavier <shiningxc(a)gmail.com> Date: Thu Oct 18 13:26:04 2007 +0200 libalpm/package.c : fix requiredby with multiple providers. The code didn't match the following comment : "A depends on B through n depends <=> A listed in B's requiredby n times" It stopped at n=1 with a break. I was surprised to see this case happens in real, that's how I noticed the bug: wine depends on both freeglut and glut, while freeglut provides glut. So when installing wine, the update_depends function listed wine twice in freeglut's requiredby. But the compute_requiredby function (used when installing freeglut, and used by testdb) listed wine only once in freeglut's requiredby. That made testdb unhappy. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> Signed-off-by: Dan McGee <dan(a)archlinux.org> commit 619bf56e6613333f319fb1add22fa989060e208d Author: Dan McGee <dan(a)archlinux.org> Date: Tue Oct 16 08:41:37 2007 -0500 Remove rounding on package size totals We print the total to two decimal places, so there is no real need for rounding of the values. Remove the rounding and switch all output to two decimal places. Signed-off-by: Dan McGee <dan(a)archlinux.org> ----------------------------------------------------------------------- Diffstat: doc/pacman.conf.5.txt | 4 + etc/pacman.d/mirrorlist.in | 2 + lib/libalpm/Makefile.am | 1 + lib/libalpm/alpm.h | 25 ++ lib/libalpm/alpm_list.c | 18 ++ lib/libalpm/alpm_list.h | 1 + lib/libalpm/be_files.c | 19 ++ lib/libalpm/db.h | 3 +- lib/libalpm/delta.c | 265 ++++++++++++++++++++++ lib/libalpm/{conflict.h => delta.h} | 33 ++-- lib/libalpm/error.c | 5 + lib/libalpm/handle.c | 6 + lib/libalpm/handle.h | 1 + lib/libalpm/package.c | 22 ++- lib/libalpm/package.h | 1 + lib/libalpm/po/POTFILES.in | 1 + lib/libalpm/sync.c | 415 +++++++++++++++++++++++++++++++---- lib/libalpm/util.h | 3 + scripts/makepkg.sh.in | 2 +- scripts/repo-add.sh.in | 56 +++++- src/pacman/callback.c | 23 ++- src/pacman/pacman.c | 3 + src/pacman/query.c | 2 +- src/pacman/sync.c | 2 +- src/pacman/util.c | 28 +-- src/pacman/util.h | 2 +- 26 files changed, 851 insertions(+), 92 deletions(-) hooks/update --- Git Source Code Management System hooks/update refs/heads/master \ 128b288ca7f396d3cbb6c9620807075274bdbe8e \ f5b77eb9894ea92a08fac586ddf03e9ba70c3de8

1 0

[pacman-dev] fglrx-utils preferred over libgl-dri
by Alexander Baldeck 19 Oct '07

19 Oct '07

Hey all, I was just about to upgrade oxine and it used to only depend on xine-lib. xine-lib pulls a great deal of xorg stuff but then one strange thing happened. pacman felt like fglrx-utils would be the right package to pull to fulfill the requirement of 'libgl' on the system. Now I would like to not have that, what would happen if I were to go and change pkgname of libgl-dri to just libgl since Jan dropped libgl-mesa long ago? Pacman should prefer the real pkgname over provides, right? Cheers, -J

5 12

[pacman-dev] Round 2: Add delta information to libalpm.
by Nathan Jones 19 Oct '07

19 Oct '07

Here are the updated patches for adding delta downloading to libalpm. I had originally thought that if a delta could not be downloaded (due to md5sum or file deleted on the server), the package file should be downloaded instead. I decided not to do this because some users may be annoyed at seeing a 7MB download size but end up downloading 100MB openoffice package. If the md5sum is not correct, the output is mangled a bit. I am not sure what causes this. Proceed with installation? [Y/n] y :: Retrieving packages from test... wbox-3-1_to_4-1-i686... 7.9K 17.1M/s 00:00:00 [#####################] 100% wbox-4-1_to_4-2-i686... 0.4K 1087.1K/s 00:00:00 [#####################] 100% :: File wbox-4-1_to_4-2-i686.delta is corrupted. Do you want to delete it? [Y/n] checking delta integrity... error: failed to commit transaction (corrupted delta) Errors occurred, no packages were upgraded. Also, is it just me, or do 78% of the pactests fail?

3 8

[pacman-dev] [PATCH] scripts/makepkg.sh.in: Strip bins/libs in all {bin, sbin, lib} directories.
by Andrew Fyfe 18 Oct '07

18 Oct '07

In one of the original clean up patches[1] I changed the search path for stripping binaries and libraries. This resulted in only usr/{bin,sbin,lib} being searched. This patch reverts that change. [1] 721ceee1e2c9b18425d84cf39f6541b2f04072b3 Signed-off-by: Andrew Fyfe <andrew(a)neptune-one.net> --- scripts/makepkg.sh.in | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index 4399076..df6c92e 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -736,7 +736,7 @@ tidy_install() { if [ "$(check_option strip)" = "y" ]; then msg2 "$(gettext "Stripping debugging symbols from binaries and libraries...")" - for file in $(find {,*/}{bin,lib,sbin} -type f 2>/dev/null || true); do + for file in $(find {,usr/{,local/},opt/}{bin,lib,sbin} -type f 2>/dev/null || true); do case "$(file -biz "$file")" in *application/x-sharedlib*) # Libraries /usr/bin/strip --strip-debug "$file";; -- 1.5.3.4-dirty

2 1

[pacman-dev] [PATCH] libalpm/package.c : fix requiredby with multiple providers.
by Chantry Xavier 18 Oct '07

18 Oct '07

The code didn't match the following comment : "A depends on B through n depends <=> A listed in B's requiredby n times" Because of the usage of break instead of continue, it stopped at n=1. I was surprised to see this case happens in real, that's how I noticed the bug: wine depends on both freeglut and glut, while freeglut provides glut. So when installing wine, the update_depends function listed wine twice in freeglut's requiredby. But the compute_requiredby function (used when installing freeglut, and used by testdb) listed wine only once in freeglut's requiredby. That made testdb unhappy. Signed-off-by: Chantry Xavier <shiningxc(a)gmail.com> --- lib/libalpm/package.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/lib/libalpm/package.c b/lib/libalpm/package.c index 38e6e4c..cd33ca2 100644 --- a/lib/libalpm/package.c +++ b/lib/libalpm/package.c @@ -543,7 +543,7 @@ alpm_list_t SYMEXPORT *alpm_pkg_compute_requiredby(pmpkg_t *pkg) _alpm_log(PM_LOG_DEBUG, "adding '%s' in requiredby field for '%s'\n", cachepkgname, pkg->name); reqs = alpm_list_add(reqs, strdup(cachepkgname)); - break; + continue; } } } -- 1.5.3.4

4 3

[pacman-dev] Add delta information to libalpm.
by Nathan Jones 18 Oct '07

18 Oct '07

This patchset adds delta information to repository databases and libalpm, allowing libalpm to download multiple delta files when upgrading a package. I still have some work to do, but I thought I would post this now to get any suggestions and criticisms that anyone has. Some things I still need to do: Add error checking/md5sum to the download function. Remove some hardcoded stuff (like ".pkg.tar.gz"). Run through valgrind to make sure everything is freed. Delete delta files and intermediate packages after they are not needed. Test... The patches in action: $ ls -lh wbox*.pkg.tar.gz wbox*.delta -rw-r--r-- 1 nathanj users 55K 2007-10-15 17:11 wbox-3-1-i686.pkg.tar.gz -rw-r--r-- 1 nathanj users 7.9K 2007-10-15 17:11 wbox-3-1_to_4-1-i686.delta -rw-r--r-- 1 nathanj users 56K 2007-10-15 17:11 wbox-4-1-i686.pkg.tar.gz -rw-r--r-- 1 nathanj users 398 2007-10-15 17:11 wbox-4-1_to_4-2-i686.delta -rw-r--r-- 1 nathanj users 56K 2007-10-15 17:11 wbox-4-2-i686.pkg.tar.gz $ repo-add test.db.tar.gz wbox-4-2-i686.pkg.tar.gz ==> Extracting database to a temporary location... ==> Adding package 'wbox-4-2-i686.pkg.tar.gz' -> Removing existing package 'wbox-3-1'... -> Creating 'desc' db entry... -> Computing md5 checksums... -> Creating 'depends' db entry... -> Creating 'deltas' db entry... -> Added delta 'wbox-3-1_to_4-1-i686.delta' -> Added delta 'wbox-4-1_to_4-2-i686.delta' ==> Creating updated database file /home/nathanj/pkgs/pkgs/test.db.tar.gz $ pacman -Q wbox wbox 3-1 $ sudo pacman -Sy wbox :: Synchronizing package databases... test 3.4K 8.2M/s 00:00:00 [#####################] 100% resolving dependencies... done. looking for inter-conflicts... done. Targets: wbox-4-2 Total Package Size: 0.05 MB Total Download Size 0.01 MB Total Installed Size: 0.08 MB Proceed with installation? [Y/n] :: Retrieving packages from test... wbox-3-1_to_4-1-i686... 7.9K 23.3M/s 00:00:00 [#####################] 100% wbox-4-1_to_4-2-i686... 0.4K 5.3M/s 00:00:00 [#####################] 100% Generating wbox-4-1-i686.pkg.tar.gz with wbox-3-1_to_4-1-i686.delta... done. Generating wbox-4-2-i686.pkg.tar.gz with wbox-4-1_to_4-2-i686.delta... done. checking package integrity... done. (1/1) checking for file conflicts [#####################] 100% (1/1) upgrading wbox [#####################] 100%

7 22

[pacman-dev] RootDir using symlinks -> problem
by Xavier 16 Oct '07

16 Oct '07

I use the following options for using an alternate rootdir and testing stuff: RootDir = /home/xav/pacman/pacman/foo DBPath = /home/xav/pacman/pacman/foo/var/lib/pacman CacheDir = /home/xav/pacman/pacman/foo/var/cache/pacman/pkg with the following symlink : /home/xav/pacman -> /data/share/devel/pacman/ Running pacman -v with this config gives the following : Root : /data/share/devel/pacman/pacman/foo/ DB Path : /home/xav/pacman/pacman/foo/var/lib/pacman/ This makes the scriptlets handling in libalpm/trans.c break : 570 strncpy(scriptfn, installfn, PATH_MAX); 571 /* chop off the root so we can find the tmpdir in the chroot */ 572 scriptpath = scriptfn + strlen(root) - 1; Because for example: scriptfn = /home/xav/pacman/pacman/foo/var/lib/pacman/local/slocate-3.1-3/install root = /data/share/devel/pacman/pacman/foo/ so scriptpath result in /pacman/local/slocate-3.1-3/install, instead of /var/lib/pacman/local/slocate-3.1-3/install

4 19