December 2010 - Arch-dev-public - lists.archlinux.org

[arch-dev-public] parallel compression in makepkg
by Andreas Radke 31 Dec '10

31 Dec '10

ArchLinux prefers xz compression now. It has a very good compression ratio but poor speed due not being threaded for smp usage. It seems the upstream xz developer lost the intention to implement it after the final xz 5.0 release: http://sourceforge.net/projects/lzmautils/forums/forum/708858/topic/4015685 I've made some local quick tests for compressing large binary packages like LibreOffice on my Intel Quad core Q6600+ system using /tmp (tmpfs): time tar -xvf ~/arch64/packages/libreoffice-3.3.0.2-2-x86_64.pkg.tar.xz real 0m11.679s - that's well acceptable user 0m10.686s sys 0m1.117s [andyrtr@workstation64 libo]$ du -sch 298M . We currently use this in makepkg: time bsdtar -cf - * | xz -zf > LibO.tar.xz real 3m53.461s - very poor! user 3m52.382s sys 0m1.287s [andyrtr@workstation64 libo]$ ls -lha LibO.tar.xz -rw-r--r-- 1 andyrtr users 80M 31. Dez 13:14 LibO.tar.xz 1st alternative: PXZ http://jnovy.fedorapeople.org/pxz/ Using http://aur.archlinux.org/packages.php?ID=36362 I get this: [andyrtr@workstation64 libo]$ time bsdtar -cf - * | pxz -zf > LibO.tar.xz real 1m9.924s - not bad :) user 3m41.709s sys 0m1.530s [andyrtr@workstation64 libo]$ ls -lha LibO.tar.xz -rw-r--r-- 1 andyrtr users 81M 31. Dez 13:16 LibO.tar.xz but I'm a bit afraid of the comments in AUR saying decompressing fails sometimes. I couldn't find something helpful in google for broken decompression. 2nd alternative: from the forum xz thread let's try parallel 7zip [andyrtr@workstation64 libo]$ time bsdtar -cf - * | 7z a -txz -si LibO.tar.xz 7-Zip [64] 9.13 beta Copyright (c) 1999-2010 Igor Pavlov 2010-04-15 p7zip Version 9.13 (locale=de_DE.UTF-8,Utf16=on,HugeFiles=on,4 CPUs) Creating archive LibO.tar.xz Compressing [Content] Everything is Ok real 1m12.984s - also nice :) user 3m19.284s sys 0m1.667s [andyrtr@workstation64 libo]$ ls -lha LibO.tar.xz -rw-r--r-- 1 andyrtr users 79M 31. Dez 13:40 LibO.tar.xz 3rd alternative: lxz from the forum thread I couldn't find so far and should get further testing before hitting production state in a distribution. What's your opinion? Is using p7zip a solution we can add to makepkg for now? -Andy

2 1

[arch-dev-public] adding hunspell-xx packages
by Andreas Radke 31 Dec '10

31 Dec '10

Right now we don't have modern dictionaries packaged. People use AUR packages or poor OpenOffice extensions that cannot be used for other apps. The proper way is to have hunspell-xx packages providing a dictionary, thesaurus and hyphenation data files. I'm going to add a base set of hunspell-{en,de,foo} packages to testing. A common PKGBUILD doesn't make much sense because the upstream sources are maintained in different countries and releases come at different time. This way we can make use of the packages in OOo/LibO, enchant and Mozilla's FF/TB/SM packages. The files will go into /usr/share/hunspell/xx_XX with symlinks to /usr/share/myspell/dicts/xx-XX for historical reason. See http://en.wikipedia.org/wiki/Hunspell for more. To compare look at http://packages.debian.org/source/sid/openoffice.org-dictionaries http://pkgs.fedoraproject.org/gitweb/?p=hunspell-XX.git http://gentoo-portage.com/app-dicts/ the myspell-xx packages. Any comments or objection? -Andy

3 7

[arch-dev-public] [signoff] cloog-ppl-0.15.10-1
by Allan McRae 31 Dec '10

31 Dec '10

Upstream "update" - only changes to configure files. Passes all testsuite. Signoff both, Allan

1 1

[arch-dev-public] Breaking the unspoken rule: AUR helper in [community]
by Dan McGee 30 Dec '10

30 Dec '10

http://www.archlinux.org/packages/community/i686/cower/ Thoughts? I was under the impression we didn't do this, and definitely on purpose, otherwise people have *no* idea the AUR is different in a lot of ways. Making people go the "hard way" to get a helper installed at least presents some (necessary) barrier. -Dan

5 4

[arch-dev-public] [signoff] cryptsetup 1.2.0-1
by Thomas Bächler 29 Dec '10

29 Dec '10

Upstream update, please sign off. From the announcement: Changes since version 1.1.3 Important changes ~~~~~~~~~~~~~~~~~ * Add text version of *FAQ* (Frequently Asked Questions) to distribution. * Add selection of random/urandom number generator for luksFormat (option --use-random and --use-urandom). (This affects only long term volume key in *luksFormat*, not RNG used for salt and AF splitter). You can also set the default to /dev/random during compilation with --enable-dev-random. Compiled-in default is printed in --help output. Be very careful before changing default to blocking /dev/random use here. * Fix *luksRemoveKey* to not ask for remaining keyslot passphrase, only for removed one. * No longer support *luksDelKey* (replaced with luksKillSlot). * if you want to remove particular passphrase, use *luksKeyRemove* * if you want to remove particular keyslot, use *luksKillSlot* Note that in batch mode *luksKillSlot* allows removing of any keyslot without question, in normal mode requires passphrase or keyfile from other keyslot. * *Default alignment* for device (if not overridden by topology info) is now (multiple of) *1MiB*. This reflects trends in storage technologies and aligns to the same defaults for partitions and volume management. * Allow explicit UUID setting in *luksFormat* and allow change it later in *luksUUID* (--uuid parameter). * All commands using key file now allows limited read from keyfile using --keyfile-size and --new-keyfile-size parameters (in bytes). This change also disallows overloading of --key-size parameter which is now exclusively used for key size specification (in bits.) * *luksFormat* using pre-generated master key now properly allows using key file (only passphrase was allowed prior to this update). * Add --dump-master-key option for *luksDump* to perform volume (master) key dump. Note that printed information allows accessing device without passphrase so it must be stored encrypted. This operation is useful for simple Key Escrow function (volume key and encryption parameters printed on paper on safe place). This operation requires passphrase or key file. * The reload command is no longer supported. (Use dmsetup reload instead if needed. There is no real use for this function except explicit data corruption:-) * Cryptsetup now properly checks if underlying device is in use and disallows *luksFormat*, *luksOpen* and *create* commands on open (e.g. already mapped or mounted) device. * Option --non-exclusive (already deprecated) is removed. Libcryptsetup API additions: * new functions * crypt_get_type() - explicit query to crypt device context type * crypt_resize() - new resize command using context * crypt_keyslot_max() - helper to get number of supported keyslots * crypt_get_active_device() - get active device info * crypt_set/get_rng_type() - random/urandom RNG setting * crypt_set_uuid() - explicit UUID change of existing device * crypt_get_device_name() - get underlying device name * Fix optional password callback handling. * Allow to activate by internally cached volume key immediately after crypt_format() without active slot (for temporary devices with on-disk metadata) * libcryptsetup is binary compatible with 1.1.x release and still supports legacy API calls * cryptsetup binary now uses only new API calls. * Static compilation of both library (--enable-static) and cryptsetup binary (--enable-static-cryptsetup) is now properly implemented by common libtool logic. Prior to this it produced miscompiled dynamic cryptsetup binary with statically linked libcryptsetup. The static binary is compiled as src/cryptsetup.static in parallel with dynamic build if requested. Other changes ~~~~~~~~~~~~~ * Fix default plain password entry from terminal in activate_by_passphrase. * Initialize volume key from active device in crypt_init_by_name() * Fix cryptsetup binary exit codes. 0 - success, otherwise fail 1 - wrong parameters 2 - no permission 3 - out of memory 4 - wrong device specified 5 - device already exists or device is busy * Remove some obsolete info from man page. * Add more regression tests for commands. * Fix possible double free when handling master key file. * Fix pkg-config use in automake scripts. * Wipe iteration and salt after luksKillSlot in LUKS header. * Rewrite file differ test to C (and fix it to really work). * Do not query non-existent device twice (cryptsetup status /dev/nonexistent). * Check if requested hash is supported before writing LUKS header. * Fix problems reported by clang scan-build.

2 2

[arch-dev-public] glibc kernel version news draft
by Allan McRae 27 Dec '10

27 Dec '10

Hi, I'm going to move the toolchain out of testing, so need a short news item about the minimum kernel version. --draft-- Minimum required kernel version increase The update to glibc-2.12.2-1 increases the minimum required kernel version to 2.6.27. This kernel version was chosen as it is currently the oldest kernel with longterm support upstream. Users of the Arch "kernel26" and "kerne26-lts" packages are unaffected by this change. Those wanting to continue use an older kernel version will be required to rebuild the glibc package with the appropriate adjustment made to the "--enable-kernel" configuration option. --end draft-- Any comments? Allan

2 1

[arch-dev-public] [signoff] elfutils-0.150-1
by Allan McRae 27 Dec '10

27 Dec '10

Upstream bugfix release. Passes testsuite. Signoff both, Allan

1 0

[arch-dev-public] extra/progsreiserfs duplicate of core/reiserfsprogs?
by Guillaume ALAUX 26 Dec '10

26 Dec '10

Hello, Correct me if I'm wrong but it looks like extra/progsreiserfs (0.3.0.5-5) is an old duplicate of core/reiserfsprogs (3.6.21-3). I search for reasons in mailing list archives for having both but found none. Nb: Old extra/progsreiserfs PKGBUILD [1] pulls a tarball from ftp.archlinux.org. Let's remember to remove the tarball if we get rid of the package. [1] http://repos.archlinux.org/wsvn/packages/progsreiserfs/trunk/PKGBUILD -- Guillaume

1 0

[arch-dev-public] [signoff] e2fsprogs-1.41.14-1
by Allan McRae 26 Dec '10

26 Dec '10

Upstream update. Also removed references to $srcdir from /usr/bin/{compile_et,mk_cmds}. Signoff both, Allan

3 3

[arch-dev-public] [signoff] rpcbind 0.2.0-3
by Andreas Radke 25 Dec '10

25 Dec '10

Minor fix: add missing man page for rpcinfo https://bugs.archlinux.org/task/21271 Please sign off. -Andy

2 2