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

Archive repos snapshot changes
by Kristian Klausen 09 Nov '25

09 Nov '25

Hi all, We (DevOps) want to reduce [the I/O stress][1] on the archive server (archive.archlinux.org) by reducing the numbers of inodes (currently 55 millions). Most of these inodes (52,6 millions) originates from the daily snapshot (64k inodes) of the repositories. Going forward only the database files and the "sources" directory will be included in the snapshot (~120 inodes + ~2400 inodes for "sources") and all HTTP requests for package files will be redirected to the relevant package at https://archive.archlinux.org/packages/. E.g. https://archive.archlinux.org/repos/2025/11/02/core/os/x86_64/linux-6.17.6.… -> https://archive.archlinux.org/packages/l/linux/linux-6.17.6.arch1-1-x86_64.… Package files from existing snapshots will be removed at a later date. The "sources" directory will be removed at a later date, when a proper replacement is in place. If no concerns are raised before 2025-11-21 EOD, we will move forward with the changes. [1] All inodes must be traveresed for backup and hardlinking Cheers, Kristian Klausen DevOps

1 0

Dropping qt5-webengine
by Antonio Rojas 08 Nov '25

08 Nov '25

Hi, Qt5 is EOL since May. In particular, qt5-webengine's Chromium ships with unfixed vulnerabilities. Its code is bitrotting, which also makes it a costly package to maintain due to the frequent API changes in its dependencies (I had to give up on system ffpmeg and switch to its ancient bundled version, and currently doesn't build against ICU 78). I therefore think it's wise to drop it from our repos. Its remaining reverse dependencies are: csoundqt - can be built without it by disabling remote debugging support, MR at [1] fcitx-libpinyin - archived upstream for over a year globalprotect-openconnect - very out of date, a new version switching to wbkitgtk and rust is available in testing. kchmviewer - dead upstream (no activity in 4 years) qmc2 - dead upstream (no activity in 1.5 years) wiznote - dead upstream (no activity in 4 years) openshot - the only case of a (semi-)maintained application still unported to Qt6 - there was an attempt 3 years ago [2] aborted due to some drama and no signs of new efforts since then, porting is non-trivial. If there are no objections (that include an offer to pick up maintenance of the package), I will drop it and all its unported reverse dependencies in a few days. [1] https://gitlab.archlinux.org/archlinux/packaging/packages/csoundqt/-/merge_… [2] https://github.com/OpenShot/openshot-qt/pull/4836

3 3

GitLab unavailable at the moment
by Jelle van der Waa 05 Nov '25

05 Nov '25

Hi All, GitLab is experiencing some high load due to unknown reasons, devops is investigating. Expect issues when updating packages so maybe postpone them if you can. Greetings, Jelle

2 1

News Draft: waydroid >= 1.5.4-3 update may require manual intervention
by George Hu 05 Nov '25

05 Nov '25

The `waydroid` package prior to version `1.5.4-2` (including `aur/waydroid`) creates byte-compiled Python files (.pyc) at runtime which were untracked by pacman. This issue has been fixed in 1.5.4-3, where byte-compiling these files is now done during the packaging process. As a result, the upgrade may overwrite the unowned files created in previous versions. If you encounter errors like the following during the update: ``` error: failed to commit transaction (conflicting files) waydroid: /usr/lib/waydroid/tools/__pycache__/__init__.cpython-313.pyc exists in filesystem waydroid: /usr/lib/waydroid/tools/actions/__pycache__/__init__.cpython-313.pyc exists in filesystem waydroid: /usr/lib/waydroid/tools/actions/__pycache__/app_manager.cpython-313.pyc exists in filesystem ``` You can safely remove the conflicting files by running the following command before upgrading: ```sh find /usr/lib/waydroid/tools/ -type d -name __pycache__ -exec sudo rm -rf '{}' + ``` -- Best Regards, George Hu / Integral

4 5

mkinitcpio v40, packaging changes and kernel-install
by Morten Linderud 04 Nov '25

04 Nov '25

Yo, I'm dropping the new `mkinitcpio` v40 release into [core-testing] right after this email. Generally the changes are not super interesting, but we got a new build system with meson for this release. This has given us the option to enable and disable the default hooks for `mkinitcpio` during building. As a result I have made two packaging decisions! # Disabled fallback image creation for new installs. Generally they are not super useful and people that need them can enable them if they would like too. This is only affects new installs as we can't change the hook files on existing installs. # Switched over to `systemd` hooks by default. This is generally a change I think we should have done a long time ago. The systemd side of our initramfs is getting a lot of changes thanks to systemd moving the needle forward, however our busybox hooks are not getting any of these improvements and have generally stalled for years. I have swapped things over on our side, and will continue to maintain the busybox hooks upstream. Further release notes can be found here: https://gitlab.archlinux.org/archlinux/mkinitcpio/mkinitcpio/-/releases/v40 # kernel-install Beyond this next on my todo is to see if we can move us wholesale away from the custom libalpm hooks we currently have to using `kernel-install`. This is going to make things much easier for us and users to swap between initrd and uki generators. It would also allow us to simplify our kernel installation and boot creation as we can probably automatically generate configs with hooks. I plan on writing a rough outline of my idea to arch-dev-public before doing a RFC, mostly because I haven't completely decided on how we would like to do this smoothly and people might have better ideas then me. -- Morten Linderud PGP: 9C02FF419FECBE16

1 0

News draft: dovecot >= 2.4 requires manual intervention
by Thore Bödecker 29 Oct '25

29 Oct '25

Hi folks, the upstream dovecot 2.4 branch comes with handful of breaking changes, mostly concerning configuration semantics and syntax. For this, upstream provides a migration guide [1]. We've kept dovecot 2.4.1 in [testing] for quite some time and it received a decent number of signoffs. There was already a bit of chatter on arch-general regarding the availability of a dovecot 2.4 package in [extra]. While slightly unintentional, I've built and pushed todays upstream release version 2.4.2 to [extra]. However, there is a feature disparity: the 2.4 branch no longer supports replication. My own setup relies on that particular feature and I've grown quite fond of it, so I'm a bit sad to see it going away. As per upstreams EOL announcement [2], the 2.3 branch is supposedly going to receive critical secrity fixes, which I considered "maintained enough" to warrant packaging it separately until upstream fully drops it. The following packages are now available in [extra]: - dovecot23 - pigeonhole23 - dovecot23-fts-elastic - dovecot23-fts-xapian Where applicable, they are configured with the necessary conflicts/provides. The most recent dovecot-2.4.2-3 build also contains the following dovecot.install notice: > ==> Dovecot 2.4 is incompatible with older configurations and requires manual intervention. > ==> Refer to 'https://doc.dovecot.org/latest/installation/upgrade/2.3-to-2.4.html' for upgrading instructions. > ==> Alternatively, there is a dovecot23 (and pigeonhole23) package available. Given the fact that any dovecot 2.4 binary *will not* start with a <= 2.3 configuration file, it seems reasonable for us to publish an announcement to make users aware of this situation while also informing them about the alternate 2.3 packages we provide. News draft below. ----8<---- # dovecot >= 2.4 requires manual intervention The dovecot 2.4 release branch has made breaking changes which result in it being incompatible with any <= 2.3 configuration file. Thus, the dovecot service will no longer be able to start until the configuration file was migrated, requiring manual intervention. For guidance on the 2.3-to-2.4 migration, please refer to the following upstream documentation: [Upgrading Dovecot CE from 2.3 to 2.4](https://doc.dovecot.org/latest/installation/upgrade/2.3-to-2.4.html) Furthermore, the dovecot 2.4 branch no longer supports their replication feature, it was removed. For users relying on the replication feature or who are unable to perform the 2.4 migration right now, we provide alternative packages available in [extra]: - dovecot23 - pigeonhole23 - dovecot23-fts-elastic - dovecot23-fts-xapian The dovecot 2.3 release branch is going to [receive critical security fixes](https://dovecot.org/mailman3/archives/list/dovecot@dovecot.org/threa… from upstream until stated otherwise. ---->8---- [1] https://doc.dovecot.org/latest/installation/upgrade/2.3-to-2.4.html [2] https://dovecot.org/mailman3/archives/list/dovecot@dovecot.org/thread/3P45L… Cheers -- Thore "foxxx0" Bödecker GPG ID: 0xEB763B4E9DB887A6 GPG FP: 051E AD6A 6155 389D 69DA 02E5 EB76 3B4E 9DB8 87A6

1 0

Remove GTK2 from the official repositories
by Balló György 26 Oct '25

26 Oct '25

Hi all, I think we've reached the point when we can remove the very old GTK2 package from the official repositories. I recently dropped the GTK2 version of LXDE, and there are only a few applications left that still depend on GTK2: - alltray - amsynth - bless - calf - eq10q - gimp-nufraw - gkrellm - gpsim - gtk2fontsel - gtk-chtheme - hexchat - hexter - leafpad - mono-tools - nitrogen - nspluginwrapper - osdlyrics - parcellite - pidgin - sbxkb - trayer - xboard - xdialog - xlockmore Most of these have no active development a long time ago. If no one has objections, I'll drop these packages to AUR in the next days if it's not possible to build them without GTK2. Any opinion? -- György Balló Package Maintainer

2 2

New CDN based mirror now available: fastly.mirror.pkgbuild.com
by gromit 20 Oct '25

20 Oct '25

Hello everyone, we [have now setup][0][1] an additional option for a worldwide mirror: https://fastly.mirror.pkgbuild.com With the caching done by the Fastly CDN, the bandwidth speeds and time to first byte should be quite good all around the globe, especially as more and more people start using the service. Thanks a lot to Fastly for sponsoring us with free access to their services! This new mirror can be seen as an orthogonal approach to the already existing, also DevOps administered geomirrors, which will not be replaced by the CDN based mirror. Please check it out and [let us know][2] if you find any problems with it! Kind regards, Antiz, Pitastrudl and gromit [0]: https://gitlab.archlinux.org/archlinux/infrastructure/-/merge_requests/1031 [1]: https://gitlab.archlinux.org/archlinux/infrastructure/-/issues/639 [2]: https://gitlab.archlinux.org/archlinux/infrastructure/-/issues

1 0

GHC 9.6.6 enters [extra-testing]
by Felix Yan 13 Oct '25

13 Oct '25

Hi all, The GHC 9.6.x updates have now entered [extra-testing]. This time, four packages have been retired, so please remove them manually if you have them installed: - haskell-configfile - haskell-monad-gen - haskell-monad-journal - haskell-success Again a special thanks to @Vekhir for helping me in preparing many batches of updates, including this one. Please let me know if you encounter any issues. Otherwise, the batch will be moved to [extra] in a few days. -- Regards, Felix Yan

1 0

New filesystem in core-testing with fully locked accounts
by David Runge 12 Oct '25

12 Oct '25

Hi all, please note, that filesystem 2025.10.12 in core-testing now features fully locked user accounts [1] based on the ongoing related TODO [2]. In case you run into any issues with those user accounts because of this, please report them as ticket on the package. Best, David [1]: https://gitlab.archlinux.org/archlinux/packaging/packages/filesystem/-/comm… [2]: https://archlinux.org/todo/change-sysusers-to-fully-locked-system-accounts/

1 0