Arch Linux User Repository

@shtrophic : does it make any difference? The reason I used 0.2510.0 is that is what is used in the tags in the upstream git repo. 25.10 type names are "fancy names" they give the release if you would, whereas 0.2510.0 type version strings are what is used in commit tags, and hence in building a download link. We honestly could use either here, but 0.2510.0 just seemed more convenient.

whats wrong with this pkgver? why not just set it to be 25.10?

Thanks a lot @marcool04 and @sukanka!

hi @jkhsjdhjs. That's interesting, and entirely unintended on my part at least (maintainer of mautrix-whatsapp-bin which has an identical .service file ; and now co-maintainer of this package, thanks @sukanka btw! 👍️) I'll make the correction on the other package, thanks for explaining this and well done for the quick fix @sukanka

Can you remove Restart=on-failure or RestartSec=30s from the service file? Auto restart with such a long delay prevents the unit to ever fail and thus won't show up as a failed unit in systemctl --failed, because it won't run into the limit defined by StartLimitIntervalSec and StartLimitBurst.

@Sir-Photch your comment is perfectly justified. I maintain the mautrix-whatsapp-bin package, and I have also made the mistake of bumping that to 0.10.8, which is not a good idea as you have said. However, in my and @sukanka's defence, the notification email which github sent our about the new release didn't say anything about it being pre-release, nor was that notice on it:

v0.10.8

Repository: mautrix/whatsapp · Tag: v0.10.8 · Commit: f0b06fa · Released by: tulir

Added proxying options to config.
Updated fallback message for live locations and polls to clarify the user should open the native WhatsApp app.
—
This release has 7 assets:

mautrix-whatsapp-amd64
mautrix-whatsapp-arm
mautrix-whatsapp-arm64
mautrix-whatsapp-darwin-arm64
sha256sums.txt
Source code (zip)
Source code (tar.gz)
Visit the release page to download them.

—
You are receiving this because you are watching this repository.
View it on GitHub or unsubscribe from all notifications for this repository.

I would imagine that what happened is upstream release 0.10.8 as a normal release, and realised after there were problems, so added the notice. However, by then, we had already published these 0.10.8 AUR PKGBUILDs, so we are a bit stuck now.

It's possible to push a "fake" 0.10.8-2 which is in fact 0.10.7 in disguise, that would be fine for anybody still on 0.10.7 or below, or new installations, but as you say that could cause issues for anybody who did install 0.10.8, if the program doesn't support downgrades and the databases get messed up.

Not sure what to do on my end. We'll have 0.10.9 in a couple weeks, I think I might just leave the broken 0.10.8 up in the interim...

v0.10.8 is marked upstream as a pre-release, with regressions explicitly noted in the release notes. Why did you update the PKGBUILD to that? I am exactly experiencing the issues described in the release notes. Eager people might as well use mautrix-whatsapp-git. This really makes for a bad experience when upstream does not support downgrades...

The PKGBUILD creates /etc/mautrix-whatsapp with mode 755, but tmpfiles define it as 750. This causes warnings on every update:

warning: directory permissions differ on /etc/mautrix-whatsapp/
filesystem: 750  package: 755

Can you make this consistent?

Also, if you add -n as a runtime argument in the systemd unit, the bot will not try to write the config file, which would take care of misleading warnings in the log. The bot functions fine with only read access to the config.

the logpath patch needs to change to /var/log/mautrix-whatsapp/mautrix-whatsapp.log, as specified in ReadWritePaths