Package Details: snapd-git 2.45.r489.gf8a32f8e27-1

Git Clone URL: https://aur.archlinux.org/snapd-git.git (read-only, click to copy)
Package Base: snapd-git
Description: Service and tools for management of snap packages.
Upstream URL: https://github.com/snapcore/snapd
Licenses: GPL3
Conflicts: snap-confine, snapd
Provides: snapd
Submitter: bboozzoo
Maintainer: None
Last Packager: bboozzoo
Votes: 7
Popularity: 0.000000
First Submitted: 2017-11-17 06:47 (UTC)
Last Updated: 2020-05-18 14:24 (UTC)

Required by (15)

Sources (1)

Pinned Comments

bboozzoo commented on 2020-04-23 15:21 (UTC) (edited on 2020-04-23 15:22 (UTC) by bboozzoo)

This package builds directly from snapd master branch. For all your regular use cases, consider using snapd instead.

Latest Comments

« First ‹ Previous 1 2 3 Next › Last »

egrupled commented on 2018-03-29 12:46 (UTC) (edited on 2018-03-29 12:47 (UTC) by egrupled)

@bboozzoo Considering below, can you add export TMPDIR="/tmp" at the beginning of check() section? This will make sure that /tmp is used for tests. Otherwise build will fail for everyone who uses non-standard TMPDIR.

egrupled commented on 2018-03-20 09:51 (UTC) (edited on 2018-03-20 18:20 (UTC) by egrupled)

I think when someone set TMPDIR to something else than /tmp they had good reason too. Hardcoding dirs isn't smart behavior. Also if they refuse to remove they should refuse to create in first place.

bboozzoo commented on 2018-03-20 06:36 (UTC)

@egrupled the cleanup in unit tests refuses, IMO for good reasons, to remove anything that does not start with /tmp/, given that the directory is created using glib's g_dir_make_tmp() and the strace, I'd guess that your TMPDIR is set to /build

egrupled commented on 2018-03-19 20:24 (UTC) (edited on 2018-03-19 21:46 (UTC) by egrupled)

@bboozzoo Yeah, in clean chroot it works. Unfortunately most AUR helpers don't support building in clean chroot.

I uploaded strace of failed attempt here: https://ghostbin.com/paste/h4n4r/raw

I suppose it's related to BUILDDIR != TMPDIR on my machine.

bboozzoo commented on 2018-03-19 18:31 (UTC)

@egrupled tried mkchrootpkg in a clean chroot without any issues. Running unit tests directly no issues either. Is this reproducible? Can you run this specific test under strace (strace -vf ./cmd/libsnap-confine-private/unit-tests -p /locking/sc_lock_unlock) and paste the log somewhere?

egrupled commented on 2018-03-19 17:40 (UTC)

Check fails on my machine with:

/locking/sc_lock_unlock: refusing to remove: /build/.LMFQFZ: Resource temporarily unavailable
make: *** [Makefile:3334: check-unit-tests] Error 1

Any clues?

bboozzoo commented on 2018-02-22 07:25 (UTC)

@mati865 can share which apps segfault? It'd be interesting to look into this in more detail. The only snaps requiring opengl that I have used are ohmygiraffe and boa. No issues there, although I'm using nvidia-340xx drivers as my hardware is quite dated.

mati865 commented on 2018-02-21 22:28 (UTC)

@bboozzoo thanks, I hoped snap run --gdb from git will help to discover why apps requiring OpenGL segfault. Unfortunately there is no stack trace avaiable. Nvidia drivers 390

bboozzoo commented on 2018-02-21 10:23 (UTC) (edited on 2018-02-21 11:29 (UTC) by bboozzoo)

@mati865 same thing happens on Arch at the moment. Master branch is in a bit of a flux, so it's best to use snapd package, which has just been updated to 2.31.1 and should work just fine.

Update: you need to use edge branch of the core snap with current master (probably at least until 2.32) is released. snap refresh core --edge will do the trick

mati865 commented on 2018-02-21 10:04 (UTC)

On Manjaro when starting any non-classic snap I'm getting:

$ spotify
execl failed: No such file or directory
child exited with status 1

strace: https://hastebin.com/fomedesawo

« First ‹ Previous 1 2 3 Next › Last »