Evolution not deleting autosave files

[Lindsay]

[ Impact ]

Autosave files are not removed from evolution local state directories (and windows leaked)

[ Test case ]

1. Open evolution, and start to compose a new email
2. Write enough text and wait few minutes so that this command returns a file
   ls -l ~/.local/share/evolution/.evolution-composer.autosave-*
3. Close the email composer window, hitting "Do not save"
4. ls -l ~/.local/share/evolution/.evolution-composer.autosave-* should list no files
5. Opening and closing again evolution should not ask to restore the previous email

[ Regression potential ]

No sounds could be performed during some UI actions

---

Running Ubuntu MATE, Ubuntu 22.04.1 LTS. Evolution v3.44.1-0ubuntu1 was installed with the distribution on a new system and received settings imported from an earlier version on another computer.

When composing an email, if the process takes long enough an autosave file is created as ~/.local/share/evolution/.evolution-composer.autosave-xxxxxx ("xxxxxx" is a random 6 character string). When the email is successfully sent, the autosave file SHOULD be deleted. It's not. When evolution is shut down and restarted I'm asked if I want to recover an unfinished email. Answering No to this will delete the autosave file, but otherwise it persists and the recovery query recurs the next time I open evolution.

Other people are having the same issue, see https://gitlab.gnome.org/GNOME/evolution/-/issues/1972. The Gnome people aren't dealing with it, perhaps considering it a distro-specific bug.

On a system used by many people this is a potential security issue.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Lindsay (fmouse)]

According to the analysis by Milan Crha at GNOME (see the link in the bug description) this appears to be associated with the use of Xorg and doesn't occur under Wayland. It seems that although Wayland is the default X server in Ubuntu-22.04, The desktop session for MATE isn't supported on Wayland, so Ubuntu-MATE still uses Xorg. See the comment by Martin Wimpress ("wimpy"), a Ubuntu-MATE project leader, at https://ubuntu-mate.community/t/ubuntu-mate-22-04-lts-jammy-jellyfish-release-notes/25318/8.

[Marian Rainer-Harbach (marianrh)]

Further analysis by Milan Crha shows that the issue under Xorg seems to be caused by Ubuntu's gtk3, see https://gitlab.gnome.org/GNOME/evolution/-/issues/1972#note_1524307.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Nathan Teodosio (nteodosio)]

Thank you for the bug report!

I rebuilt and installed GTK3 with Ubuntu specific patches disabled, but the problem still occurs in Xorg. :/

In two occasions I got

(evolution:3367): evolution-mail-composer-CRITICAL **: 20:45:02.635: e_msg_composer_is_soft_busy: assertion 'E_IS_MSG_COMPOSER (composer)' failed
(evolution:3367): module-composer-autosave-CRITICAL **: 20:45:02.635: e_composer_save_snapshot: assertion 'E_IS_MSG_COMPOSER (composer)' failed

in stderr. We could hardly wish for a better trail to follow.

With fatal-criticals one can backtrace that, though I still don't know what triggers it.

[Lindsay (fmouse)]

I notified Milan Crha (@mcrha) who's a GNOME maintainer of Evolution on the GNOME branch of this bug, https://gitlab.gnome.org/GNOME/evolution/-/issues/1972#note_1529229. He's written new features for Evolution, which have been pushed into the GNOME upstream pacakge, so he knows the code. I know I initiated this report, but you and @mcrha are far more capable of getting to the bottom of this bug than am I. I would hope that he would be reading this and possibly work with you on stomping it.

[Nathan Teodosio (nteodosio)]

Hey Lindsay, don't worry, this is just meant to keep information out there. :)

So, although disabling the patches didn't make an effect for a typical "Ubuntu on Xorg" (as in GDM's cog option), if I use a more minimal environment, such as if I log in a console and just `startx /bin/dwm`, the bug doesn't show up, either with patched or not patched GTK.

Another difference is that upon closing Evolution the following warning is logged in stderr:

    Your application did not unregister from D-Bus before destruction. Consider using g_application_run().

[Lindsay (fmouse)]

libcanberra, "translates GTK+ widgets signals to event sounds" wha???

Does anything important depend on libcanberra, or can it simply be uninstalled until a fix is pushed into the distribution? I'd rather have my MUA behaving properly than to hear widget event sounds.

Thanks for your work on this, and thanks to Milan Crha with GNOME Gitlab for wrangling this bug!! I truly hope you had fun with it. I've done my share of programming and debugging, but you folks are at a whole different level.

[Nathan Teodosio (nteodosio)]

Hey Lindsay, judging by

$ apt rdepends libcanberra-gtk3-module
libcanberra-gtk3-module
Reverse Depends:
  Depends: ubuntu-mate-desktop
  Depends: ubuntu-mate-core
  Depends: screenruler
  Depends: python3-goocalendar
  Suggests: budgie-control-center
  Suggests: gnome-control-center
  Recommends: libcanberra-gtk3-0

You can't uninstall it without uninstalling your MATE desktop as well.

That doesn't mean you can't disable it, as in

  cd /usr/lib/x86_64-linux-gnu/gtk-3.0/modules
  mv libcanberra-gtk3-module.so libcanberra-gtk3-module.so.disabled

However you would then you would be on your own.

This was really mostly tracked down by Milan, kudos to him!

[Lindsay (fmouse)]

On Fri, 2022-08-19 at 16:47 +0000, Nathan Teodosio wrote:
> Hey Lindsay, judging by
>
> $ apt rdepends libcanberra-gtk3-module
> libcanberra-gtk3-module
> Reverse Depends:
> Depends: ubuntu-mate-desktop
> Depends: ubuntu-mate-core
> Depends: screenruler
> Depends: python3-goocalendar
> Suggests: budgie-control-center
> Suggests: gnome-control-center
> Recommends: libcanberra-gtk3-0
>
> You can't uninstall it without uninstalling your MATE desktop as
> well.

Ubuntu-MATE has some very sensible features, especially the drawer
system, which seem to be decades ahead of other GUIs, especially the
boys and girls from Redmond.

> That doesn't mean you can't disable it, as in
>
> cd /usr/lib/x86_64-linux-gnu/gtk-3.0/modules
> mv libcanberra-gtk3-module.so libcanberra-gtk3-module.so.disabled
>
> However you would then you would be on your own.

Yeah, one can get into trouble trying to do an end-run around
distribution packaging systems. Been there, done that, when
circumstances required it, but there are frequently "issues".

I'm sure y'all will be pushing a fix for libcanberra into the Ubuntu
22.04 update stream, so I'll probably just put up with the annoyance in
Evolution and wait for the fix.

> This was really mostly tracked down by Milan, kudos to him!

I dropped him a thank you note too :)

[Lindsay (fmouse)]

It looks as if the source of this problem has been identified. How long should it take to get the fix pushed into Ubuntu updates? I assume that it will be.

[Lindsay (fmouse)]

Nathan, it looks as if a fix for this problem was "released" a week and a half ago. Do you have any knowledge of the progress on this? I have one system here running Ubuntu MATE w. Ubuntu 22.04.1 LTS and am manually tracking released updates/fixes every few days, and the problem persists.

I have another system running 20.04.5 LTS, which doesn't have the problem, and which I'd like to bring up-to-date. I'm reluctant to do so until this problem is fully addressed since Evolution is my everyday go-to MUA, and I send a lot of emails, and having to deal with this problem every time would be a major PITA.

[Jeremy Bícha (jbicha)]

Lindsay, this bug isn't fixed. The upstream bug was closed. Gitlab uses closed for both fixed and wont-fix bug statuses so Launchpad reports it as "Fix Released" which is correct some of the time.

The bug is still open for Ubuntu's libcanberra package.

[Lindsay (fmouse)]

Can you cite the bug report on the libcanberra package which relates to this issue? There are LOTS of them.

FWIW, the only component of the several components of the evolution suite which EXPLICITLY depends on libcanberra (ldd ....|grep canberra) is /usr/libexec/evolution-data-server/evolution-alarm-notify. Disabling the loading of this module by renaming it does NOT fix the problem, so if the problem is indeed in libcanberra then it must be used implicitly by some other part of the suite.

.... or else something much more esoteric is going on :)

[Jeremy Bícha (jbicha)]

This bug **is** the tracker bug for the issue we believe was triggered by libcanberra.

[Nathan Teodosio (nteodosio)]

I'm forwarding the program provided by
https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1949200/comments/26
as a useful minimal test case for the regression introduced by the
Libcanberra handle-display-closing patch.

[Marco Trevisan (Treviño) (3v1n0)]

Yeah, thanks I was aware of this and it was indeed my starting point.

[Steven Wormuth (swormuth)]

Is there some setting in Evolution for "Autosave"? And would disabling it or setting to a longer interval negate some of this while we wait for a fix?

[gug09 (gug)]

I, i have installed libcanberra-gtk-3-0 on my ubuntu 22.04.1 (Linux version 5.15.0-50-generic) for an other program and i have the same bug (i think ...)

evolution:
  Installé : 3.44.4-0ubuntu1

Gnome 42.4
Serveur graphique X11 (Nvidia T600)
xserver-xorg:
  Installé : 1:7.7+23ubuntu2

thanks

[[e-mail] (mike-papersolve)]

Thanks, uninstalling libcanberra-gtk-3-0 helped with this. (I'd only installed it way back in the distant past to stop those stupid warnings on the command line!)

[Esben Høg (esben-l)]

To try to get rid of this problem with Evolution I uninstalled libcanberra-gtk-3-0 as <email address hidden> suggested, and after that I can only log in to my Ubuntu 22.04 in terminal mode! I have reinstalled libcanberra-gtk-3-0 but it doesn't help. I am stuck in terminal at start-up !!

(I realize now that I should never have uninstalled libcanberra-gtk-3-0)

How do I get the GUI back ??

[Christoph Petzold (christoph-de)]

For me

  sudo apt remove libcanberra-gtk3-module

solved the problem.

[Launchpad Janitor (janitor)]

This bug was fixed in the package libcanberra - 0.30-10ubuntu2

---------------
libcanberra (0.30-10ubuntu2) lunar; urgency=medium

  * d/p/gtk-module-Handle-display-closing-gracefully: Do not leak objects on
    destruction.
    As per previous fix for LP: #1949200 we were leaking Gtk Widgets when they
    were manually destroyed, because we never released the reference that we
    added previously.
    Thanks to Alberts Muktupāvels for providing the test case and some analysis.
    (LP: #1983794)

 -- Marco Trevisan (Treviño) <email address hidden> Tue, 10 Jan 2023 12:59:22 +0100

[Timo Aaltonen (tjaalton)]

Hello Lindsay, or anyone else affected,

Accepted libcanberra into kinetic-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/libcanberra/0.30-10ubuntu1.22.10.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-kinetic to verification-done-kinetic. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-kinetic. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Timo Aaltonen (tjaalton)]

Hello Lindsay, or anyone else affected,

Accepted libcanberra into jammy-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/libcanberra/0.30-10ubuntu1.22.04.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-jammy to verification-done-jammy. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-jammy. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Ubuntu SRU Bot (ubuntu-sru-bot)]

All autopkgtests for the newly accepted libcanberra (0.30-10ubuntu1.22.04.1) for jammy have finished running.
The following regressions have been reported in tests triggered by the package:

mutter/42.5-0ubuntu1 (arm64)

Please visit the excuses page listed below and investigate the failures, proceeding afterwards as per the StableReleaseUpdates policy regarding autopkgtest regressions [1].

https://people.canonical.com/~ubuntu-archive/proposed-migration/jammy/update_excuses.html#libcanberra

[1] https://wiki.ubuntu.com/StableReleaseUpdates#Autopkgtest_Regressions

Thank you!

[Lindsay (fmouse)]

On Fri, 2023-01-27 at 17:06 +0000, Timo Aaltonen wrote:
> Hello Lindsay, or anyone else affected,
>
> Accepted libcanberra into jammy-proposed. The package will build now
> and
> be available at
> https://launchpad.net/ubuntu/+source/libcanberra/0.30-10ubuntu1.22.04.1
> in a few hours, and then in the -proposed repository.
>
> Please help us by testing this new package.

I'll be happy to test this out, but I'm a bit out to sea on building
from source these days. A couple of issues:

First, I received an email about 4 hours after yours which references the
following "excuses":

  libcanberra (0.30-10ubuntu1 to 0.30-10ubuntu1.22.04.1)

      Migration status for libcanberra (0.30-10ubuntu1 to 0.30-
  10ubuntu1.22.04.1): BLOCKED: Rejected/violates migration
  policy/introduces a regression

Is there an issue here about which I should be concerned? is the .deb
in the -proposed repository, or has it gone back to the shop for minor
repairs?

Second, I downloaded libcanberra_0.30-10ubuntu1.22.04.1.debian.tar.xz,
unpacked it, and apparently times have changed. I've built dozens of
packages using the autogen.sh; configure; make; sudo mail install
routine but although this source has an autogen.sh script, it lacks a
configure.in file and autogen.sh fails.

No need to bring me up to speed in public. A private reply re. the
required build technique will do.

Thanks for your work!!!

--
Lindsay Haisley | "The world is full of monsters with friendly
FMP Computer Services | faces and angels with scars."
512-496-7118 |
http://www.fmp.com | - Heather Brewer

[miguelquiros (mquiros)]

Updated packages from proposed repository.

In a quick first test, the bug has not shown up. If it does not appear next days, we may consider the issue as solved.

Thanks.

[Jeremy Bícha (jbicha)]

You shouldn't build the package from source; it is already built in the proposed updates repository.

miguelquiros, please specify what version of Ubuntu you are using. There are proposed packages for both Ubuntu 22.04 LTS and Ubuntu 22.10 available now to fix this bug.

[miguelquiros (mquiros)]

I am using Ubuntu 22.04.1 LTS

Best wishes.

[Lindsay (fmouse)]

On Fri, 2023-01-27 at 17:06 +0000, Timo Aaltonen wrote:
> Please help us by testing this new package.

On first look, it does the Right Things in Evolution here with regard
to autosave files. Don't know about the system sounds since I don't pay
attention to these.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Marco Trevisan (Treviño) (3v1n0)]

miguelquiros, can we mark this as verified?

[Zsolt (muzso)]

Hi!
I just found this bug report.
I use a clean installed 22.04 (not upgraded from 20.04) and a clean setup of Evolution (no config was copied from an earlier release) and I do experience this issue, it's pretty straightforward to reproduce, i.e. every time a composed email is saved into an autosave file and I send the email, the autosave file remains and upon restarting Evolution I get asked whether I want to recover. If I say yes, all emails open up that I've sent since the last restart (and "recovery").

I've tested the 0.30-10ubuntu1.22.04.1 version of the libcanberra-gtk3-module package (from the jammy-proposed repo) and I can confirm that it fixes the issue. I also hear sounds from Evolution, when a new mail arrives. I'm not sure whether this qualifies as a "libcanberra-gtk3-module still works as intended" test or not. I'm happy to do more testing if somebody gives me a testcase description.

Since I've just upgraded libcanberra-gtk3-module to 0.30-10ubuntu1.22.04.1 a few minutes ago, I've no long-term experience whether anything else got accidentally broken.

[miguelquiros (mquiros)]

I have not experienced the bug since I updated libcanberra from proposed repo. So, as far as I am concerned, the bug seems to be solved.
Thanks a lot.

[Chris Halse Rogers (raof)]

Can anyone do the verification for kinetic? Releasing this for 22.04 is blocked waiting on successful testing for 22.10.

[Jaap Keuter (jaap-keuter)]

This bug is biting me multiple times per week. Came here through https://gitlab.gnome.org/GNOME/evolution/-/issues/1972 . Digging through the comments there found that is was solved here, now to find out the package with solution is stuck in the release workflow for two month. Who or what do we need to poke to get this to completion?

[Andreas Hasenack (ahasenack)]

Jaap, what's needed is a verification for kinetic. Please see the test plan in the bug description at the very top, and try out the package from proposed as outlined in comment #24.

And we also need a kinetic verification on https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1949200 which is being fixed in the same upload.

Since jammy comes before kinetic, and both are affected, we need the verification to be complete for both, to avoid jammy users upgrading to kinetic and experiencing the bug again.

[Andreas Hasenack (ahasenack)]

This is still needing a kinetic verification.

[Łukasz Zemczak (sil2100)]

Seeing that kinetic will go EOL next month, I'll conditionally skip the requirement for kinetic to be verified before this can go out. Releasing.

[Launchpad Janitor (janitor)]

This bug was fixed in the package libcanberra - 0.30-10ubuntu1.22.04.1

---------------
libcanberra (0.30-10ubuntu1.22.04.1) jammy; urgency=medium

  * d/p/gtk-module-Handle-display-closing-gracefully: Do not leak objects on
    destruction.
    As per previous fix for lp:1949200 we were leaking Gtk Widgets when they
    were manually destroyed, because we never released the reference that we
    added previously.
    Thanks to Alberts Muktupāvels for providing the test case and some analysis.
    (LP: #1983794)

 -- Marco Trevisan (Treviño) <email address hidden> Wed, 11 Jan 2023 15:48:13 +0100

[Łukasz Zemczak (sil2100)]

The verification of the Stable Release Update for libcanberra has completed successfully and the package is now being released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.

[Utkarsh Gupta (utkarsh)]

Ubuntu 22.10 (Kinetic Kudu) has reached end of life, so this bug will not be fixed for that specific release.