bug#71154: slim: Two regressions on v.1.4.1

bug#71154: slim: Two regressions on v.1.4.1

[André Batista]

Hi Guix,

I've found two regressions after upgrading to v. 1.4.1 of slim.

The first one is that it is no longer possible to use a secondary monitor
with it. If it is connected at the same time as a the primary one, only
the later gets to show the slim greater. If the primary one is disabled
via kernel command line (eg. "video=LVDS-1:d"), slim does not start and
the xorg-server crashes. Shepherd keeps trying to restart it to no avail.

Curiously enough the changelog for the current version mentions that
there has been some change to "how/when the pseudo-root window is created
and removed, in preparation for handling multiple monitors"[*]. I, for one,
have been using multiple monitors without issue until now.

The second regression is more subtle and annoying. After logging in with
one account, sometimes, in what appears to be related to have used any
gtk application, the user session is not properly terminated on logout
and herd looses track of xorg-server. If one tries to stop it with herd,
the session remais alive and there is only one bit on slim.log:

'WaitForServer: Not seen SigUSR1 from Xserver'

and nothing useful on Xorg.log.

Only by manually killing the xorg process, I've been able to get back
to a sane state again.

Any thoughts?

For now I have reverted the upgrade to its current upstream version.

1. https://sourceforge.net/projects/slim-fork/files/

bug#71154: slim: Two regressions on v.1.4.1

[Z572]

[-- Attachment #1: Type: text/plain, Size: 1761 bytes --]

André Batista <nandre@riseup.net> writes:

> Hi Guix,
>
> I've found two regressions after upgrading to v. 1.4.1 of slim.
>
> The first one is that it is no longer possible to use a secondary monitor
> with it. If it is connected at the same time as a the primary one, only
> the later gets to show the slim greater. If the primary one is disabled
> via kernel command line (eg. "video=LVDS-1:d"), slim does not start and
> the xorg-server crashes. Shepherd keeps trying to restart it to no avail.
>
> Curiously enough the changelog for the current version mentions that
> there has been some change to "how/when the pseudo-root window is created
> and removed, in preparation for handling multiple monitors"[*]. I, for one,
> have been using multiple monitors without issue until now.

maybe you can report it to upstream?

>
> The second regression is more subtle and annoying. After logging in with
> one account, sometimes, in what appears to be related to have used any
> gtk application, the user session is not properly terminated on logout
> and herd looses track of xorg-server. If one tries to stop it with herd,
> the session remais alive and there is only one bit on slim.log:

I'm not exactly familiar with shepherd, maybe replacing
fork+exec-command with make-forkexec-constructor would solve this
problem?.

>
> 'WaitForServer: Not seen SigUSR1 from Xserver'
>
> and nothing useful on Xorg.log.
>
> Only by manually killing the xorg process, I've been able to get back
> to a sane state again.
>
> Any thoughts?

We can revert this commit or add a slim-1.4.0 package?

>
> For now I have reverted the upgrade to its current upstream version.
>
> 1. https://sourceforge.net/projects/slim-fork/files/

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]