Status on GNOME for Jessie

[Niels Thykier]

On 2015-03-13 20:44, Simon McVittie wrote:
> (Non-subscriber, replying via the list archives; sorry if this breaks
> threading.)
> 

Hi Simon,

Thanks for following up, especially with half the bugs fixed already. :)

(Re: the follow-up itself, the only thing I noticed is that pkg-gnome
seemed to get dropped, I have re-added them).

> Niels wrote:
>> [...]
> 
>> * #775877 - gnome-session: No mouse pointer after login
> 
> I cannot reproduce this, but I've found a bug that seems like a
> plausible reason for the crash reported by the bug submitter, and had a
> 2-line patch accepted upstream. I'm currently waiting for the bug
> submitter to test the potential fix. Would you like me to upload it
> blindly in the hope that it is in fact the root cause?
> 

This one: https://bugzilla.gnome.org/attachment.cgi?id=298916&action=diff  ?

Seems reasonable.  Especially if we know the lack of an error can
trigger an assertion.  If nothing else, we fixes someone else's problem
before they can report it. :)

>> * #776746 - gnome-session: GNOME crashes during a remote desktop
>>   access
> 
> As far as I can tell, the issue here is that GNOME Shell requires the
> decade-old Composite extension, but neither of our Xvnc implementations
> seem to have that, because they're both based on X server codebases
> older than that extension. I don't see how this can be resolved from the
> GNOME side, other than wontfix; Shell is a compositing manager, and just
> won't work without compositing. It doesn't seem particularly realistic
> to fix this from the Xvnc side for jessie either.
> 

This is certainly unfortunate but I agree that that fixing it in Jessie
is unlikely to happen.

What are the prospects on getting a more informative error message in GNOME?
  Failing (or in addition to) that, it might be worth documenting it in
the release-notes - apparently we got a user base relying on VNC and
GNOME working together.

> [...]
> My suggestion would be to reassign this to our two Xvnc implementations
> with "affects xrdp" and perhaps "affects gnome-session", and drop it to
> a non-RC severity.
> 
> [...]

Or downgrade and block it on the tigervnc RFP?  (Just a suggestion - not
sure which is better).  Anyhow, please go ahead with that.

>> * #770130 - gnome-shell: fails to start on i386 when built with [...]
> 
> Two separate issues, probably. I've unmerged #775235, which has a tested
> (but upstream-rejected) patch, and reassigned to mesa; this seems worth
> fixing for jessie, since it affects fairly normal VM configurations and
> there's a known-working patch.
> 

Ok?  It is currently tagged "moreinfo", but it does not seem to be
intentional based on your message above and the last message in the bug.

Is the patch the one from #775235 comment #32 (linking to
https://freedesktop.org/patch/34445/)?

> #770130/#776911 looks like an Intel driver bug (could be either kernel
> or X, I can't tell). The report that has hardware details was on Pentium
> III-era i386 hardware, and I can reproduce something very similar
> (#780413, with which the others should perhaps be merged) on a disused
> Pentium IV that I happened to have in my office. Given that that's a
> decade old at this point, and would probably not provide great
> performance for GNOME anyway, I'm not convinced this deserves to be RC.
> 
> Thanks,
>     S
> 
> 

Ok.


Thanks for the work so far, :)
~Niels