[Xcb] New release of xcb ?

Christian Linhart chris at DemoRecorder.com
Sat Sep 5 05:42:11 PDT 2015


Hi Olivier and Uli,

I have reviewed the branch "xcb-1.11" in Olivier's repo http://cgit.freedesktop.org/~ofourdan/libxcb

This looks good.
* The 64-bit patch is identical to the one in the main branch of the official repo.
   So it applied cleanly to the version 1.11, which is good.
* setting the new version in the buildscripts and the change in the NEWS file look good.

Therefore:
Reviewed-by: Christian Linhart <chris at demorecorder.com>

I am also fine with including the threading-fixes. However the following rather obvious things should be taken care of:
* these fixes should be committed before the commit that bumps the version number.
   (this is possible as long as this is not pushed to the official repo )
* The release date in the "NEWS"-file should be updated to the date of merging these fixes.

For reviewing/testing that:
Olivier or Uli, I suggest that you set up a new private repo with all the fixes that should go in this version.
So that the version with all of its fixes can be tested or reviewed again before merging this to the official repo.

To the time-plan for reviewing and the timeout according to the "by crickets" approach:
I'll have time tomorrow (Sunday), and on Tuesday for a re-review of the updated version, so there shouldn't be much delay for my re-review.

Given the history of this thread, I doubt that anybody except us three will do a review.
So, I think, that you Uli can make the release as soon as Olivier and me will have posted the reviews for the updated release with the threading fixes.

If somebody else wants to do a review, please post a message with your intention, so that Uli can wait until that review,
at the longest until next weekend.

Cheers,
Chris






On 09/03/15 09:36, Olivier Fourdan wrote:
> Hi Uli
>
> ----- Original Message -----
>> However, looking through the changes in master, I propose to also include the
>> following two commits:
>>
>>  5b40681c88719 "Fix a thread hang with xcb_wait_for_special_event()"
>>  f85661c3bca97 "Call _xcb_wake_up_next_reader from
>>  xcb_wait_for_special_event"
>>
>> These two commit fix multi-thread hangs with code using
>> xcb_wait_for_special_event(). If I remember correctly, the combination of
>> Mesa
>> and the PRESENT extension managed to run into this and it's a problem "in the
>> real world". Also these changes seem low-risk to me (famous last words...).
> I reckon it makes sense to include bug fixes in this bug fix release :-)
>
> Seriously though, this is an issue I faced myself while adding Xpresent support to xfwm4 so I would definitely appreciate these fixes to be added.
>
>  > [If no one complains, I will just do this myself. I guess I'd create my own
>> branch for the release and just cherry-pick cb621341a62e "expose 64-bit
>> sequence
>> numbers for XLib" by hand as well.]
> The cherry-pick is straightforward and will apply without change, it's a no-brainer here.
>
>>> I've also signed the tag on my mirror repo, but if this is not suitable, no
>> problem at all, I can easily wipe this out and redo it (as long as it's not
>> merged in the official repo ^_~)
>>
>> I don't remember what the ReleasingXCB docs say, but I think it would be nice
>> if
>> the tag is signed by the same key that also signs the release announcement
>> mail.
>> So whoever does that job should also create the tag (that's me, right?).
>>
>> For next steps, the usual "by crickets" approach is to wait a while to give
>> others the chance to complain and then just "do it".
> No complain from me, but you knew that already ^_~
>  
>> So... if no one minds: Next weekend?
> Sounds great.
>
> Cheers,
> Olivier
> _______________________________________________
> Xcb mailing list
> Xcb at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/xcb



More information about the Xcb mailing list