On Fri, Jun 26, 2015 at 11:29 AM, Emil Velikov <[email protected]> wrote:
> Hmm I thought I've commented on this one, but here is goes:

I thought you had too :) But it's not in my e-mail, perhaps I started
a discussion thread or had an earlier version of the patch?

>
> On 26 May 2015 at 00:51, Ilia Mirkin <[email protected]> wrote:
>> Currently a piglit_report_result in init (or earlier) would cause the
>> framework to not get torn down. This ensures that the teardown happens
>> on exit, which also enables usage of valgrind to detect memory leaks.
>>
>> This is a refinment of commit 121fb6b7d84 for winsys and commit
>> 50402d6c7 which introduced the fbo framework.
>>
>> Signed-off-by: Ilia Mirkin <[email protected]>
> As piglit_report_result uses exit() rather than calling back to the
> gl/cl framework, one needs to explicitly call the destructor
> beforehand. So the message might come as a bit misleading. Perhaps one
> day piglit_report_result() will can explicitly foo->destroy(), but
> until then this looks great.

FWIW I looked into doing this, but it was going to be a pain.
piglit_report_result knows nothing of GL, and this is a GL-specific
thing. One would have to create an abstract framework concept as well,
and modify a ton of little things. This seemed easier :)

>
> Iirc older Windowses (~ Windows 2k era) had some problems with
> atexit() but I'm pretty sure that Brian/Jose will give us some tips if
> similar problems exist on supported platforms.
>
> Fwiw
> Reviewed-by: Emil Velikov <[email protected]>

Thanks!
_______________________________________________
Piglit mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/piglit

Reply via email to