[Mesa-dev] [PATCH 09/14] egl: Lock the display in _eglCreateSync's callers

[Emil Velikov]

On 14 September 2016 at 13:36, Adam Jackson <ajax at redhat.com> wrote:
> On Wed, 2016-09-14 at 11:15 +0100, Emil Velikov wrote:
>
>> Nice one... I wonder if your view will be the same if you were never
>> involved in distribution packaging? Guess we'll never know :-\
>> In case you've forgotten things have been like that for a long time -
>> long before I jumped in.
>
> I wasn't accusing you of anything. I said _I_ am not the one making the
> decision, that's all.
>
It's surprising that you haven't heard about this, considering it's
been in use for more than three years. Guess you simply forgot ?

> Obviously I can't make definite assertions about counterfactuals about
> my work history, but

> I think considering all "new features" equally
> destabilizing is wrong.
Fully agree. The point is that defining which new things are "more
destabilizing" than others is a never ending topic. There are a few
major influences:
 - humans always have a subjective view on things, always
 - due to ^^ the feasibility and impact of backporting is related to
the personal interest in the feature

Thus, such topics are better left to distributions to lobby, discuss,
vote and/or other, as they seem fit.

> Why have an extension model if you're not going
> to use it to make assertions about the orthogonality of feature sets?
> Why refuse to reason about the code, unless you don't have any
> confidence that it's something that can be reasoned about?
>
> Yes, we do backport features, it works pretty well. If one does so
> enough times, a sense develops of how "big" of a feature it's possible
> to backport reasonably. I have my own opinion about this one, and I was
> asking what the rule was for mesa stable. Since the rule seems to be
> "no", fine, not for stable.
>
With the above said:
Yes, we are aware of (at least some) the backports that you do. Even
though I don't use such packages, I think it's quite reasonable thing
to do. Yes, your view about the severity/feasibility of such backports
is (in all likelihood) going to be spot on, yet there is the
subjective element.

Hope the above provides some clarity on the topic.

Thanks
Emil