[Mesa-dev] GLSL IR is no longer cool where to from here?

[Ian Romanick]

On 02/05/2016 06:57 PM, Timothy Arceri wrote:
> For the past couple of months I've been working away solely in the
> wasteland that is GLSL IR and one things seems clear. No one wants to
> review this code anymore. A lot of the original developers have either
> moved on or are busy with other things.
> 
> The difference between sending a patch with nir: ... vs glsl: ... is
> very noticable.
> 
> Its not impossible to get reviews for patches, especially if they are a
> small part of a bigger series not just confined to GLSL IR, but
> anything involving a refactoring can be difficult as no-one wants to
> relearn how this code works, step up to the ast code and things are
> even worse.
> 
> So I guess the discussion I'm trying to kick off is, with the CTS, dEPQ
> and piglit all saying no regressions (or even reporting fixes \0/).
> Should one still be forced to go around hassling people for a rubber
> stamped r-b? Or can we relax the criteria for pushing bug-fix/refactor
> type patches for GLSL-IR?
> 
> The other thing I've consider is maybe this I just don't review enough
> patches for people to reciprocate, although I've made an effort to
> review patches where I feel I can since being employed to work on Mesa
> so hopefully thats not it. Maybe it's time for Ken to run his script
> again :-P

At this point, most of it is just twitchy, annoying code that only a
couple people still really know.

I think one or two lazy people (eh-hem... maybe just one) need to get
off their butts and start actively reviewing your patches.  Please don't
start pushing unreviewed code.  That never turns out well.  There have
been enough reviewed patches this year already that have had problems...
some of them leading to reverts.  We don't need unreviewed code too.

> Tim
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev