Is Slow Patch Review For Mesa Driving Away Developers?

[phoronix]

Phoronix: Is Slow Patch Review For Mesa Driving Away Developers?

While most everyone would agree Mesa could benefit from more developers of this important piece of the open-source Linux desktop stack for providing OpenGL/3D graphics drivers, it seems slow patch review times are frustrating at least some casual developers wanting to contribute...

http://www.phoronix.com/scan.php?page=news_item&px=Mesa-Slow-Patch-Review

[BtbN]

It's not only their patch review. Over the years, i reported several bugs via their Bugtracker, so far not even the 3 year old ones have been addressed.
Most recently, i stumbled upon a bug which i'd call a release blocker for mesa 11:

https://bugs.freedesktop.org/show_bug.cgi?id=91646

It basicaly boils down to "If something linking against libudev also links against any mesa library, it will run into an infinite loop".
So far this bug is present in every mesa 11 release candidate, and will most likely make it into the release.
It affects basicaly every video player using OpenGL for rendering. I found it while testing VAAPI/EGL/HEVC interop in Kodi, took me several days to figure out what the hell was going on.

[Krejzi]

I have found out that bothering the right people in the right place (other than mailing lists) can speed things up and usually works most of the time. So, in rare cases I submit something, I go to IRC, find the devs that could review it and bother them (although not frequently, once in a day or two is enough) until I get a review.

I remember when everyone was ignoring Canonical when Mir EGL backend + other useful patches were submitted until I specifically asked on IRC about that and got the attention of one or two devs which reviewed "other" patches (new EGL image version which helped wayland too) from the series and they got merged after 2-3 revisions.

This might be related to G-Man's "The right man in the wrong place can make all the difference in the world."

[marek]

I have found out that bothering the right people in the right place (other than mailing lists) can speed things up and usually works most of the time. So, in rare cases I submit something, I go to IRC, find the devs that could review it and bother them (although not frequently, once in a day or two is enough) until I get a review.

Yes, and this is what people should do if there seems to be no other option.

[FireBurn]

Isn't a good way of getting your patches reviews to be a good reviewer. Review some other people's patches and they're more likely to return the favour...

[anarki2]

Another great example of why mailing lists are absolutely, totally inappropriate for project management purposes.

[carewolf]

This is common on many open source projects. It works best if the review tool forces you to pick specific reviewers, because without one assigned patches get lost. Even in professional settings this is common.

[F1esDgSdUTYpm0iy]

Honestly, this sounds like the project either has a lack of maintainers/reviewers or just poor maintainers/reviewers. We all know (and hopefully love) another open source project which sees far more activity (risk of noise drowning your ideas/patches) which does not suffer from this kind of phenomenon (that I know of at least) -- the Linux kernel. And it is sad to see Mesa suffering from this, to me it suggests that the project could be propelled forward if something was done about this.

[Creak]

I really like the GitHub approach. Is there an open source flavor for that? GitLab?

[cb88]

I really like the GitHub approach. Is there an open source flavor for that? GitLab?

Gogs is the easiest to setup... its just a static binary written in go (assuming you live in x86 or arm land).