Gallium Nine, end of the road

[Pavel Ondračka]

On 8/22/24 1:11 AM, Axel Davy wrote:
> 
> For all these reasons, unless there is vigorous protestations here, I
> will propose a PR to remove gallium nine.
> 

Hi Axel,

thank you for all your great work on nine and mesa and I also hope you
stay around.

However, I would like to argue that nine should stay for now.

I fully agree with you that nine is no longer needed for modern
hardware, however I have to second Filip here that nine still has its
uses for old HW (like the r300 driver I maintain where modern stuff
like DXVK is not an option). There has been a discussion about making
new amber branch and getting rid of the old drivers few weeks ago and I
feel like nine is in the same group, i.e., while there is extra
maintenance burden, it is not that large enough to really require
nuking it right now. However, to be honest, I'm not the one working on
core mesa changes, therefore maybe I will be corrected here about the
actual burden of having nine in tree.

I'm also obviously biased, since I've invested quite some time into
fixing r300 for nine and quite recently also into having some basic
r300/nine CI testing (although I don't have resources to make it
reliable enough for pre-merge). So I can promise to continue keeping an
eye on regressions and maybe occasionally fix something in parts of
mesa I understand (which is unfortunately not nine itself).

Best regards
Pavel

P.S. Sorry for posting in a new thread, I actually only subscribed
today to write this answer so I could not post a proper reply.