Re: Regarding EGLStreams support in KWin

Hi,

just adding my opinion. As my role of maintainer I kind of invited 
NVIDIA to contribute the patches. While I am no longer maintainer I 
think it's crucial for FLOSS communities to stick to their word when 
working with hardware vendors. We all want NVIDIA to use gbm, we all 
want NVIDIA to drop their blob and work on nouveau instead. Getting 
NVIDIA to work in the public together with FLOSS projects is very 
important and I am extremely happy to see these patches here for KWin. 
We have seen contributions from NVIDIA in the past and this has always 
been good for all users. Refusing patches for political reasons is not a 
good idea IMHO.

Also it doesn't make sense as the patches are out.

Cheers
Martin


Am 2019-02-20 19:13, schrieb David Edmundson:
> Whilst I welcome your opinion, I don't really agree.
> 
> It is an open standard that happens to only have one vendor.
> 
> We currently have a libhybris backend, which shims propreitory blobs
> directly. That was done by the then kwin maintainer to no noise at
> all. Having a different stance here would really be an unfair double
> standard.
> 
> A lot of the arguments apply there and overall, despite being a whole
> separate backend that's hard to even build, has proven to be much more
> of an asset than a burden.
> 
> In terms of this code, it's very well encapsulated, and very tidy. I
> don't see anything scary. The existing files are scarcely touched
> aside from some getters. There's a thin argument about use of
> DrmOutput::present, but we will follow that up on the thread.
> 
> Relatively, I think that's quite a minor thing that we should be able
> to resolve.
> 
> If we are going to discuss technical merits, we should include them
> for EGLStreams too, such as the deferred texture decompression that
> don't exist in GBM.
> 
> In terms of debugging it's not like we, as kwin, have a precedent of
> doing much other than saying "it's an upstream issue" for even the
> open source drivers.
> 
> ---
> 
> In terms of maintenance, the last time we had a huge regression where
> we broke the lockscreen on nvidia, I went out and bought an nvidia
> card and used their driver on one of my machines. Since then I've
> implemented support for NVidia's glGetGraphicsResetStatus throughout
> Qt and in Plasma, as well as doing the software fallback work to
> handle bad upgrades.
> 
> I'll continue to work on these things, my priority is my userbase, and
> overall I think this patch is a good thing.
> 
> At least two other senior KDE devs has helped test this patch too. I
> don't see a reason why it would be less maintained than any other part
> of Plasma.
> 
> I want to welcome the open source contribution that Nvidia is making.
> Embracing open source outreach is the way to generate more.
> 
> --
> 
> David Edmundson