Moving amber into separate repo?

[Daniel Stone]

Hi,

On Sun, 25 Sept 2022 at 06:22, Jason Ekstrand <jason at jlekstrand.net> wrote:

> On September 24, 2022 04:04:48 "Filip Gawin" <filip at gawin.net> wrote:
>
>> > 2. repo is growing large. Amber kinda requires long history, modern
>> > mesa not. This may be good spot to split if cleanup is required.
>>
>> mesa absolutely uses long history. there is nothing to clean up. those
>> bytes of disk space are well worth it.
>>
>> (Neutral on the other points, I don't work on stuff suported in Amber)
>>
>> I thought here about traces in issues. (Sometimes traces are uploaded
>> onto gitlab by users.) I'm guessing it should be possible to reedit really
>> old closed issues and remove attachments.
>>
>
> If the fd.o admins are complaining about disk usage, we can take steps to
> reduce that but it's entirely irrelevant to the discussion of whether amber
> should get it's own repo. Pulling amber into a separate repo won't reduce
> disk usage. If anything, it'll slightly increase it but not by an amount
> that's likely to matter.
>

We're not complaining about disk usage. As pointed out, splitting Amber and
mainline Mesa wouldn't decrease that, because both of them are linear since
the dawn of Mesa. We already have repository pools to share storage between
forks, so the net storage change would be zero.

Given that, any suggested changes should be focused on users, e.g.
discoverability of Amber vs. Mesa, making sure there are no hidden traps to
lead users towards useless configurations (like building modern drivers
from the Amber branch), etc.

Cheers,
Daniel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20220926/38d44917/attachment.htm>