[aur-general] TU Application - Seblu

[Seblu]

On Mon, Jan 17, 2011 at 10:04 AM, Ronald van Haren <pressh at gmail.com> wrote:
> On Mon, Jan 17, 2011 at 12:41 AM, Seblu <seblu at seblu.net> wrote:
>> On Sun, Jan 16, 2011 at 10:09 PM, Ronald van Haren <c> wrote:
> Some pointers
>
> - cairo-xcb is broken and unmaintained by upstream
http://cgit.freedesktop.org/cairo/log/
i see several commit from this january about xcb. Reading changelog,
the backend does not seem abandoned.

i see activity on ml:
http://lists.cairographics.org/archives/cairo/2010-December/thread.html
http://lists.cairographics.org/archives/cairo/2010-November/thread.html
http://lists.cairographics.org/archives/cairo/2010-October/thread.html

i see patched on last release changelog:
http://cairographics.org/news/cairo-1.10.2/

> - cairo-xcb is known to cause X crashes and rendering issues
Crawling arch bugtracker is not really revelent of this.
I don't find many bugs around this issue (maybe not detected as corollary).
RedHat bug is not really accurate.

I also use cairo-xcb from aur everyday and i don't see issues.

> - crashes and redering issues will probably appear in other
> applications than awesome
Sure! This is to avoid.

> - people don't know that they are using something that is unsupported
> (at least by upstream)
Upstream speak about experimental feature, not unsupported (see README).
By our bleeding edge point of view, we enable a lot of experimental
feature (in particular in linux kernel).
Experimental "tag" is not always a synonym of unstable (or unsuported)
but more like really Bleeding edge.

> - it is generally a bad idea to support something which is encouraged
> not to by upstream
I agree. But i'm wondering why debian include it by default in next
release (squeeze)
http://git.debian.org/?p=collab-maint/cairo.git;a=blob;f=debian/rules;h=9270f7e33e8c15510e2c74d4c73a65b886d802dd;hb=HEAD

I see in configure, they include --enable-xlib and --enable-xcb.
_Maybe_ we can enable the two to have the best of both worlds.
This need more investigation.

> So unless you want to fix all issues in cairo-xcb, it is probably not
> a good idea to support it. In case you will, upstream is also looking
> for a maintainer of the xcb backend if I'm not mistaken.
I don't see the job offer :)

> But then
> again, you may need to first fix all issues before you bring it in :-)
>
I see the point.
I would contact cairo debian maintainers about that. To have their
point of view.

-- 
Sébastien Luttringer
www.seblu.net