[Xcb] Deprecations in xcb-demo removed, etc

[Jamey Sharp]

On Sun, Dec 11, 2005 at 02:46:34PM -0800, Barton C Massey wrote:
> I've systematically gone through xcb-demo and made it
> conform to Jamey's new stable API.

Thanks much!

> One conclusion I came to as a result of that exercise was
> that XCBAuxGetScreen() should be moved to XCB proper (and
> renamed appropriately?)---most X programs seem to need it,
> and it doesn't seem like this will change with toolkits,
> etc.  Should all of xcb-utils/convenience/xcb_aux.c
> (uggh---stupid CVS) be moved into core?  It's not much code,
> and it would be convenient.

GetScreen seems like a win to have in XCB, I guess. GetDepth and
GetVisualtype seem to me like they're not commonly used (or at least
that they shouldn't be in new code); I'd like some evidence to the
contrary before adding them to the API. They also don't look ready yet:
GetDepth needs some thought put into error handling (instead of using
perror/exit) and off-hand it seems wrong that GetVisualtype only looks
at the first depth on the screen. Perhaps the latter needs a depth
argument, or needs to iterate over all depths? Also, I think
GetVisualtype should take an XCBSCREEN instead of a screen number.

I've just committed a version of GetScreen that uses the code I posted a
couple of days ago for Vincent's tutorial. I wouldn't mind having that
moved over to xcb_util.c at some point.

> Also, there's a very minor error handling issue when
> XCBConnect() fails; what (if anything) to report to the
> user?  See some kludgy, UNIX-centric code I added to
> xcbrandr.c to see what I mean.

Yup, I tell you something bad happened but don't tell you what. Any
suggestions?

> Finally, some bugs.  On my server (X.org 6.8.2, Debian
> 6.8.2.dfsg.1-10), rendertest core dumps the server after a
> few seconds(!).

I noticed that too, shortly before posting my announcement. I lost a lot
of state to that. :-( I'm using Debian xserver-xorg version
6.8.2.dfsg.1-6.

Seems pretty certain it isn't "our" problem, but rather a server bug.
I'm not especially interested in tracking it down right now.

> Also, tests/lissajous 1 (shared memory)
> doesn't seem to display anything for me, although 0 (normal
> connection) works fine.

I've been meaning to poke at that for months. I think I inspected the
code and it looked reasonable to me; maybe you could try debugging it?

--Jamey
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.freedesktop.org/archives/xcb/attachments/20051212/98d106bc/attachment.pgp