[Xcb] Where's XCBRenderFindStandardFormat ?

Barton C Massey bart at cs.pdx.edu
Tue May 17 14:39:46 PDT 2005


In message <428A4337.20807 at brandeis.edu> you wrote:
> Keith Packard wrote:
> > XCB isn't big on convenience functions.
> 
> We seem to be hitting this quite a lot when converting from xlib to xcb. 
>   I know there are some record, and randr that I've had to reimplement 
> in my test programs.  Do we need a more proper set of libraries ontop of 
> xcb?  Some XCBAux<Extension><FunctionName> (to go along with our 
> libraries?

You bet!  Vincent started on something like this; what we
might want is some coherent analysis about what might be
missing that we're not needing right this second :-), that
we can start to turn into a collection of small API designs.

> Or is this outside of the scope of the XCB project?

Absolutely not.  As long as we make it sit above XCB, rather
than in or under it, we're happy to have all the useful
stuff we can gather.

> I feel 
> like the stuff currently in xcb-util is a bit unwieldly and certainly 
> doesn't provide the same extension support as xlib.  But if we follow 
> this approach then we'll have just another Xlibs with different names.

Not at all.  First of all, these libraries will be atop XCB,
instead of aside it as with Xlib.  Second of all, multiple
single-purpose libraries---don't want it or like it?  don't
link against it.  Third of all, small size.  Fourth, thread
transparency and latency hiding.  Shall I continue? :-)

> Second, the naming convention as mentioned earlier... is this 
> standardized somewhere?  If not we need to come up with something.  All 
> the autogenerated stuff is fine but is there a convention for 
> xids/typedefs/structs/enums/etc?

Yes, it's conventional, but obviously no one including me is
very happy with the conventions.  There needs to be some
hard thinking here, I think.

    Bart


More information about the xcb mailing list