[Xcb] [RFC libxcb 0/5] Fix GE events from the ground up, leading to a full 64bit sequence API

[Keith Packard]

Bart Massey <bart at cs.pdx.edu> writes:

> The other consideration was that you know exactly what is coming in on
> the wire and what you are putting out. This is useful in the case of
> protocol analyzers, test code and the like. It's just a convenience,
> as you can always read the spec to find out what's actually going on,
> but I consider it a convenience feature.

That's why I want the XML file to match the wire exactly. Yes, it would
be "nice" for the XCB exposed data structures for events and errors to
match the wire exactly, but that's not the case today; xcb exposes the
full_sequence value as being at byte offset 32 in all events and errors.

> I'm pretty agnostic about the proposed change. It seems to me that if
> we are ever going to fix it, we should do it now: it will only get
> harder later. But I'll leave it to others to decide whether it's worth
> fixing at all.

It will require a change to the library major version number as it is
an incompatible change to public data structures. That's pretty high on
the 'don't ever do that' list of library maintainer rules.

Btw, I just verified that Xlib deals with full_sequence being in the
middle of larger XGE events, so changing libxcb would require that we
synchronize the release of that with an update to Xlib that handled the
proposed change. I should have anticipated that this would be the case
as there are lots of Xinput2 applications using the Xlib binding that
work fine today.

-- 
keith.packard at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 827 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/xcb/attachments/20130605/f38be7e0/attachment.pgp>