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

[Keith Packard]

Daniel Martin <consume.noise at gmail.com> writes:

> But, they see it in the generated event structure.

It's already visible in the xcb_generic_event_t and xcb_generic_error_t
structures, which is part of the XCB API and ABI today. Frankly, I think
we're stuck with having the full_sequence after the first 32 bytes of
events and should just live with the presence of this member in larger
events as the best way to ensure ABI compatibility.

> I'll adopt the "2 step" suggestion of Josh. That won't be such a hard
> change.

Frankly, the only reason XCB exposes the native event wire format to
applications is because that's the 'cheapest' way of doing it; it
doesn't provide any practical benefit for applications.

I don't think any of this is worth a SONAME version change. Live with
the 'mistake' and move on. Changing SONAMEs is an enormous burden on
distributions; they'll end up shipping both versions for years to come
to provide ABI compatibility with existing applications (and existing
libraries).

I suggest the only change necessary here is to make the header
generation for libxcb automatically insert the full_sequence into the
event structure in the right place. That makes the .xml files accurately
represent the wire format.

-- 
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/b2ca4f04/attachment.pgp>