On Thu, 4 Sep 2014 22:05:00 -0500 Paul Sbarra <sbarra.p...@gmail.com> wrote:
> I agree it's a hack, but it's also one that invalidates the protocol > specification. If the wire protocol requires four arguments then the > specification needs to reflect that. Currently if another tool or protocol > implementation (like the gowl example previously mentioned) attempts to > generate code from the specification they end up code that is not > compatible with libwayland applications. In my opinion that's a pretty bad > outcome for a hack, documented or not. No, we are not going to break everyone else's generator by going back to a version that existed *before* Wayland 1.0, while 1.0 was the first stable release of Wayland. This XML definition or generator behavior has not changed during the whole lifetime of the stable Wayland protocol. This special case does not invalidate "the protocol specification". It is part of the Wayland XML format, the only problem being that the XML format is not documented. We now have bug about that: https://bugs.freedesktop.org/show_bug.cgi?id=83478 See the commits mentioned in http://lists.freedesktop.org/archives/wayland-devel/2014-September/017087.html for more information on why this particular design choice was made. It was very deliberate. While wl_registry.bind happens to be the only user of the special case, you must not assume it will be the only one forever, or that no-one's private extension would not use this feature. > Attached is a patch that resolves the problem (at least it limits the scope > of the hack to the wayland code generator) while maintaining the existing > api. Thank you in advance for reviewing and considering this patch. NAK. Again, fix your generator instead. Thanks, pq _______________________________________________ wayland-devel mailing list wayland-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/wayland-devel
