[PATCH] load_cursor_argb is supposed to return a Bool, not void

Keith Packard keithp at keithp.com
Fri Apr 18 11:38:29 PDT 2014

Michael Thayer <michael.thayer at oracle.com> writes:

> On 14/04/14 21:35, Keith Packard wrote:
>> Julien Cristau <jcristau at debian.org> writes:
>>> Only since
>>> http://cgit.freedesktop.org/xorg/xserver/commit/?id=901fbfbbbd71c0d82080957f8ba09eebbc786f2b
>>> Which could probably have used a different name to avoid silent
>>> breakage.
>> Yeah, that probably would have been a better change.
> Sorry, my fault: it didn't occur to me that this would fail to break 
> driver builds.

Turns out to be tricky to make this warning into an error because it's
enabled by default, and not with an explicit -W switch. The only way to
get it treated as an error is to turn on -pedantic-errors, but that also
turns on -pedantic, which we don't want as that complains about all of
the GCC extensions that we use. So, you turn that back off with

        CFLAGS = -pedantic-errors -Wno-pedantic

We're definitely not ready for this though; it also turns casts from
'const char *' to 'char *' into errors. Getting one 'this can never
work' bug converted into a raft of 'this essentially always works'
spurious error messages just isn't helpful.

> Shall I submit another patch to change names?  I presume that would
> still be allowed after the end of the release window.

I think so; the alternative is to break hardware cursors on any driver
which hasn't been updated. Anyone else have a second opinion?

keith.packard at intel.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 810 bytes
Desc: not available
URL: <http://lists.x.org/archives/xorg-devel/attachments/20140418/6dabf424/attachment.sig>

More information about the xorg-devel mailing list