[sane-devel] Heads up: fixing compilation warnings

Julien BLACHE jb at jblache.org
Wed Jun 23 16:32:02 UTC 2010


I've just pushed out a series of 50 patches fixing compilation warnings
in a number of backends.

Most of these patches are trivial, so no ill-effects are anticipated in
the general case. If your backend, or a backend you use, is the subject
of one of these patches, please take a couple minutes to test it.

Please, pretty please, fix warnings in your backends! It's not like the
compiler complains just for the sake of complaining. Those warnings can
cause real bugs on some architectures.

Left over:
 - 1 warning in hp5400, there is a comment in the source
   (hp5400_internal.c) about a (correct) fix but someone needs to try it
   out and make sure it doesn't break calibration (no idea why it would

 - 1 warning in niash, due to some disabled options. It'd be nice if we
   could find a better way to disable options than fiddling with enum
   values triggering compiler warnings on array bounds.

 - constness issues. Mostly affecting option handling code. We probably
   need to use a different definition inside SANE (non-const) and in the
   public API (const) to fix this. Other cases need some more work to
   fix the const poisoning that is plaguing the codebase.

FWIW I've built sane-backends with both gcc and clang while fixing those
warnings. I haven't actually tested the resulting binaries built with
clang, though. No reason why they wouldn't work.


Julien BLACHE                                   <http://www.jblache.org> 
<jb at jblache.org>                                  GPG KeyID 0xF5D65169

More information about the sane-devel mailing list