Re: [lmi] wxmsw-2.9.0 (svn trunk) anomaly

[Greg Chicares]

On 2009-03-14 19:15Z, Vadim Zeitlin wrote:
[...]
>  I think you might be not be taking into account the fact that
> -Wno-uninitialized _suppresses_ the warnings about uninitialized variables
> which are enabled by -Wuninitialized or by -Wall (used by wx). So the above
> does explain it: wx build does not use -Wno-uninitialized and hence
> generates the warnings while LMI build suppresses the warning with
> -Wno-uninitialized and so no warnings are output.

Ah. You're right.

I had thought you meant that this flag caused the warnings in the
wx build, but actually you meant that it explains why I don't see
the same warnings in the lmi build.

[The ultimate cause is this compiler defect
  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22207
anyway.]

So, restoring the end of your message:

>>>  My point was that it would be nice to avoid having to
>>> suppress this warning as it can be, and often is, useful and if using
>>> -isystem could suppress all the false positives it might be worth it.

Yes, I agree that it would be nice to suppress them more delicately
in lmi. But perhaps Cygwin will soon publish a native cross-compiler
(an oxymoron, but I'm sure it's transparent to you); and if I save
time by not working around this gcc-3.4.4 defect, that makes more
time for moving to gcc-4.x, or for considering your patches that fix
problems detected by another compiler, or for fixing the compilo I
somehow introduced for Comeau....