Removal of -pedantic and -pedantic-errors from acinclude.m4.in
zipzippy at sonic.net
Sun Aug 17 11:11:38 BST 2003
On Sat, Aug 16, 2003 at 01:04:32PM +0200, Dirk Mueller wrote:
> I'm very much aware of that. However, I still didn't see a case where things
> actually go wrong for you. Like, config.log of an incorrectly failing test.
What are you talking about?
I just gave a specific example (actually two). The random and srandom
tests FAIL because of some errors in stdlib.h. These ERRORS are triggered
by the pedantic flag. These ERRORS prevent dcopserver from compiling, and
will probably prevent just about every other piece of KDE from compiling
if I were to run make -k.
I don't have logs right now, because gosh I had to go back and FIX
acinclude.m4.in so I could build kdelibs (surprisingly enough, the fix was
to remove -pedantic from acinclude.m4.in).
The point is, using -pedantic triggers improper (warnings -> errors)
behavior on the part of gcc. This is already known. Saying that code
should avoid tripping the GCC's inner pedant doesn't do anything to help.
The fact is -pedantic causes trouble with gcc 3.3.x. Check out the
kde-freebsd mailing lists, this is well known. I believe their solution
was to remove -pedantic from acinclude.m4.in all together (as
--disable-warnings obviously does not work).
More information about the kde-core-devel