> -----Original Message----- > From: ext Andy Lester [mailto:[EMAIL PROTECTED] > Sent: 28 April, 2005 17:29 > To: Hietaniemi Jarkko (Nokia-NVO/Helsinki) > Cc: perl5-porters@perl.org > Subject: Re: (in blead) a missing const in a prototype > > > On Thu, Apr 28, 2005 at 08:52:06AM +0300, > [EMAIL PROTECTED] ([EMAIL PROTECTED]) wrote: > > > If PADOFFSET was a pointer, it would be different. > > > > All true. But now if the declaration has const but the proto doesn't, > > compilers will whine. (E.g. Visual C does.) > > Oof, I didn't know. gcc doesn't mind.
In general, I wouldn't trust gcc to tell that much, even with -Wall. The -Wall does *not* mean "all the warnings". Gcc 3.x and 4.x are getting better at whining, though. Especially naughty gcc is at silently enabling gcc-specific extensions, and not warning about their portability problems. Some extra warning/strictness options, for varying levels of pain (don't try these all at once with the Perl sources or you will go blind :-) -Waggregate-return -Wbad-function-cast -Wcast-align -Wcast-qual -Wconversion -Wdeclaration-after-statement -Wmissing-prototypes -Wnested-externs -Wpointer-arith -Wredundant-decls -Wshadow -Wsign-compare -Wstrict-prototypes -Wwrite-strings -ansi -pedantic >