For some reason, I decided to give a little more thought to this thread.
I'm not sure anyone cares, but I thought I'd point out some things.
On Tue, 20 Sep 2005, Joel E. Denny wrote:
On Mon, 19 Sep 2005, Paul Eggert wrote:
"Joel E. Denny" <[EMAIL PROTECTED]> writes:
int* p = malloc(n*sizeof(int));
Are you saying g++-4 and g++-5 don't catch it as they should?
I just tried g++ 4.0.1, which is the current release series. It catches
the above missing cast as it should.
Someone mentioned g++-5 earlier in this thread. I didn't think to
question it before, but g++-5? Is this a typo?
Just out of curiosity, why is your test suite calling g++ for that test
case anyway?
Some people like to use C++ compilers when C compilers are called for.
Personally I don't think it's worth catering to such usage, but I
guess others disagree.
I'm certainly guilty of using g++ when compiling bison-generated code...
even from yacc.c and glr.c. That seems necessary given that the C++
skeletons aren't ready. However, when building bison itself, I don't
understand why g++ would be used. I guess it's a moot point, but it's
curious.
I now finally realize that make maintainer-check runs the test suite with
g++. This makes sense to me given that many users, such as myself, use
bison's yacc.c and glr.c with C++. Moreover, many of us will probably
continue to do so long after the C++ skeletons are complete. It takes a
while to upgrade old code.
Who knows why, but I had it in my head that someone had actually
configured with g++ in place of gcc for all of make and make check.
That's what I was actually perplexed about.
Now that I understand maintainer-check, I'll be more careful with my test
cases.
Sorry for the stupid questions.
Joel
