This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] C++ FE warning control cleanup
Lee Millward <lee.millward@gmail.com> writes:
| I'm not so sure that it is my patch which has caused these failures,
| when running the test suite with the current revision I'm experiencing
| those same failures whereas when I ran the test suite after I created
| the patch I experienced no new failures.
bootstrap and regtest just finished. I can definitely see the new
failures. They are procuded by "excess errors". Many of them are of
the form:
/home/gdr/redhat/trunk.gcc/gcc/testsuite/g++.dg/charset/extern2.cc:5: warning: deprecated conversion from string constant to 'char*''
Notice the double apostrophe at the end. Granted, you did not
introduce that bogus apostrophe, but your hunk to cp/typeck.c did
"touch" it:
- warning (0, "deprecated conversion from string constant to %qT'", totype);
+ warning (OPT_Wwrite_strings, "deprecated conversion from string constant to %qT'", totype);
So I suspect that you may have uncovered some latent bugs (in either
the testsuite or the diagnostic machinery) with passing non-zero
option code.
I also see:
/home/gdr/redhat/trunk.gcc/gcc/testsuite/g++.dg/init/ctor4.C:11: warning: non-static reference 'int& bar::a' in class without a constructor
which is a warning "touched" by your hunk:
- warning (0, "non-static reference %q+#D in class without a constructor", x);
+ warning (OPT_Wextra, "non-static reference %q+#D in class without a constructor", x);
The reason I suggested you ran the testsuite -- apart from the fact
that it is our commit procedure -- is that I do remember when I did
similar patch I saw those failures and since I did not have time then,
I decided it was someone else's (hi DJ! :-) job.
Another set of regression seems to be caused by SSA failure.
-- Gaby