[Bug testsuite/25241] DejaGNU does not distinguish between errors and warnings

["manu at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #41 from manu at gcc dot gnu dot org  2007-03-24 17:55 -------
(In reply to comment #38)
> For comment #32 I get the failure but don't understand the problem; the regexp
> matches the message, doesn't it?  This doesn't seem any different from other
> Fortran tests but it's the only one that fails; perhaps there's something
> special about specifying the line number in the test directive, or that the
> message comes from a file with a different name.  I'll stare at it again later.

No, it doesn't match. If you look closely to this message compared to other
fortran messages:

* Broken match:

Warning: /home/manuel/src/trunk/gcc/testsuite/gfortran.dg/badline.f:2: file
src/badline.F left but not entered

* Correct match:

/home/manuel/src/trunk/gcc/testsuite/gfortran.dg/blockdata_1.f90:17.7:

 common j ! { dg-warning "cannot contain blank COMMON" }
      1
Warning: BLOCK DATA unit cannot contain blank COMMON at (1)


The difference is that in the first case the filename plus position is between
the "Warning:" and the message to match. Actually I think the deja-gnu or some
*.exp file does some preprocessing of the output. For example, if you look at
the output of gcc/testsuite/gfortran.dg/continuation_1.f90

/home/manuel/src/trunk/gcc/testsuite/gfortran.dg/continuation_1.f90:11.10:

         world!" ! { dg-warning "Warning: Missing '&' in continued character co
         1
Warning: Missing '&' in continued character constant at (1)

FAIL: gfortran.dg/continuation_1.f90  -O0   (test for warnings, line 11)
FAIL: gfortran.dg/continuation_1.f90  -O0  (test for excess errors)
Excess errors:
/home/manuel/src/trunk/gcc/testsuite/gfortran.dg/continuation_1.f90:11:
Warning: Missing '&' in continued character constant at (1)

The message shown after "Excess errors" is slightly different than the
original. 

I don't think that we should handle this corner case in our patch: either it is
handled by whatever function preprocesses fortran output or the output itself
is changed by the fortran front-end. We could as well use the original
dg-warning / dg-error directives for fortran.

So, gcc/testsuite/gcc.dg/20041213-1.c is the issue number 1 now. I think it
should match without modifying the testcase.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25241