This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH, Fortran] Some work on PR 29835: Improve the error message for an invalid format


FX wrote:
The two execute-tests don't work anyway--any suggestions on how to make those do what expected? Otherwise I'll remove them if that's ok rather than leaving them in without any effect. Or should I fix the output-clauses accordingly, leave them and replace the XXX comment by an informative one stating that and why they are not really testing the runtime behaviour at the moment?

I don't know why it's failing. Have you looked at the testsuite log file (${builddir}/gcc/testsuite/gfortran/gfortran.log)? It'd be better to have runtime checks, if you manage to have them work. If you can't, well, I don't believe they should be included at all.

It's because the Fortran runtime-error exists with non-zero exit code I suppose; and the output is printed to stderr, so I'm not even sure if we could match it if the program would not "fail".


Any ideas? Otherwise I'll take the tests out completely.

BTW, I won't do anything to the PR as it is not fixed fully, ok? Or should I mark it fixed and comment that further changes won't be done or something like that?

As a general rule: after committing, please add a comment saying what you've done and what still needs to be done to close the PR. In that case, we don't want to implement the Q descriptor, so unless I'm mistaken, you should indeed close it (another maintainer may want to reopen it if they see fit, but I think it's highly unlikely).

Ok. In this case, I'm more thinking about improving the error locus further (the current state is a compromise, because locating to the character would mean much effort) than about implementing Q.


While I'd rate this probably "won't fix" or at least not soon (because of the minor benefit to effort), I'd not say the PR is fixed completely as stated. But I'll resolve it to fixed, commenting what I did and what is still missing together with this reasoning.

That's also why I left the TODO in the source; but I could remove this as well if you want.

Cheers,
Daniel

--
Done:     Arc-Bar-Sam-Val-Wiz, Dwa-Elf-Gno-Hum-Orc, Law-Neu-Cha, Fem-Mal
Underway: Cav-Dwa-Law-Fem
To go:    Cav-Hea-Kni-Mon-Pri-Ran-Rog-Tou


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]