Re: [Patch, Fortran] PR 39577 - fix -fcheck=recursion

[Tobias Burnus <burnus at net-b dot de>]

Dominique,

my current conclusion is that today's version of the patch works and it
has no known problems. Or have I misread something?

> With your new patch, most of the "Recursive call to nonrecursive procedure"
> I saw are now gone. I think the error is right on the first two tests at the
> end of this mail where the recursivity is obfuscated.
I'm not sure whether I understand the last part. Using all your three
test cases [(a) to (c)], I get with -fcheck=recursion a run-time error.
That matches my expectations as the in the programs procedure is indeed
called recursively without having a "recursive" attribute.

> (a monster coming form comp.lang.fortran, likely authored by James Van Buskirk!-)
That one is cool - it tests whether the check also works with threading - and it does.


> Note that the ICE with -fwhole-file on the test given in:
> http://gcc.gnu.org/ml/fortran/2009-03/msg00310.html
> is back despite the Paul's patch in
> http://gcc.gnu.org/ml/fortran/2009-03/msg00312.html
>   
That seems to be some kind of endless loop/recursion. But it does not
seem to be related to my patch.

>> I think there are still a couple of bugs in gfortran with
>> ENTRY as it is only rarely used.
>>     
> I see pr34500 (ICE) and pr34663 and pr23280 and pr37134 (for missing
> debug info).
>   
Well, I was more refering to the problem in general and not talking
about specific bugs; especially I was thinking about undiscovered ENTRY
bugs. Not that I won't work on ENTRY but it does not necessarily have a
higher priority than other bugs.

Thanks again for the testing.

Tobias