Flushing libgfortran from C

[N.M. Maclaren]

On Mar 3 2011, Tobias Burnus wrote:
>
>For flushing the buffer, you could simply call from C:
>   _gfortran_flush_i4 (NULL);
>
>It should solve at least some of your problems. However, I do not 
>understand why you get an empty stdout. To my knowledge, stdout is not 
>buffered. (Unless that's different in 4.4.)

Er, no.  That's stderr you are thinking of.  Also, even stderr is buffered
on many non-Unix systems - and damn what the C standard says, because it's
unimplementable!

I must say that I am surprised that stderr is unbuffered rather than line
buffered in gcc, even under Linux, but that's another matter.

>(I admit that having a call which properly shuts down the library would 
>be sometimes useful.)

I would say "badly needed".

On this matter, my suspicion is that the Rice people have got it wrong.
I can't think of any good reason to call _exit in that case.  To give
them some slack, here is a description of the C standard situation:

C90 had the exit function, which calls atexit and flushes C's buffers;
C99 added the _Exit function, which does not call atexit functions or
honour signal handlers and it is implementation-defined whether it
flushes C's buffers; this was copied from the _exit function, which was
introduced by POSIX; C1X is adding the quick_exit function, which does
not call atexit functions but does call at_quick_exit ones, does not
honour signal handlers, and it is UNDEFINED whether it flushes C's
buffers.  It is undefined behaviour to call exit from quick_exit or
vice versa, but apparently not to call _Exit from either or indeed
conversely!  God alone knows what implementors will do with that mess.


Regards,
Nick Maclaren.