Re: gfortran "partial" recursive I/O support

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

FX wrote:
> >  So, until we have implemented derived type I/O we can only claim partial
> > recursive I/O.
> >     
> I don't think that's it: Cray, g95 and Intel all have no support for
> derived-types I/O and yet have a "yes" for recursive I/O.
>   

The table follows ftp://ftp.nag.co.uk/sc22wg5/N1601-N1650/N1648.pdf, 
which has:

"4.10 Recursive input/output -- A recursive input/output statement is 
one that is executed while another input/output statement is in 
execution. We met this in connection with derived-type input/output 
(Section 4.1). The only other situation in which it is allowed, and this 
is an extension from Fortran 95, is for input/output to/from an internal 
file where the statement does not modify any internal file other than 
its own."

Our conclusion was: "Yes, but" and playing cautious we put a "partially" 
there. Maybe we can come to a new conclusion.  See for the initial 
discussion:

http://gcc.gnu.org/ml/fortran/2007-02/msg00109.html
http://gcc.gnu.org/ml/fortran/2007-02/msg00086.html

I do not know whether there is a problem on some systems or whether 
there could be a race condition. For the target-dependency, we could 
simply add a test case to test suite and wait ;-)

Tobias