IMPORT or C Binding bug?

[Christopher D. Rickett]

On Thu, 28 Jun 2007, Steve Kargl wrote:

> On Thu, Jun 28, 2007 at 03:28:08PM -0600, Christopher D. Rickett wrote:
>> i don't think the problem has anything to do with C binding.  for example,
>> the following simplified code behaves the same:
>>
>
> I agree with your analysis.  I found PR 31154 after I posted the
> code.  BTW, I posted a call for real-world codes using ICB on
> comp.lang.fortran.  I've gotten a few responses so I trying to
> break your patch :-)

thanks for the testing.  i'll nervously wait to see what bugs get reported 
back and hope it's only a few.  :-)

> One response stated:
>
>   I hope gfortran implementation of ISO_C_BINDING permits the use of
>   private components in a derived type with BIND(C) attribute (to my
>   best knowledge g95 doesn't ).
>
>   In a recent discussion, Richard Maine clarified that
>
>   "a derived type with the BIND(C) attribute"
>
>   1) is allowed to be a "PRIVATE type" and/or
>   2) is allowed to be a type with "PRIVATE" components.
>
>   The second one is used in the above Pthread Interface implementation
>   (which makes perfect sense).
>
> I checked your testsuite codes (with a quick glance) and could not
> find anything that matches this description.  Should we add a few
> more tests?

i'm pretty sure this will be allowed by gfortran -- i can't think of 
anything i did to prevent it.  i don't think the public/private attribute 
for type components was implemented yet in gfortran when i did the bind(c) 
work, so adding more tests now would probably be good.

Chris