[Patch, Fortran] PR 42353: [OOP] Bogus Error: Name 'vtype$...' at (1) is an ambiguous reference ...

[IainS]

On 31 Dec 2009, at 16:41, Janus Weil wrote:

>>> Harald, is it possible for you to try out the fortran-dev branch  
>>> with
>>> this patch? If yes, it would be good to know if it works on your
>>> codes.
>>
>> It took me some time to get a build of fortran-dev, but I succeeded.
>>
>> However, the patch fails with the original testcase in the PR:
>>
>> % /opt/gfortran/dev/bin/gfortran gfcbug96.f90
>> gfcbug96.f90:43.23:
>>
>>  use concrete_gradient
>>                       1
>> Error: The element in the derived type constructor at (1), for  
>> pointer component '$extends', is DERIVED but should be DERIVED
>
> Argh. Should have checked this myself, sorry. Here is an updated
> patch, which kills this error by adding a special case for vtypes in
> 'gfc_compare_derived_types'. Hope this one works better. I made sure
> that it has no testsuite regressions at least.


apropos "gfortran.dg/gomp/recursion1.f90"

I think you might also need something to tell gfortran where to find  
libgomp for uninstalled testing

This is the first "dg-do run" test added  and it fails on all my  
systems (when uninstalled) owing to not being able to locate libgomp

I've just made this small patch - which is one way - but you might  
prefer a different solution (it could be considered heavy-handed  
checking each time).

gcc/testsuite/Changelog:
2009-12-31  Iain Sandoe  <iain.sandoe@sandoe-acoustics.co.uk>

	* libs/gfortran.exp: Check for -fopenmp option and add libgomp paths  
if found.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: gfortran-exp-diff.txt
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20091231/7770536f/attachment.txt>
-------------- next part --------------



Happy New Year,
Iain