Hi Andre,

I have posted my results with your patch (and those for pr63851) at

https://gcc.gnu.org/ml/gcc-testresults/2014-12/msg02408.html.

I don’t see any problem with unlimited_polymorphic_2.f90. However the character
lengths are now wrong (they are 0) with your old patch for pr60289 at

https://gcc.gnu.org/ml/fortran/2014-08/msg00022.html.

I have also noticed that you don’t comply to the GNU policy about spaces,
in particular there should be no space at the end of a line, see the patch for
gfortran.dg/unlimited_polymorphic_18.f90.

Thanks for working hard on these issues,

Dominique

> Le 18 déc. 2014 à 19:41, Andre Vehreschild <ve...@gmx.de> a écrit :
> 
> Hi all,
> 
> here is my next try on proposing a patch for the issue in pr60255. It took me
> quite some time to understand the intricacies with handling variables
> associated in a select type. I think I got most of the issues fixed now:
> 
> - Added generation of _len component for each unlimited polymorphic pointer.
> - Removed (my own) _len component creation routine.
> - Removed the double underscore in get_len_component().
> - Associating an unlimited polymorphic entity to a deferred char array now 
> lets
>  the deferred char array use the actual string length from the '_len'
>  component of the unlimited polymorphic entity for the charlen instead of the
>  size component of the vptr.
> - Removed: Generating a special vtab name for deferred strings. A deferred
>  string assigned to the unlimited polymorphic entity is now stored as having 
>  charlen zero again.
> - Basic support for char array arrays (No stuttering here) in u-poly 
> variables.
> 
> Bootstraps ok on x86_64-linux-gnu. Comparing regtests I get a difference in
> unlimited_polymorphic_2.f90 that I don't understand yet. May be that is only,
> because one error message disappeared.
> 
> Attached is the full patch for trunk and a delta patch for those of you who
> already have my pr60255_3 added.
> 
> I don't provide a changelog entry yet, because I think review will find some
> issues still to fix. So, comments welcome!
> 
> Regards,
>       Andre

Reply via email to