Hi Tobias,
I don't know why that bootstrapped initially. I fixed the patch (naming a
```
else
/* Prevent warning. */
cdesc = NULL_TREE;
```
obvious) and rerun bootstrap making sure to purge everything beforehand. It did
not break bootstrap on x86_64-linux/f35. Hope it doesn't elsewhere with
Hi Tobias,
ups, sorry, reverted immediately.
Regards,
Andre
On Fri, 28 Jan 2022 10:27:26 +0100
Tobias Burnus wrote:
> Hi Andre,
>
> your patch breaks bootstrapping:
>
> ../../repos/gcc/gcc/fortran/trans-array.cc: In function ‘tree_node*
> structure_alloc_comps(gfc_symbol*, tree, tree
Hi Andre,
your patch breaks bootstrapping:
../../repos/gcc/gcc/fortran/trans-array.cc: In function ‘tree_node*
structure_alloc_comps(gfc_symbol*, tree, tree, int, int, int,
gfc_co_subroutines_args*)’:
../../repos/gcc/gcc/fortran/trans-array.cc:9200:42: error: ‘cdesc’ may be used
uninitialized
Hi Harald,
thanks for the fast review. I have submitted as c9c48ab7bad.
Will wait for two weeks (reminder set :-)) before backporting to gcc-11.
Thank you and regards,
Andre
On Tue, 25 Jan 2022 22:30:22 +0100
Harald Anlauf via Fortran wrote:
> Hi Andre',
>
> Am 25.01.22 um 17:32 schri
Hi Salvatore,
That's correct. It's what ifort does too. Thus far I have found only one or
two minor niggles with the Intel implementation of finalization. Since, as
often or not, the two compilers are frequently used by the same users, I
will attempt to make them compliant with one another. I will
So, you are saying that three calls is the correct number, one for the LHS,
one for the RHS, and one for the deallocation in the main program.
I have (re)-read the standard, I would concur with your interpretation.
On Thu, Jan 27, 2022 at 11:10 PM Paul Richard Thomas <
paul.richard.tho...@gmail.co
