[patch, fortran] PR 80687, 80696

[Janne Blomqvist]

On Wed, May 10, 2017 at 12:14 AM, Thomas Koenig <tkoenig@netcologne.de> wrote:
> Hello world,
>
> it appears that using VLAs also caused the strange errors reported
> in PR 80696.  This, together with Janne's comment, has made me decide
> that maybe using malloc/free is not such a bad idea after all.
>
> So, here is the updated patch.  Running regression-testing with -m32
> at the moment.  OK for trunk if it passes?

Ok.

Have you done any benchmarks to see whether it makes any performance
impact? If so, I suppose you could make some approach where you use a
smallish constant sized stack buffer, and malloc/free only if the size
exceeds the static buffer.


-- 
Janne Blomqvist