Le 13/08/2026 à 10:57, Mikael Morin a écrit :
Le 13/08/2026 à 09:50, Mikael Morin a écrit :
Le 12/08/2026 à 20:29, Steve Kargl a écrit :
On 8/12/26 10:24, Mikael Morin wrote:
Hello,
this caused a regression on something else I was working on.
I found a failing example on the trunk, and here we are.
Fortran-tested on aarch64-unknown-linux-gnu. OK for mainline?
Mikael
Yes. The patch verges on "obviously correct".
The change itself is definitely trivial; the context and impacts are
less so. Thanks for the review, pushed as r17-3254-gfbc358491eac26.
There is a similar problem in the CO_BROADCAST implementation for non-
character arrays, and I'm currently testing the attached fix. It could
be extended to the character case as well, but there is a strlen call
that I'm not brave enough to touch. Will push if tests are clean.
Now done: r17-3255-g26ba7ae2c5dafa
