Florian Deconinck

Unreliable information, please disregard ~~Hi everyone.~~ ~~Here's a workaround dug from the depth of [openssh bug list](https://anongit.mindrot.org/openssh.git/commit/?id=8d0578478586e283e751ca51e7b0690631da139a)~~ ~~Tested on windows 10 2h22 / GitBash~~ ~~* do your `ssh -MNf` with...

> @FlorianDeconinck Not clear how it is supposed to work and what you were testing. `-O` option requires setting ControlPath (`-S` option). Setting ControlPath makes OpenSSH fail, as it is...

@BenWeber42 / @phschaad / @tbennun Gents, The bug on compilation has been solved, so I am resuming work on this. Unfortunately, it seems settings have changed and the Github actions...

Hello, Looking to also build GDR_COPY on SLES (12 or 15). @bcfriesen did you get the chance to test the fix? Still no official support for SLES @drossetti ? Thank...

Fixing test it seem my original logic is flawed. Type hinting an array as `Optional` doesn't tell us anything more than "we should check at runtime". The way parsing is...

Seems fair, I will write the code that go down that path and we can reason on it.

TL;DR: to make sure AOT based on type hints only is supported, no inference can be done, including in the `OptionalArrayInference` which might be a bug. Going deeper on this,...

The only option left is to restrict this pass to nested calls made with clear `None` type e.g. ```python def nested(A. B): .. @dace.program def program(A, B): nested(A, None)

Agreed this would solve the original issue in Pace. That said I believe the nested call version described above should still resolve at parse time. Now I don't have non-ugly...

Behavior broken in `main` as of commit ` b7325e`