Jean-Christophe Léchenet
Jean-Christophe Léchenet
The same failure on GitLab: https://gitlab.com/jasmin-lang/jasmin/-/jobs/2340273098.
It should. Apparently the problem was the invocation of `nix-shell -p` which was not reproducible. Now we always rely on `default.nix` which pins the versions of the dependencies. So as...
Sorry for that, I'm updating the status of the issue.
What @maximedenes suggests for point 3 of the issue is to keep the two CI systems during a certain transition period (e.g. a few weeks), just to have the time...
One way to have an automatic invalidation of the cache when changing the hash in `nixpkgs.nix` would be to include this hash in the key, either manually or automatically (probably...
@maximedenes you may want to know that not only https://github.com/jasmin-lang/jasmin/pull/109 but also PR https://github.com/jasmin-lang/jasmin/pull/110 touched the GitLab CI. We certainly have to reflect these changes on GitHub's side. Would you...
@vbgl found a way to report the CI status (run on Gitlab) to the corresponding PR (on GitHub). See for instance PR https://github.com/jasmin-lang/jasmin/pull/158. This is not as well integrated as...
@vbgl do you have an idea to improve the current situation?
This may be possible to reject this kind of assignment by type-checking, but this is not trivial, since `stack array = reg ptr array` is allowed in general. One would...
I don't understand the code enough to answer properly, but this is plausible. Ok, I'll go with the lightest solution.