Sebastiaan Huber

Just to mention that I stubbed my toe on this very problem today. My tests miraculously were failing when applying a fixture using `mark` but when used directly it was...

Found the source of the issue none of the `zend_entry_function` stubs were defined in `php_git2.c`. Also the implementation of `git_branch_upstream` was missing from `branch.c`

Seems like this would have to wait for https://github.com/python/importlib_metadata/pull/342 to be merged and released

We added this back in the day because the performance of adding to groups is really bad. The raw SQL will often take order(s) of magnitude less compared to the...

Yep, still a problem, `skip_orm=True` is 50 times faster. ```python In [1]: group = Group('test') In [2]: group.store() Out[2]: In [3]: nodes = [Data().store() for _ in range(1000)] In [4]:...

This PR so far just adds the regression test showing the problematic behavior introduced in v2.0. The actual fix still needs to be added. See the issue description for analysis...

I support this feature request. It is indeed not easy to invoke the `click` commands directly from Python and actually directly invoking a subcommand will have unexpected results. For example,...

I am not sure if this can be considered a bug per se. In a sense, it only makes sense for an expression that is evaluated in the `while()` to...

Thanks @dev-zero . I think it is kind of nice that the `Scheduler` is independent of any direct AiiDA concepts, so frankly I would be a bit hesitant to require...

I just had a read through the entire stack of issues and the discussion, including on `aiida-quantumespresso` to try and understand the use case and I think I got it,...