Janick Gerstenberger

Results 6 comments of Janick Gerstenberger

Unfortunately it is not as simple. Both `pytest-cases` and `pytest-lazy-fixture` have bugs with transitive dependencies, because `pytest-cases` explicitly and `pytest-lazy-fixture` implicitly assume that the dependencies form a tree. In reality...

There are a few plugins, e.g. pytest-lazy-fixture, pytest-cases that manipulate `item.funcargs` to inject (lazy/transitive fixture) values. Will this change will eventually break them? I am also slightly confused. What is...

> Well,those are hacks and they really need a better api from pytest, > > As long as we keep the mess good things are blocked > > Id like...

> Before approving, we should check @jgersti comment regarding pytest-lazy-fixtures. Specifically how it uses `funcargs` and whether `getfixturevalue` is a viable replacement for it. See also @jgersti's post here: #11412...

> @jgersti Only had a quick look now at lazy-fixtures, I do think this will break (after the change is done fully): > > https://github.com/TvoroG/pytest-lazy-fixture/blob/18ec85edb5e27c933733f748c685b2fd083198d7/pytest_lazyfixture.py#L50-L54 > > That's assuming `is_lazy_fixture`...

While looking through pytest i found only one location where `item.funcargs` is read and that was to fill the initial test function call. Additionally the lines you hightlighted can be...