Update handler for mkdocstrings 0.28 #23
Conversation
|
Oh, let me fix linting. |
| Returns: | ||
| The combined options. | ||
| """ | ||
| return {**self.default_config, **local_options} |
There was a problem hiding this comment.
I don't think we need to deepcopy here, but I've been bitten by this previously. This creates a new dict, but with references to mutable structures from both default config and local options. The only time config is mutated is below in render with config["members_order"] = Order(config["members_order"]), which I believe doesn't cause any issue.
We can still deepcopy here just to be safe. Let me know.
There was a problem hiding this comment.
Maybe the real solution is to not mutate the config at all. Otherwise yes, let's use deepcopy.
| try: | ||
| data = self.collect(identifier, {}) | ||
| except CollectionError: | ||
| return () | ||
| return data.path.as_posix(), *(p.signature.name for p in data.procedures) |
There was a problem hiding this comment.
I have replicated previous behavior here, but I have to note that this seems incorrect to return both the path of the file, as well as all the procedure names within it. It's possible that the previous method (get_anchors) was confusing, and the "anchors" concept not well explained. The renaming to get_aliases is to make it more clear: it's not really about HTML anchors, it's about obtaining all the different "aliases" ("locations") for a given object, to correctly populate mkdocs-autorefs data. I know nothing about VBA, but surely Procedure1 is not an alias of Procedure2?
Aliases (and the identifier) are supposed to be strings that you can pass to collect.
Here I'd either just return (data.path.as_posix(),) (or even an empty tuple ()), or add a way to support collecting Procedure1 without a file path (or these procedure objects should expose fully qualified names that can be collected).
| @@ -28,7 +28,7 @@ | |||
| "setuptools_scm", | |||
| ], | |||
| install_requires=[ | |||
| "mkdocstrings>=0.26.1,<1", | |||
| "mkdocstrings>=0.28,<1", | |||
There was a problem hiding this comment.
0.28 is not published yet so it will probably fail in CI.
Following discussion in mkdocstrings/mkdocstrings#727.