amol- edited a comment on pull request #11726:
URL: https://github.com/apache/arrow/pull/11726#issuecomment-1010050324
@jorisvandenbossche
Looking at `numpydoc` there are a few things that come to my mind:
1) We should avoid the form `list of int` because `of` is nowhere recognised
for containers, containers are dealt with `list[int]` form. Se should probably
migrate all cases to that form because it solves some misunderstandings from
numpydoc
2) numpydoc is somewhat buggy, `or` is only detected as a literal in some
cases, in some other cases is parsed as a reference and becomes ``obj:`or` ``.
3) Some words it's correct that they are identified as generic references
(`function`, `Mapping`, `iterator`) as they are in practice
interfaces/protocols, so they might not have a reference in the Python docs but
we want to identify them as code anyway.
My suggestion would be to apply (1) and solve (2) using
`numpydoc_xref_ignore = {"or", "and", "of", ",", "default", "optional"}`. For
`bool`, `True`, `False`, `object` I think it would be wrong to place them in
ignore as they are currently correctly detected and linked to the Python
documentation.
That solves the majority of cases, few instances of problems still remain
like
where `depending on` still remains as code
or
where the whole parameter is misinterpreted for unknown reasons.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
