Repository navigation
DOC: more nitpick follow up - #15048
WeatherGod merged 3 commits into
Conversation
Sorry, something went wrong.
The Axes classes have a multi-layered internal class structure and we generate documentation for the top-level public classes. However, we also auto-generate property tables by querying the methods on the objects which resolve to the base class they are defined on which leads to incorrect references. This is a very quick hack which special-cases two of the Axes classes and replace their name with just ``Axes``.
|
@ImportanceOfBeingErnest RE #15042 (comment) can you try building the docs on windows from this branch? Sometimes the path includes ':' which can not be random places in windows paths. |
Sorry, something went wrong.
ImportanceOfBeingErnest
left a comment
There was a problem hiding this comment.
I built the docs off this branch locally, and it does not show the error I previously got.
Sorry, something went wrong.
|
|
||
| # sometimes the 'path' can contain ':' which are forbidden on | ||
| # windows, but on posix just passes through. | ||
| path, *post = path.partition(':') |
There was a problem hiding this comment.
uh, how can there be colons in the path, even on posix? a priori this should be a real filesystem path given that we call Path(path).resolve() just below
Sorry, something went wrong.
There was a problem hiding this comment.
Yes, it's confusing that this is named "path", while in reality just being some string which should contain a path at the start.
Sorry, something went wrong.
There was a problem hiding this comment.
As always I'm a bit confused by the magic of escaping, but : appears to be kosher in posix paths:
sharon@21:09 ➤ touch "/tmp/path:o:log:i:cal" sharon@09:03 ➤ echo "hi there" > "/tmp/path:o:log:i:cal" sharon@09:03 ➤ ls /tmp/ | grep path path:o:log:i:cal sharon@09:03 ➤ cat /tmp/path\:o\:log\:i\:cal hi there
Sorry, something went wrong.
There was a problem hiding this comment.
It's certainly kosher (anything but / and \0 are for posix), but how is that relevant here? if we're running on linux, then the path is what it is and it will become relative to the mpl root just below (in relative_to), and there are no directories in mpl with colons; if running on windows, get_source_line should be returning a real path on the windows filesystem and that has no colons.
Sorry, something went wrong.
| attrs = sorted(self._get_setters_and_targets()) | ||
|
|
||
| names = [self.aliased_name_rest(prop, target) | ||
| names = [self.aliased_name_rest(prop, target).replace( |
There was a problem hiding this comment.
it doesn't have to be this PR, but perhaps we should just fix Axes.__module__ to be matplotlib.axes, not matplotlib.axes._base? see numpy/numpy#12382 for similar stuff on numpy's side.
Sorry, something went wrong.
There was a problem hiding this comment.
That seems reasonable, but also push to another PR.
Sorry, something went wrong.
This fixes an exception just merged to master and the other half of
the issues @ImportanceOfBeingErnest is having in #14917.
PR Summary
PR Checklist