Closed tacaswell closed 1 month ago
FWIW I'm strongly +1 to requiring this filter, both for avoiding the time and bandwidth of irrelevant downloads, plus aligning better with 'explicit is better than implicit'.
I could see value in some sort of special all
behavior, though, if it were useful for internal testing purposes.
+1 to the behavior, do you think we want a short but existing deprecation period even if the package is brand new ?
do you think we want a short but existing deprecation period even if the package is brand new ?
The project is still at v0.0.4 on PyPI, with a couple dozen downloads per day. While it would be strictly correct to deprecate and then change, I vote -0.5 to a deprecation period given the newness and light uptake as yet. But, I vote for emitting a custom-crafted error message when the filter is missing (perhaps with a link to more information, if that makes sense) so that users don't have to search around for the fix.
It's not like it will give wrong behavior or break live code. Some users' docs builds will fail; they'll realize they need to add the config item, and then all should be well.
Ok, I pushed a suggestion of new behavior, let me know.
I'm normally very (very) 👎🏻 on CalVer, but in this case I think it makes sense as the functionality is thin and the thing that will actually change is the data (and it should be something that should be left un-pinned in downstream libraries).
For calver, let's discuss on #4 there are a few possibilities for calver format.
I made all keys lowercase (pillow, pypug), and added a test they are all lowercase.
Ok, let's try that.
I think defaulting to expecting a list is good because:
conf.py
what external links should workThe con is that it is a bit more verbose, but still an improvement over every project maintaining the urls.