Pin Sphinx 3.x to docutils <0.17 #9071
Conversation
|
Could you rebase this into 3.5.x branch? Then I'll merge this soon. Note: 3.5.4 will be the last release of 3.x series. So I will merge this to the package. But I will not merge this into 4.x series because it should go forward. No reason to restrict the version. |
ericholscher
force-pushed
the
pin-docutils
branch
from
2d0ad02
to
2433db8
Compare
This will address the issues that the latest docutils release caused. I think thi si a good practice in general to make sure we have a defined range of docutils versions, given that they might change in the future. Having this defined will mean that 3.x versions of Sphinx will always work, even when docutils has advanced with additional backwards incompatible features. Refs sphinx-doc#9065 sphinx-doc#9063 sphinx-doc#9061 sphinx-doc#9051
ericholscher
force-pushed
the
pin-docutils
branch
from
2433db8
to
025f26c
Compare
|
@tk0miya Done 👍 |
|
I do think that specifically defining a upper bounds of docutils support in Sphinx is a good ongoing policy, but that's a larger discussion. It certainly has downsides, but it could be useful to have Sphinx increase the docutils dependency after it's been tested against the currently released version. Currently I don't believe docutils is testing Sphinx in it's releases, so it seems we'll hit this issue again whenever docutils makes a useful but breaking change. |
|
Thanks! |
This will address the issues that the latest docutils release caused.
I think this is a good practice in general to make sure we have a defined range of docutils versions,
given that they might change in the future.
Having this defined will mean that 3.x versions of Sphinx will always work,
even when docutils has advanced with additional backwards incompatible features.
We've had more than 10 support requests about this in the past week, so I think this is hitting a lot of users throughout the Sphinx ecosystem and it's very hard to debug if you don't understand what is happening.
Refs #9065 #9063 #9061 #9051