Use latest versions of sphinx and sphinx-rtd-theme #2802
Conversation
humitos
added
PR: ready for review
and removed
PR: work in progress
|
Does conda have the proper packages now? We could always setup a |
| 'mock', | ||
| 'pillow>=3.0.0', | ||
| 'sphinx_rtd_theme==0.1.7', | ||
| 'sphinx-rtd-theme<0.3', |
There was a problem hiding this comment.
One of these is underscore, the other is dashes. We should be consistent.
There was a problem hiding this comment.
The latter is more consistent because _ is replaced by -: http://stackoverflow.com/questions/19097057/pip-e-no-magic-underscore-to-dash-replacement
| @@ -193,12 +193,12 @@ def install_core_requirements(self): | |||
|
|
|||
| # Use conda for requirements it packages | |||
| requirements = [ | |||
| 'sphinx==1.3.5', | |||
| 'sphinx==1.5.3', | |||
There was a problem hiding this comment.
Should we do sphinx<1.6 on both?
There was a problem hiding this comment.
Wouldn't it be nicer to pin Sphinx because it's so vital that bumping the version should be challenged every time by a deliberate action (PR) ?
There was a problem hiding this comment.
@benjaoming Given the frequent release cycle for Sphinx, semver, as well as the workload on the core team here, I concur with @ericholscher's recommendation to float up to the next major release. Moving to support Sphinx 1.6 would be a deliberate action.
There was a problem hiding this comment.
@willingc good point, and actually looking at a search result on the Sphinx issue tracker reveals that their 1.5 series does not seem to carry many issues specific to patch versions, so it's low risk.
This issue in Sphinx 1.5.2 might be an example of how RTD would have to skip a patch release due to an issue, however that could be fixed reactively when necessary.
There was a problem hiding this comment.
Also +1 on <1.6 here. The Sphinx devs are very deliberate with versioning these days, I trust that bugfix versions won't have a strong effect here. Additionally, the number of issues we've had from indescriminately selecting versions are outweighted by the number of bugfixes we haven't adopted.
|
@agjohnson @ericholscher I'm marking this as |
|
I'm closing this one because of #2876, where we discuss at PyCon Sprint about this issue. |
I upgraded the conda pinned versions as in pip environments.
Again, we should consider to remove all of this pinned versions (#2566 (comment)). In the meantime, I upgraded them.
This need testing with a project that uses this.
Closes #2714
Closes #2584