should update CSS to reflect new docutils 0.17 HTML5 tags

[drammock]

The HTML5 writer in docutils 0.17 writes the .. sidebar:: directive as the HTML5 aside tag (used to be div). This means that the rule in basic.css_t for styling those elements no longer applies, since the rule is written for div.sidebar instead of aside.sidebar:

sphinx/sphinx/themes/basic/static/basic.css_t

Lines 322 to 331 in f7768d8

	div.sidebar {
	margin: 0 0 0.5em 1em;
	border: 1px solid #ddb;
	padding: 7px;
	background-color: #ffe;
	width: 40%;
	float: right;
	clear: right;
	overflow-x: auto;
	}

According to the docutils changelog linked above, other tags now being output are main, section, header, footer, figure, and figcaption. I haven't systematically checked for where / whether similar CSS mismatches are happening as a result, but I'd guess there are a few more besides the div -> aside issue that I ran into.

EDIT: crossref to mne-tools/mne-python#9031 where there are screenshots of a sidebar directive that is failing to get styled properly.

[Daltz333]

A ping on this. docutils 0.17 has just gotten published and now breaks generation

[mgeier]

I've just created #9059, which hopefully fixes this issue.

@drammock

According to the docutils changelog linked above, other tags now being output are main, section, header, footer, figure, and figcaption. I haven't systematically checked for where / whether similar CSS mismatches are happening as a result, but I'd guess there are a few more besides the div -> aside issue that I ran into.

If you find more errors, can you please create a new issue?

[ericholscher]

Another option for now is just to pin to docutils <0.17. We did this on the RTD theme, until we have a chance to fully test and implement the changes.

[return42]

Another option for now is just to pin to docutils <0.17

Another option I would vote for is to maintain a own fork and decouple docutils development from David Goodger: Each minor release brings new incompatibilities, each release of docutils cause problems in hundred of projects.

[tk0miya]

I feel such a reaction is very hysteric and intolerant. If you'd like to get the docutils project better, please join the project and help them. (or please vote to decouple Sphinx development from me also)

[tk0miya]

I merged #9059 to 3.5.x branch. So closing this now. Please file another issue if you find other troubles. Thanks,

[return42]

I feel such a reaction is very hysteric and intolerant.

Apologies my inappropriate comment .. but I'm annoyed that each minor release from docutils caused a cascade of problems.

If you'd like to get the docutils project better, please join the project

I had a look years ago ...the access to become team member is unusually difficult .. and personally I have a problem with sourceforge. / I know of some developers who would like to contribute to the project but don't, because the project management is a little bit idiosyncratic.

[ericholscher]

I think the main goal here is more testing and communication, which is something everyone in the community can help and contribute to. Wearing my Read the Docs hat, we'd love to be able to work with sphinx & docutils to increase testing beta & development across the ecosystem, so hopefully we can catch some of these issues prior to them getting released.

We can talk more about it over at #9056 perhaps, instead of this old issue, but I think there's some productive work to be done here.

[andredias]

Disclaimer: I'm the author of rst2html5

I'd like to suggest using it instead of docutils' rst2html5. It has been stable for years and produces better results than docutils. Please, see its README.