Tskit: RTD build failing

[jeromekelleher]

https://readthedocs.org/projects/tskit/builds/10785421/

I think this is probably a pinning issue for sphinx also?

[petrelharp]

You know, it'd be pretty easy to build docs ourselves and serve them with gh-pages.

[benjeffery]

Yes, that's the exact same error. I'm looking throught the readthedocs docs, but can't see where you pin the version.

[benjeffery]

Ahhh, found it, will PR

[jeromekelleher]

You know, it'd be pretty easy to build docs ourselves and serve them with gh-pages.

RTD is annoying when it doesn't work, but it's served us well and it mostly does just work. It's bloody conda is the problem over in https://github.com/popsim-consortium/stdpopsim/issues/468

[molpopgen]

gh-pages also blows up your repo size quite a bit.

[molpopgen]

You know, it'd be pretty easy to build docs ourselves and serve them with gh-pages.

RTD is annoying when it doesn't work, but it's served us well and it mostly does just work. It's bloody conda is the problem over in https://github.com/popsim-consortium/stdpopsim/issues/468

This is an RTD issue, which has been partially fixed recently. You can request more memory for the project, but the best way to avoid it is to become a paying supporter of RTD.

[petrelharp]

gh-pages also blows up your repo size quite a bit.

Really? docs/_build/html is only like 12M on msprime and tskit, and seems to be self-contained.

[molpopgen]

gh-pages also blows up your repo size quite a bit.

Really? docs/_build/html is only like 12M on msprime and tskit, and seems to be self-contained.

Every time you rebuild and commit, the diffs are massive because of all the HTML gibberish that changes inside.

[benjeffery]

That's easy to get round - just don't keep the history on that branch.