Doc build failing

[valeriupredoi]

fail xgboost is at version 1.6.2 on conda-forge, but at 1.7.0 on PyPi - a bit not ideal, but the build gets very confused about that. I recommend we pin <1.7.0 what do you reckon @bouweandela @zklaus

[bouweandela]

Just wait a bit for the conda package to appear? 1.7 isn't a day old yet

[valeriupredoi]

yeah that was my plan as well, takes about a day or so between PyPi and c-f - but in case stuff's still not OK I was thinking about pinning temporarily until they fix their business 👍

[zklaus]

Hm. I wrote about the failing nightlies on Discord. As far as I can see this has absolutely nothing to do with xgboost, but everything with docs.scipy.org being down and hence failing to deliver intersphinx data. This is fixed now and the builds seem fine.

What led you to xgboost?

[zklaus]

Ah, I see it now. I was only looking at our nightlies, where the doc build also failed. On readthedocs they gloss over the intersphinx problem and then run into the xgboost one. Let's see where we land with the nightlies. Btw, someone is working on that xgboost release, see conda-forge/xgboost-feedstock#104

[valeriupredoi]

all working now, cheers, gents 🍺

[valeriupredoi]

seems that enetering November was not auspicious - Docker broke last night, docs broke, my travel expenses claim submission portal was FUBAR, even my mum has a broken water pipe at home 🤣

[valeriupredoi]

nevermind, spoke a bit too soon, let's keep this open until it's actually fixed (by xgboost folk)

[valeriupredoi]

xgboost is at 1.7 across the board, and our doc builds work again 👍

[bouweandela]

Now they've released 1.7.1 on PyPI while conda is still on 1.7.0 and the problem is back

[valeriupredoi]

agh! Was just aboot to come back here too 🤦‍♂️

[valeriupredoi]

conda-forge/xgboost-feedstock#105 - I may fix it for them in a couple days