Nikita Titov
Nikita Titov
@jameslamb Great findings! > it seems like `LGBM_NetworkFree()` is only ever getting called on the machine that is rank 0 Hmm, so that might be a reason of failing tests...
@jameslamb Seems that #4132 fixed this issue, didn't it?
Just seen this error at `master`: Logs: ``` 2021-07-31T19:38:41.5284774Z Found online and idle self-hosted runner(s) in the current repository that matches the required labels: 'self-hosted , linux' 2021-07-31T19:38:41.5284859Z Waiting for...
> but macOS Dask tests are still the main source of flakiness I'm seeing in LightGBM. Also this one #6703 🙈
@jameslamb Could you please enable RTD builds for this branch for testing?
One more issue is that `stable` leads to `4.0.0` (and not `4.5.0`) version of R-package docs. https://lightgbm.readthedocs.io/en/stable/R/reference/ 
I believe it's ready for reviews.
@jameslamb Hey, great idea about build time changes! I'll try to implement it in `conf.py` since that file already contains a lot of installation code and it won't be neccesary...
> One more issue is that `stable` leads to `4.0.0` (and not `4.5.0`) version of R-package docs. Looks like RTD really treats `stable` as `4.0.0`. Build hashes are the same...