Compatibility for xgboost 0.90 #90
Conversation
xgboost_ray/compat/tracker.py
Outdated
| # flake8: noqa | ||
| # Copied from https://github.com/dmlc/xgboost/blob/master/python-package/xgboost/tracker.py |
There was a problem hiding this comment.
Do you need to add some licensing notes here?
There was a problem hiding this comment.
Right, this is Apache 2.0 so probably yeah. I'll quickly check how to attribute this correctly
There was a problem hiding this comment.
Ok since our own license is also Apache 2.0 we shouldn't run into any problems.
I still added a license header to this file and linked to the correct (hash-identified) file source and license file on the xgboost repo.
There was a problem hiding this comment.
(pushed but github seems to need some time to update this PR)
xgboost_ray/compat/__init__.py
Outdated
| logging.warning( | ||
| f"You are using `xgboost_ray` with a legacy XGBoost version " | ||
| f"(version {xgb.__version__}). While we try to support " | ||
| f"older XGBoost versions, please note that this library is only " | ||
| f"fully tested and supported for XGBoost >= 1.4. Please consider " | ||
| f"upgrading your XGBoost version (`pip install -U xgboost`).") |
There was a problem hiding this comment.
i wonder if this will get printed on every single Ray worker?
There was a problem hiding this comment.
I don't remember seeing that in my runs but I can check again later to make sure.
There was a problem hiding this comment.
Turns out you were right - I moved the warning to the predict() and train() methods, so they will only be printed once.
cc @mmui