Logrun

Available on PyPI: https://pypi.org/project/logrun/0.2.0
Available on conda-forge: https://anaconda.org/conda-forge/logrun

Additions:
- By default, reads `LOGRUN_ROOT` from a JSON configuration file in the current directory. The environment variable still works, and if set overrides the path specified in the config file.

Changes:
- `experiments_by_{input,output}_file` now use text files with an experiment ID in each line, instead of symlinking to the experiment targz

Fixes:
- Make eval_checksum on directories be deterministic and more robust

Available on PyPI: https://pypi.org/project/logrun/0.1.6
Available on conda-forge: https://anaconda.org/conda-forge/logrun

Fixes:
- When importing `logrun.internals` from a REPL, do not activate logging code
- Avoid crashes on source code logging, in particular when there are empty `*.py` files
- Avoid duplicate input or output files

8fdcc50
Available on PyPI: https://pypi.org/project/logrun/0.1.3/
Available on conda-forge: https://anaconda.org/conda-forge/logrun

Fixes:

- `outpath` ensures that the *parent* directory of the given path exists
- Source logging now skips a module if it can't get its source (this only happens with modules that don't have a corresponding `*.py` file)

Available on PyPI: https://pypi.org/project/logrun/0.1.2/

Changes:
- `LOGRUN_ROOT`'s existence is now only checked when saving the experiment

Fixes:
- `inpath` and `outpath` are now working again
- Typos in the README