starling-lab / BoostSRL

BoostSRL: "Boosting for Statistical Relational Learning." A gradient-boosting based approach for learning different types of SRL models.
https://starling.utdallas.edu
GNU General Public License v3.0
32 stars 21 forks source link

BoostSRL README overhaul, contribution guidelines, reorganization, repository optimization. #6

Closed hayesall closed 6 years ago

hayesall commented 6 years ago

README.md changes:

New file: .github/CONTRIBUTING.md

.github/ directory is for markdown and information GitHub uses. CONTRIBUTING.md contains some contribution guidelines for BoostSRL, notably How to ask a question, How to file a bug report, and How to suggest a new feature.

New directory: docs/

A large portion of the documentation for WILL was buried in the code (/src/edu/wisc/cs/will/ILP/willmanual/), this documentation may be appropriate to have around, so documentation was moved to a directory specifically for this type of information. docs/ will be used for this purpose.

Removing the Trains dataset:

In a continued effort to separate data and the code, this update removes the trains dataset from /src/edu/wisc/cs/will/ILP/SampleTestbeds/trains/ (refer to Issue #5). The trains dataset is now hosted on the BoostSRL-Misc repository.

hayesall commented 6 years ago

Clearing out large files

This pull request will rewrite a large portion of the repository history, clearing out datasets that were written to the history at various points in time.

This was done with the bfg-repo-cleaner on the HumanInTheLoop branches.

Overview of steps (after downloading the jar):

Clone a mirror of the repository (i.e. just the .git contents): git clone --mirror https://github.com/HumanInTheLoop/BoostSRL.git

Use the bfg to delete the contents of SampleTestbeds: java -jar bfg.jar --delete-folders SampleTestbeds --no-blob-protection BoostSRL.git/

Reflog and cleanup: git reflog expire --expire=now --all && git gc --prune=now --aggressive

hayesall commented 6 years ago

BFG the master branch

Merging development branches back into the boostsrl_readme_1.1 undid the results of "Clearing out large files" comment. These steps are easy to repeat though, perform them on the BoostSRL master branch after all of these changes are merged in for version 1.1.

hayesall commented 6 years ago

Closing this pull request without merging

Performing some merges after trying to optimize the branch caused the unexpected changes documented above. Because of this I am going to close this request without merging, the useful commits have been cherry picked and are now available via pull request #8