Improving CM core

[gfursin]

Aggregating tasks to improve CM core based on user feedback (this ticket is being gradually updated):

Functionality

• [x] Add support to check min CM version for automations and scripts
• [x] Move string checks to belong to [True, yes, on] to cmind.utils to avoid duplicates
• [x] Add cm checkout repo to checkout pulled repos
• [x] Add support to debug CM core and scripts (cmind.utils.debug_here) (tested with Visual Studio)
• [x] Develop alias to pull mlcommons@cm4mlops instead of mlcommons@ck (when -b is not used)
• [ ] Add portable md5sum to cmind.utils from here - it is often used in various CM scripts
• [ ] Add portable tar and zip extractor to cmind.utils based on ABTF prototyping - used in many projects
• [x] Add logging to CM core
  • [ ] Agree on CM logging names/streams
  • [ ] Convert CM prints to logging
  • [ ] Convert CM script prints to logging
• [x] Print internal status (version, paths to repos, python used, etc): cm test core
• [ ] URGENT: setup.py will be deprecated soon and we need to provide a new way to install CM - help is appreciated!

Tests

• [ ] Unify GitHub test names
• [ ] Add more unit tests to cover most/all CM internals across different platforms
• [ ] Agree how to keep control of code quality of CM scripts and improve quality of existing CM scripts (from prototype to production)

Docs

• [ ] Agree on how to show optional flags. Should be similar to Python parser: cm run script ... [--help] [--path PATH] ...
• [x] Generate CM API docs per version with sphynx and upload to website
• [ ] Improve basic CM framework docs/tutorials based on cmake/bazel/workflow concept
  • [ ] CLI
  • [ ] API
  • [ ] add new recipe
  • [ ] share new recipe
  • [ ] work with CM scripts
    • [ ] add new script
    • [ ] reuse new script
    • [ ] CM cache
    • [ ] workflow (env/state)
      • [ ] deps with conditions
      • [ ] customize
• [ ] Provide guide for user contributions and to report issues

Tutorials

• [ ] Basic CM concepts
• [ ] CM automations (use, extend, add)
• [ ] CM scripts

Brainstorm

• [ ] Add possibility to have strict validation of inputs and types for production environments
• [ ] Move CM index to sqlite to speed up search by tags and enable parallel execution - need to brainstorm
• [ ] Check how to parallelize cmind.access runs (if needed)
• [ ] Simplify get git, download file, extract file with custom CM cache entry ...
• [ ] Simplify docker generation via separate docker automation
• [ ] Test CM scripts with different versions of dependencies
  • [ ] Enable crowd-testing and reporting similar to cTuning