- COME UP WITH A GOOD NAME! (
pandasql
is already taken in PyPI) - Fix up
setup.py
- Stream BCP stdout to logs in real time asynchronously
- Upload to PyPI
- Upload to conda-forge
- Set up CI/CD
- Benchmarks vs. regular pandas methods
- Find way to automate creation of temp SQL Server db for testing
- Use pandas internal code to build new SQL tables, instead of just all NVARCHAR
- Better docs
- Remove SqlCmd and all references to it
- Remove the deprecated
read_sql
and all references to it - Test inserting dataframe with (all) boolean values
- Implement if dataframe is missing some database columns and/or the columns are out of order
- Add section in docs detailing testing and Docker
-
1.0.0
Release
- Explore alternatives to
INFORMATION_SCHEMA
SQL views - Write more tests for 100% code coverage
- Build in auto-retry with native pandas methods if BCP method fails
- Add instructions for contributing (setup conda env, how to write tests, pre-commit, etc.)
- Add to docs the exact required SQL permissions
- Add the benchmarks to CI/CD
- Find way to use Linux containers on Windows OS in CI/CD
... and of course, anything marked # TODO
in the code.