Fix connection pool

[stefan-jansen]

Addresses #209.

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: +14.69 🎉

Comparison is base (206888c) 73.43% compared to head (03bd61e) 88.12%.

Additional details and impacted files

@@             Coverage Diff             @@
##             main     #210       +/-   ##
===========================================
+ Coverage   73.43%   88.12%   +14.69%     
===========================================
  Files         171      171               
  Lines       16057    16058        +1     
===========================================
+ Hits        11791    14151     +2360     
+ Misses       4266     1907     -2359     

Impacted Files	Coverage Δ
src/zipline/utils/sqlite_utils.py	100.00% <100.00%> (+4.16%)	⬆️

... and 75 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[stefan-jansen]

@MBounouar - adding the fix proposed by @RichardDale; just let me know if you have a better alternative in mind. Looks like something our tests missed.

[RiseT]

@MBounouar (sorry, just fixing alias typo 😉 )

[MBounouar]

@MBounouar - adding the fix proposed by @RichardDale; just let me know if you have a better alternative in mind. Looks like something our tests missed.
Unfortunately, I don't have a better alternative. I vaguely remember about this issue when playing with SQLAlchemy 2.0 several months ago, but I completely forgot about it and never encountered any issue with Linux or WSL.

The whole sql reading and writing in zipline is a bit awkward, even more so with SQLAlchemy 2.0. Refactoring the code or at least investigating potential improvements would make sense. I have the impression that the whole ingestion process is suspiciously slow. However, I don't think I will be able to pick that up any time soon

[RichardDale]

Thanks Stefan and Mehdi - looking forward to this being pushed to conda-forge so our users can upgrade.