mayeut
commented
1 month ago Sorry it took me so long to look a bit more into this. I confirmed the hunch I got when reading the issue, that is, it would grow the size of processed data too much (around 30GB/day this week). This would undeniably get the project over the free-tier limit for big query. Even if dropping the pip version (which will happen eventually), this would be an additional 20 GB/day.
I'm going to leave this open anyway in order to get a reminder once in a while to check what can be done about this.
Currently there's a "Python implementations" plot under "package statistics", but something similar is missing for distinguishing e.g. CPython/PyPy/GraalPy on the consumer side.
The number between package & consumer statistics are wildly different (compare e.g. % of python 3.11 package builds vs 3.11 consumers), so that information cannot be inferred.
CC @mayeut