@slott56 Yes, It's mainly data scientists bringing up the issues...
@sullybiker spent a bunch of years with them. We strongly encouraged the use of conda (either full Anaconda or miniconda). And spent a lot of time making sure they could build virtual environments.
It was banking and regulatory oversight meant reproducibility. And constant retesting and retraining with new versions. And. No disclosures to mere mortal devs. Trade secrets. And. Customer info. We could help but we couldn’t touch. Or see.
@slott56 I am a huge, huge fan of conda for this reason.
@slott56 The field is outpacing the packaged software, at an unprecedented rate. This isn't a new problem, but it keeps us in work...
@sullybiker as Python’s popularity continues to grow, there will be adjustments, I’m sure. The analytics/data science toolkit involves a very tall tech stack and innovations at a rapid tempo. To me, it sounds like these kinds of problems may be trending.