crflynn / pbspark

protobuf pyspark conversion
MIT License
21 stars 5 forks source link

Support for spark==3.1.x? #47

Closed jelos98 closed 1 year ago

jelos98 commented 1 year ago

Hello!

I'm interested in using pbspark in a project I'm working on. Unfortunately, I'm running in an spark environment that's not in my control, and as a result I need to be able to run with pyspark==3.1.2, which is problematic given that the requiremetns for pbspark list pyspark>=3.2.0

I've tested locally (though admittedly non-comprehensively) by modifying the requirements to allow pyspark==3.1.2, and it appears to work fine, at least for the cases I need.

Was the requirement set at 3.2.x because you have specific reason to believe it doesn't work in all cases on 3.1.x? Or was it simply a matter of 3.2.x being the version that it was developed against, and thus it's not tested against earlier versions?

If it's the latter, would you consider relaxing the requirements (perhaps with a README caveat that 3.1.x isn't thoroughly tested?) If it's the former, do you recall what doesn't work on earlier versions, so I don't wind up doing something that's going to come back to haunt me?

Either way - I wanted to thank you for your work on this package. I'm working in a world where I'm consuming literally dozens of distinct, large protobuf messages in Spark, and this package as you might imagine, becomes an enormous time saver.

Kudos, and thanks! -James

crflynn commented 1 year ago

I think that's just what the latest version was when I created this.

I just tested locally with python 3.8 and pyspark 3.1.1 and the tests all passed. I can relax it and make a new release for you a bit later today.

jelos98 commented 1 year ago

Awesome!

Thanks so much!

crflynn commented 1 year ago

Try 0.9.0. Feel free to re-open if you have any issues