S.Lott
Sorry for the long delay. (I've got a million excuses, all of them really shabby.) I appreciate your enthusiasm and your patience. Can you execute the CLA agreement?
With PR #131, we now have tooling (and test cases) for the last CEL specification. (Thanks @hudlow and @stefanvanburen for doing so much work on this.) This means we "simply"...
It's important to benchmark against the Go version of CEL to be sure what it does with these kinds of things. A PR is welcome.
One performance drag is the CEL spec mandates a number of features of Go. This leads wrapping some Python objects to create Go-language semantics. (The `%` operator is a specific...
To answer the question as asked: Yes, poor performance is expected. I doubt, however, that's what this issue is really about. I'm adding this detailed example to the benchmarking suite...
Closing this to permit focus on compliance.
After looking at the `Makefile`, my suggestion on deleting was a bad idea. I don't think it should be deleted. I think it needs to be rewritten to use poetry...
I'm still trying to understand the ask and the implications. Specifically, I don't know if *What is needed is for `has` to return false while still populating the default value.*...
> it seems obvious that you have to special case protobuf messages for has. Okay, I'm with you so far. I understand CEL-Python is different. I'm totally convinced it's different...
Update ci.yaml and tox.ini (among others) at scheduled end-of-life 2025-October