Version 1.0.0rc1

[aryan26roy]

Summary

This is the first release after the switch to Lark.

What's Changed

• ci: add trusted publisher and attestations by @henryiii in ci: add trusted publisher and attestations #44
• Fix trusted publisher delivery by @jonas-eschle in Fix trusted publisher delivery #47
• Update version and .gitignore by @aryan26roy in Update version and .gitignore #48

New Contributors

• @aryan26roy made their first contribution in Update version and .gitignore #48

Full Changelog: v0.1.1...v1.0.0rc1

---

This discussion was created from the release Version 1.0.0rc1.

[jonas-eschle]

I was just wandering @aryan26roy what's the plan for the 1.0 release? How long do you wanna keep the RC only?
Also, I would suggest to add it to conda-forge once 1.0 is out, I can do that and add you as a maintainer as well, okay?

[jpivarski]

Maybe @jpivarski can help us out here and add any other action items he would like to see go in before v1.0.

I can't think of anything that's missing.

[jonas-eschle]

I think the one thing that would break at least my code is the variable attribute. Since it's parsing the expressions, it seems reasonable to know about them (similar to how numexpr, sympy or any other expression knows about its input variables).

No strong opinion, but this is the one thing I've been regularly using from formulate (i.e. to let uproot know which variables to load or which variables I will need from a df etc) and seems fairly simpel, as we're parsing the expression anyways (the operations are probably harder/longer)

Other than that it's fine I think

[aryan26roy]

Hi @jonas-eschle , I think the functionality to know about variables can be easily added to the current code. It would have to be a recursive call over the whole AST, perhaps something we can do internally before we give the AST object to the end user. @jpivarski I think we can recreate the exact API with this new implementation since they were originally creating a set of all variables and exposing it as a data member, unless I missed something.

[jonas-eschle]

Hm, yes, that could make sense. I think it's not a stopping item IMHO, but probably the one most reasonable feature for something like a formula (but the argument can be made that any of the libraries, numexpr or ROOT should be able to parse a corresponding string anyways, so that's somewhat of a doublication, as one of the libraries will always be used (otherwise formulate is meaningless)

[jonas-eschle]

@aryan26roy I think enough time has passed and I would suggest to make the 1.0 release?

As mentioned, the only thing could be to add the variable attribute, by just using numexpr (as a runtime dependency). It's quick and dirty, but probably the only thing that actually is needed from the old API, even thinking about it since a while now, that is the only thing I would still miss.