Release 0.18.0
Closed this issue · 3 comments
@eitsupi in the end I'm not sure it's worth implementing the checks for empty dots now that the rewrite has started. I don't want to invest time on this if it turns out that the whole codebase will change in a few weeks.
Therefore, unless there are things you want to address in particular, I think we can release 0.18.0.
Thank you. However, the rewrite will take a few months (I was hoping to complete the work while Python Polars is in v1)
Well, in any case, we can release 0.18.0 and check the dots afterwards, and I will work on the release.
Ok, I am hesitating on where I should put my focus now. There are new functions to implement from upstream but if I do it on main then they will have to be rewritten in the next weeks/months, so it seems like a waste of work. Do you think I should now focus more effort on the next branch?
This is a difficult decision because we cannot predict at this time when the switch to the next branch will be possible.
In my opinion, if the feature is very important and easy to implement (i.e., if most of it is handled upstream in Rust and only a small wrapper is needed here), then the changes in the current main branch are worthwhile.
I feel that more than half of the work to add the functionality added upstream to the R package as well would involve adding tests and documentation, which should be mostly copiable to the next branch.
On the other hand, if a feature is described by a large amount of Python code in Python Polars, it might be better to hold off on adding it to the main branch.
So I will continue to update the Rust Polars, and I think that the rest of the work should be minimal.