Noticed that

approx_count_distinct

isn't in our docs, PR: https://github.com/Eventual-Inc/Daft/pull/2805 cc @jay @Raunak Bhagat Also, this and

approx_percentiles

aren't exposed as methods on Grouped Dataframe and Dataframe, was there a reason for this?

Jay and I discussed about writing docs for

approx_count_distinct

but I wasn't able to get around to it. I also wasn't aware that we had to add it to Grouped DataFrame as well, that's my bad.

> Jay and I discussed about writing docs for

approx_count_distinct

Like separate from the expression docs?

I wrote the expression docs, but not sure how they get propagated to the docs available at getdaft.io... Are those expressions docs automatically integrated into the webpage documentation available at getdaft.io?

you have to put them manually in

docs/source/api_docs/expressions.rst

. Only then, do they get auto generated onto the website.

I’m trying to build a mechanism that will at least give us a preview of the website on your PRs

Ahh okay, makes sense. I'll add the group-by stuff in a PR soon too. Sorry about that y'all.

I actually think it might be ok to not have those on the dataframe-level

Sure but we should make the aggregation expressions more discoverable though: https://www.getdaft.io/projects/docs/en/stable/api_docs/expressions.html#aggregation For example, we should link those agg exprs in the

.agg

expression docs: https://www.getdaft.io/projects/docs/en/stable/api_docs/doc_gen/dataframe_methods/daft.DataFrame.agg.html, and probably in the

aggregations and grouping

user guide as well. https://www.getdaft.io/projects/docs/en/stable/user_guide/daft_in_depth/aggregations.html Currently this one only talks about the agg methods, not the expressions

^ yes for sure

Ok modified the PR to reflect making the agg exprs more discoverable, and improve the grouping/agg user guide. @Raunak Bhagat could you give it another pass? also for reference, here's how you can build the docs locally: https://github.com/Eventual-Inc/Daft/blob/main/docs/CONTRIBUTING.md

Looks good to me!

Interestingly doctests are failing (on my local machine) because of precision issues I think 😛