Open harold opened 2 months ago
I did the absolute minimum possible to make the docstrings clearer - so this is partially addressed at this point - feel free to repoen if you want to have someone tackle this is more depth.
https://github.com/techascent/tech.ml.dataset/commit/bd8dacb2b75241791b6695d763d8b0ebc91b3106
Agreed that's an improvement, I still think examples would go a long way. I'll put the good first issue
tag on here and some enthusiast will come and do it. (:
Autocomplete in my editor helpfully offers these suggestions:
The docstring for
ds/concat
says this:Which I find sort of confusing given that the three functions
concat
,concat-inplace
, andconcat-copying
seem to imply that in place and copying are different.Improve arglists
Also, the arglist of all three functions is:
[dataset & args]
, which sort of obscures the fact that you're supposed to pass in many arguments each a dataset. Would[& datasets]
be better? ...clojure.core/concat
has[x y & zs]
which I also sort of don't like because it obscures the fact that the arguments are supposed to be sequences.Improve docstrings
My proposal would be for re-written docstrings on all three of these functions with 1) mention of the input and output return types 2) a short working example of usage, so users don't have to wonder if they need to call
apply
3) an indication in all three to which one is the one you probably want (is itds/concat
?) with a short explanation of why you occasionally might want the other two, with, if possible, a short example of their ideal use...