Vitalie Spinu

@lionel- I believe, if there is no strong reasons for a restriction, then it better not be there. There are plenty of use cases for modify_in to be able to...

`map` is a bad parallel. In all programming languages it does preserve the input size and it does so by very definition. Programming should be practical; not enforcing unnecessary invariants...

I see. Valid point indeed. This transitivity didn't occur to me. But then, a separate function `update_in` could be doing what the similarly named function in other languages does. Can...

@lionel- ping. In the past two years I hardly ever used modify_in and assign_in. Inability to create nested elements is a killer. Any possibility to prioritize this issue?

Any news on this? You can do all other directions, `child`, `follow` and `precede`, but not `parent`. So you can currently get both A and C in (A B C)...

Right. That's exactly what I was looking for. Don't understand how I missed it in the docs. Thanks!!!

> and you'll be returned a Clojure zipper that you can do modifications on. Thanks. I will try it out and will post an example here for the reference.

I finally got down to figuring this out: ``` clj (defn hickory-remove [selector-fn hickory-tree] (loop [zip (hickory.zip/hickory-zip hickory-tree) next (hs/select-next-loc selector-fn zip)] (if next (let [new-zip (zip/remove next)] (recur new-zip...

> I would use is_list(x) && !is.data.frame(x) This would be an unnecessary restriction. In case of data.frames and other containers which don't behave as we expect on `[]

Ok, applied all you suggestions except for abort and data.frame check as per my comment above. The `modify.R` still uses stop throughout, please modify all of them yourself as you...