Edits containing root map replacement

[arichiardi]

Hi there, I am trying to get all the keys (recursively) that have changed by parsing the get-edits vector and it all works - plus it reads nice.

However, there is an exception: get-edits returns

[[] :r {:foo :bar}]

when the first in the diff is empty and the second populated (in my case {:foo :bar}.

Now, I can special case on that [] but I wonder if I am missing more of the way editscript returns its edits.

The I have two side question 😄

Did I understand things correctly above?

Should we mention this special case in the README, right after:

The first element of an edit is the path, similar to the path vector in
the function call update-in. It might be confusing/unexpected to others.

[huahaiy]

What's the data that produces this diff? Please provide more context. Thanks.

[arichiardi]

Thank you @huahaiy I definitely can provide more context!

(editscript/diff {} {:foo :bar})
;;=> [[[] :r {:foo :bar}]]

But this also happens for more complex things - especially when there is a big number of keys - it seems like editscript is "giving" up at some point.

Would that be possible @huahaiy?

[arichiardi]

Maybe I need to tweak the vec-timeout param somehow, i basically want to avoid replace the whole thing especially for top-level keys.

[huahaiy]

It's a matter of counting the cost. For example, we can tweak the cost of :r to penalize replacement, or tweak the cost of adding a container object, etc. I don't have time to work on these right now, but would be happy to take a PR.

[arichiardi]

I see - yeah no time on my side as well for now