If want to use specter to replace the nth value in deeply nested array, is there a better way to navigate to that value than using (srange x (inc x))

Ideally I’d like to be able to transform the value there, or put in default if it doesn’t exist

@conaw: there's a custom nth-elt navigator in the wiki at https://github.com/nathanmarz/specter/wiki/List-of-Macros#defnav

I don't know of a better way to do it, personally

awesome

under the hood it does basically what you said

if you want default element behavior, you could modify it to do that

@conaw by array do you mean vector? if so, I believe (keypath 3) works

yeah, I’m actually working with a matrix, so words slipped

thanks a bunch

np

So I can reasonably do a (if-path (keypath 4) LAST) right?

or I would do (if-path (keypath 4) (keypath 4) LAST)

or rather END

nvm (if-path (must 3) (keypath 3) END)

ok, here’s a question, it seems that I can substitute a value into a vector using setval , but if I want to use sp/END i have to have it wrapped in a seqable, so couldn’t insert it in in a flat way, is there a way I might get around that

(setval [(sp/if-path (sp/must 3) (sp/keypath 3) sp/END)] :newd  [:a :b :c :dd])

=>  [:a :b :c :newd]

(setval [(sp/if-path (sp/must 3) (sp/keypath 3) sp/END)] [:newd]  [:a :b :c])
=> [:a :b :c :newd]

basically, in assoc, if I assoc to the size of a vector, I conj onto the end, I’m looking for a similar behavior here

@conaw: I think you can do [END (keypath 0)]

@codonnell: sorry for disappearing last night, I ran the benchmark with (sequence (comp ...) %) substituted for (->> % ...) and it actually came out faster, 97ms vs 110 ms

that's really surprising to me

Does (comp ...) take some time to initially build up the transducer?

@aengelberg: yes, most noticeable when operating on relatively little data

@aengelberg: the benchmarks I pasted before show a large performance difference between "precompiling" the transducer and doing it inline