how to navigate clojure code in emacs? cider ? or there maybe some other method

@haiyuan.vinurs haven't completely got a question.

@haiyuan.vinurs emacs and install cider that should be enough to get you going https://www.braveclojure.com/basic-emacs/

with cider that i must start a repl, is there a tool like global gtags that can parse the code statically

my normal workflow with cider is: open clojure project, start cider repl with :cider-jack-in, start to eval my code with :cider-eval-defun-at-point (mapped to ,ef hehe), finish the cider repl with C-c C-q

I'm confused, how does (reduce conj () [1 2 3 4]) reverse the sequence? (yes, walking through 4clojure problems!)

it's the reduce part that confused me, not the conj

(conj (list 1) 2) in your repl

and then (conj (conj (list 1) 2) 3) in your repl

the tricky thing is that (conj '(1) 2) appends while (conj [1] 2) prepends

conj has performance guarantees not location guarantees. Vectors are easy to append to. lists are easy to prepend

Ahh, I see, thank you Dan and Will. Both very insightful.

I will ponder (hammock?) the code further 🙂

@dharrigan you can also replace the reduce with reductions to get an idea of what is happening at each step of the reduce.

that's handy to know! thank you kindly!

I was confused about that one until I realized that the () was just the initial value for the conj... Realizing that (reduce conj () [1 2 3 4]) is the same as (reduce conj [() 1 2 3 4]) made it clearer for me

often if a reduce is confusing you can s/reduce/reductions/ to get some clarity - (reductions conj () [1 2 3 4])

shows you all the intermediate accumulator values

That's a great tip!

I spent a nice debugging session with cognitect's AWS library wondering why its profile-credentials-provider cannot find my profile. Heh, a rookie mistake:

:aws-profile local-dynamodb

:aws-profile "local-dynamodb"

(get (config/parse f) profile-name

Yihaa!

And I really love Cursive. I created a "reset" command and a simple hot key for it - I'm able to "stop states, refresh namespaces, start states" with one keystroke, i.e. e.g. switch the data source from my local dynamodb docker container to the real AWS DynamoDB. Yihaa!

... and if you see Stuart Halloway there in Cognitect corridors send him my regards regarding his "Repl Driven Development" - it totally changed how I use REPL.

If I have a vector like

[{:pgid {:S "1"}, :pgname {:S "Books-local-aws"}} {:pgid {:S "2"}, :pgname {:S "Movies-local-aws"}}]

{"1" "Books-local-aws", "2" "Movies-local-aws"}

why is reduce not idiomatic?

This works:

(reduce
        (fn
          [mymap item]
          (conj mymap {(get-in item [:pgid :S]) (get-in item [:pgname :S])}))
        {}
        items)

it's the best way to make an associative object out of a seq

... but is there some standard library function that could be used instead?

Ok.

no

you can use assoc instead of conj

So, my reduce above is just fine?

(conj foo {bar baz}) is the same as (assoc foo bar baz) except it creates a hash-map nobody needs

Ok.

yes, the reduce is fine, but I'd use assoc

Ok. Thanks!

So:

(reduce
        (fn
          [mymap item]
          (assoc mymap (get-in item [:pgid :S]) (get-in item [:pgname :S])))
        {}
        items)

also, this is open to judgment, but (-> item :pgid :S) might be more readable than (get-in item [:pgid :S]) for this specific case

a line break might help readability as well

(assoc foo
       bar baz)

Ok. Thanks!

So:

(reduce
        (fn
          [mymap item]
          (assoc mymap 
            (-> item :pgid :S) (-> item :pgname :S)))
        {}
        items)

Good that I asked. 🙂

So, occasionally if I'm wondering if my small code snippet is idiomatic Clojure or not, you guys don't mind me asking here? I learn a lot with you gurus. 🙂

This is exactly the right place for that sort of question

All right! I'll be a lot more active here in the future! And I'll be learning Clojure 2x than I previously did. Yihaa!

i would do this

there's also #code-reviews but usually that's usually used for bigger things

@nedcg my only problem with that is that putting an arg to a parent on the same line as a child that had a line-break in the middle feels strange

to me a line break mid form introduces a new level of nesting, so keeping an upper level on the same line has a weird friction to it - but of course that's subjective

@noisesmith the line-break is for visually associate the operations over the same parameter(%2), it feels weird to me decompact with line-breaks such a simple function

right, but {} and items are for reduce, not for #(assoc ...)

so I would remove the line break, or add another for each arg to reduce

but as I said at the start, it's subjective, we don't need to have consensus, but it's informative to hear each others reasons I think

absolutelly, thanks for the feedback

another way to implement this pattern is with (into {} (map (juxt key-fn val-fn) coll)), where key-fn and val-fn are the corresponding “extractions” on each value in coll

(->> items
     (map (juxt (comp :S :pgid) (comp :S :pgname)))
     (into {}))

or even (into {} (map (juxt kfn vfn)) items)

nice

out of curiosity, does that provide a benefit if there’s only one map?

yes, it avoids creating a lazy-seq nobody needs

clojure doesn't do this collapsing of intermediary values automatically (many other languages do), but we can do it explicitly

Interesting. I need to try that tomorrow (it's midnight here in Finland).

How do I learn to see things like "it avoids creating a lazy-seq nobody needs"?

specifically, reading about transducers (https://clojure.org/reference/transducers)

You just looked at the code and immediately saw it?

i personally wouldn’t worry too much about it if you’re learning clojure

transducers can help performance for multi-step list processing

Ok. I read it tomorrow. I think I'm not absolute beginner, I have used Clojure a bit but now want to be more fluent and learn more idiomatic Clojure coding...

over time you’ll learn what the code actually does: (map f coll) creates a lazy sequence, whereas (map f) does not. so if you’re going to do multiple steps (e.g. map then filter then into), you can avoid creating those intermediate sequence structures and use the more advanced transducer form in order to make it faster

I only really started to learn this stuff after I started digging into how I can make my horribly slow code, less so 😛 but 90% of the time it doesn’t matter