Clojurians Log v2

Convert your threading macro into a let block like this: (defn fetch-fastmail-emails! [ctx] (let [ 1 (fastmail-prep-auth-request ctx) 2 (fastmail-call-api 1) 3 (fastmail-parse-auth-response 2) 4 (fastmail-prep-email-request 3) 5 (fastmail-call-api! 4) 6 (fastmail-parse-emails 5)] ;; Debug return a datastructure with what you want to see, I do this in the repl as you are building it up or debugging [1 2 3 4 5 6] ;; Otherwise return what would have been the final form in the threading macro ;;6 ))

At least that is what I do for visibility. You could add a "7" to the let block with [1 2 3 4 5 6] as the bound value. Then have another function to error test it. either with Malli/spec or manually via another function. Then add the :error key to the ctx that you return based on whatever that function does

if you wanted to break the flow along the chain from 1-6 and stop processing due to an error, I wouldn't mix side effects and checks in a threading macro or function with let blocks. You'd need separate functions, and the orchestration uses something like multi-methods or core.match to dispatch to the right function based on the state of the data

Don’t know if I’d recommend it, but you could do something like that with reduce as well:

(def todo [fastmail-prep-auth-request
      fastmail-call-api!
      fastmail-parse-auth-response
      fastmail-prep-email-request
      fastmail-call-api!
      fastmail-parse-emails])

(defn execute-step [ctx step]
   (let [r (step ctx)]
     (cond-> r (not (ok? r)) reduced)))

(reduce execute-step ctx todo)

Basically leveraging reduced to short circuit reducing the list of steps

And as a bonus, you can use reductions to debug 🙂

I like it the reduce approach quite a bit. @slipset why do you say "I don't know if I'd recommend it". What are the downsides that you have from this approach that I may be missing?

Not sure, it seems to me like a nifty little trick which might lead to harder to read code?

I kind’a like the straight forwardness of the original code, and the reduce thing might feel a bit like golfing?

There is also https://github.com/fmnoise/flow But it feels in a way like you’re running into monadic territory.

Another approach would be to some-> and rely on nil returns from the fns you’re threading.

Which then makes you wonder wether nil-punning and monads are related, and there goes your evening 🙂

Or you could just have the fns throw exceptions, and you’d be good to go…

I love how readable threading macros are. When you're threading a single "ctx" map through them they are straight forward to understand. What I didn't like, is 1. Visibility - I couldn't easily see each step without a debugger or putting in printlns/debug statements. 2 - Error catching also is hard. With this reduce approach you handle 1 & 2, and the data structure you pass to it, it is just as easy to read as the threading macro. But, it has the added benefit that you can construct it a lot easier than the internal implementation of a threading macro.

Thanks for taking the time to answer this. Oh, I didn't know about reduced, that is neat. I'll experiment with that reducing approach; it looks like a simplified version of the think-do-assimilate.