cond->: Using of threading expression in tests?

[ru]

Hi,

Can I use in tests threading expression of cond->, and how?

Thanx in advance,

  Ru

[Colin Yates]

Hi Ru - I am not sure I understand the question - can you expand on the use-case a bit please.

--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clo...@googlegroups.com
Note that posts from new members are moderated - please be patient with your first post.
To unsubscribe from this group, send email to
clojure+u...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups "Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email to clojure+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[ru]

Hi, Colin

For example, (cond-> {:x 5} (> (get ??? :x) 3) (assoc :y 6))

пятница, 2 октября 2015 г., 17:49:28 UTC+3 пользователь ru написал:

[Colin Yates]

Ah I see - maybe https://clojuredocs.org/clojure.core/as-%3E will help (untried):

  (cond-> {:x 5} (as-> x1 (> (get x1 :x) 3)…)

[ru]

Thank you, Colin.

This works:

user=> (as-> {:x 5} x1 (if (> (get x1 :x) 3) (assoc x1 :y 6) x1) (if (<= (get x1 :x) 3) (assoc x1 :y 12) x1))

{:y 6, :x 5}

But without cond-> at all :( :)

пятница, 2 октября 2015 г., 17:49:28 UTC+3 пользователь ru написал:

Hi,

[Colin Yates]

Alternatively you can do (cond-> {:x 1} (fn [x] (= (:x x) 1)) (assoc :x 2))

[ru]

This is remarkable improvement and it shows that test is a function of the threading expression.

user=> (cond-> {:x 1} #(= (:x %) 1) (assoc :y 2))

{:y 2, :x 1}

also works.

Thanks a lot

пятница, 2 октября 2015 г., 17:49:28 UTC+3 пользователь ru написал:

Hi,

[Jason Felice]

This doesn't work, as (fn [x] (= (:x x) 1)) is never evaluated and the function itself is always truthy.  (Same for the #(= (:x %) 1) version).

[Sean Corfield]

We needed this functionality too and ended up writing our own utility macro:

(defmacro condp->
"Takes an expression and a set of predicate/form pairs. Threads expr (via ->)
through each form for which the corresponding predicate is true of expr.
Note that, unlike cond branching, condp-> threading does not short circuit
after the first true test expression."
[expr & clauses]
(assert (even? (count clauses)))
(let [g (gensym)
pstep (fn [[pred step]] `(if (~pred ~g) (-> ~g ~step) ~g))]
`(let [~g ~expr
~@(interleave (repeat g) (map pstep (partition 2 clauses)))]
~g)))

So (condp-> {:x 1} #(= (:x %) 1) (assoc :x 2)) produces {:x 2} as desired.

We have a condp->> version as well.

Sean Corfield -- (904) 302-SEAN
An Architect's View -- http://corfield.org/

"If you're not annoying somebody, you're not really alive."
-- Margaret Atwood

[Colin Yates]

Well I feel a numpty :-).

[ru]

So am I! Sean, can you share code for condp->> too, please.

пятница, 2 октября 2015 г., 17:49:28 UTC+3 пользователь ru написал:

Hi,

[Oliver Hine]

You may also find a use for condas-> which I describe in a mini blog post here: http://blog.juxt.pro/posts/condas.html

It combines cond-> and as-> which I find super useful when dealing with various fn type signatures.

[Sean Corfield]

So am I! Sean, can you share code for condp->> too, please.

Oh, sorry, I thought it was obvious from condp-> since the only difference is that it uses ->> instead of -> in one place:

(defmacro condp->>

  "Takes an expression and a set of predicate/form pairs. Threads expr (via ->>)

  through each form for which the corresponding predicate is true of expr.

  Note that, unlike cond branching, condp->> threading does not short circuit

  after the first true test expression."

  [expr & clauses]

  (assert (even? (count clauses)))

  (let [g (gensym)

        pstep (fn [[pred step]] `(if (~pred ~g) (->> ~g ~step) ~g))]

    `(let [~g ~expr

           ~@(interleave (repeat g) (map pstep (partition 2 clauses)))]

       ~g)))

Sean

[ru]

Thank you very much, Sean and Oliver!

condas-> really awesome!

пятница, 2 октября 2015 г., 17:49:28 UTC+3 пользователь ru написал:

Hi,