@ghadi @triss - here's a little more on debugging from within a go block or thread: http://stackoverflow.com/questions/32715869/clojure-log-to-console-when-code-is-being-executed-in-a-go-block-or-thread/32717490#32717490

I prefer (case choice :set value "") over (if (= choice :set) value ""). Am I strange?

Yep.

Just strange, or dangerously misled?

Extremely amazing and convenient that (assoc ["foo"] 1 "bar") adds an element to the vector!

why not using (conj [“foo”] “bar”)?

@delaguardo: It's convenient that I don't have to distinguish between modifying an element of the vector and adding one to the end.

Except for the first element 😕 (assoc nil 0 "bar") is {0 "bar"} instead of ["bar"]

because (not= nil []) )

fnil would work well there

and in doc section for assoc - Note - index must be <= (count vector).

for vectors

@bhagany: Thanks! I have (update-in db [:product :times] (fnil #(assoc % index time) [])) now

@mbertheau: alright, nice. Maybe I don't quite understand your goal, but it makes more sense to me to wrap the '%' in fnil… otherwise (I think) you're going to end up with an empty vector if you don't already have a vector, instead of a one-element vector

(assuming that's what you want)

like (update-in db [:product :times] #(assoc (fnil % []) index time))

bhagany: but % is not a function - am I missing something?

hahaha, you're right, I'm forgetting the signature

it's been awhile since I've had a need, but I remember it fondly

you were correct, I retract my "help"

@bhagany: Thanks for the effort anyway!

Hmm, for filter there's filterv, but no removev for remove?

Hi - really simple question -- why doesnt this code work: http://pastebin.com/cFu7juQH

you are basically calling (map inc 2) since you destructure the args [a [b]]

make it [a b] and it works

lein repl
nREPL server started on port 54991 on host 127.0.0.1 - 
REPL-y 0.3.7, nREPL 0.2.10
Clojure 1.7.0
OpenJDK 64-Bit Server VM 1.7.0-u60-unofficial-b30
    Docs: (doc function-name-here)
          (find-doc "part-of-name-here")
  Source: (source function-name-here)
 Javadoc: (javadoc java-object-or-class-here)
    Exit: Control+D or (exit) or (quit)
 Results: Stored in vars *1, *2, *3, an exception in *e

user=> #?(:clj 1 :cljs 2)

RuntimeException Conditional read not allowed  clojure.lang.Util.runtimeException (Util.java:221)
IllegalArgumentException Wrong number of args passed to keyword: :clj  clojure.lang.Keyword.throwArity (Keyword.java:97)

I can iterate over a map with for: (for [e {:a 1 :b 2}] e). How can I put the result back into a neat map?

can use into

(into {} (for [e {:a 1 :b 2}] e))

Ah! Thanks. I wish there was a for-into or for-map.

Yeah, there are utility libraries like https://github.com/Prismatic/plumbing that provide map-keys and map-vals.

Hmm, I have a map {:mon {:a true :b [2 4 3]} :tue {:a false :b [1 5 2]}} and I want to return the same map with :b of the weekday ordered, but leave the current weekday (`current-weekday`) alone.

That seems so easy to put in words, but hard to express with clojure. I have (reduce-kv #(assoc %1 %2 (case %2 current-weekday %3 (update-in %3 [:b] order-b))) {} my-map)

If you are trying to update a value in a map, you probably want an update-in like function. (update-in {:a {:b 10}} [:a :b] #(+ 5 %)) ;=> {:a {:b 15}}

@mbertheau: Can the ordering you want be permanent? If so perhaps easier to use a sorted-map... (not sure your business use)

@mbertheau: maybe specter can help you there https://github.com/nathanmarz/specter

(require '[com.rpl.specter :as sp])
(sp/transform [sp/ALL sp/LAST :b] (comp vec sort) {:mon {:a true :b [2 4 3]} :tue {:a false :b [1 5 2]}})
=> {:mon {:a true, :b [2 3 4]}, :tue {:a false, :b [1 2 5]}}

😊 great minds think alike

anybody using pedestal here? I have an url like "/campaign/:code-name/", and requests to that url with trailing slash return 404. While requests without trailing slash are handled normally. Any ideas how can I handle trailing slash?

explicit route that 301s? never alias bad for seo

^ so am told

@asolovyov: they’re different routes in Pedestal’s world. i recommend pinging @ohpauleez on twitter for a quick answer

they're different routes in any world. I think the think that apache does and I guess nginx with try_files followed by index directive leads to this idea that both should work as some standard practice

@robert-stuttaford: yeah, I had an interceptor strip-trailing, which stopped working... 😕

well, the problem is that pedestal just drops trailing slash

@asolovyov ah I see, that is odd

when I try to put an interceptor which would strip trailing slash from requests, it never gets called - pedestal returns 404 before that

I wonder if I should update my :path-re by hand... 😕

when I want override a specific implementation of a multimethod in a test, is defing it in the test namespace “clean”? are there other ways?

@martinklepsch: you can alter-var-root on it, or I think you can setDynamic on the var and then use binding

unsure about the latter though

@tcrayford: what would I bind it to though? if I alter-var-root the var/dispatcher the already defined implementations will probably just stick around?

@noonian, @cjmurphy, @meikemertsch: Thanks for your input :) specter looks like the most readable. I'll see if I can get the conditional in as well. Thanks

😊

@martinklepsch: rebind it to a function, and the already defined implementations do stick around, but you don't hit them

@tcrayford: if I do that all other implementations will not work anymore and they’re needed 😄

ah

@martinklepsch: in which case I'd do the typical thing of "change your design so that this kind of var based thing can be swapped out in tests without actually using vars"

this is about clojure.test/report — guess I’ll stick with def’ing it in the test namespace

@martinklepsch: I hate clojure.test/report

bolth directly monkeypatches the method definitions in it: https://github.com/yeller/bolth/blob/master/src/bolth/runner.clj

but that's suitable for a test runner, and probably not for what you're doing

@martinklepsch: I am ashamed to admit it, but I got around this by declaring a ‘normal’ defn which simply dispatched to a defmulti and then in tests overrode the defn. For an individual defmethod I had the implementation call out to a defn- which I could override in a test … yeah, not great but it worked

@colin.yates: funky! 😄

specifically the usecase is allowing users of test.chuck to customize how their smallest failing thing is getting printed

@martinklepsch: huh, so they could pretty print it? Is this for checking or something else?

(yeller has a fork of checking internally)

@martinklepsch: btw Gary and I have been talking a lot about making defspec work significantly better with clojure.test, including incorporating the ideas in checking with regards to report and such. It seems like pretty printing or whatever would also be a good thing to throw in that (might wanna ping him about it)

yes for checking… Mostly stumbled upon it because I tried making collection-check and then test.chuck work with cljs

@tcrayford: have some PRs here https://github.com/gfredericks/test.chuck/pulls (in good company haha)

@martinklepsch: that must be a lein thing - this works in 1.7 without lein

@martinklepsch: in reference to your reader conditional repl question earlier that is

Now that we have update shouldn't there also be an update! for transients?

? ¯\(ツ)/¯

if you want to file a jira, you could see if anyone votes for it as useful. not sure what Rich would think about it.

Done: http://dev.clojure.org/jira/browse/CLJ-1848

I'm fully aware this is a somewhat stupid question, but is there any JVM/Clojure hook that gets called on program exit?

@jaen: I think Runtime.addShutdownHook() might do - e.g. https://stackoverflow.com/questions/11709639/how-to-catch-ctrlc-in-clojure

@itruslove: looks legit, thanks

jaen: Works great, but won’t be invoked with “lein run” then Control-C. Does work is you “lein trampoline run” before Control-C.

I don't use lein, but still - good to know.

jaen: Yeah, one can get, uh, frustrated, wondering why it doesn’t work before one tries running from an uberjar or something. ;)

Yeah, I imagine it's probably not so obvious to get why it's not working

quiz: I have a vector [:a :b :c :d] and an integer 6, and I want to “extend” the vector repeating its last argument until its length is 6, like this: [:a :b :c :d :d :d]. What’s the most concise way?

(vec (take 6 (concat v (repeat (last v)))))?

wow, nice

I had (map-indexed (fn [i e] (nth v i e)) (repeat n (last v)), v being my original vector and n the integer

yours is better! than you @eraserhd 😄

*thank

welcome 😄

@eraserhd: +1 nice example of thinking functionally rather than idiomatically

or maybe ‘declaratively’ is a better work than ‘functionally'

this one’s not as good as @eraserhd’s but: (into v (repeat (- n (count v)) (last v))))?

how about this: (if (<= 6 (count v)) v (ask-slack)))

OH: Col, you need to stop trying to be funny!

@colin.yates: good one! that’s the one I went with 😉

if you have a vector, never use last instead use peek

@ghadi oh?

last is O(N), peek is O(1) on vectors

oh snap

last is a seq operation, uses linked list traversal

well i’m convinced

unless

you're in a macro... then who cares, it's run only once. butlast is prevalent in macros

are macros really only run once?

if you’ve got something like (defn foo [& args] (my-very-inefficient-macro…)), wouldn’t the macro get run every time the function gets called?

only once, the compiler runs the macro and uses the result to create the runtime function

It's worth mentioning that if e.g. last is used in the code returned by the macro (i.e. if it's in the macro-expanded form), it will be run whenever the macro is run. But if it's just in the code that generates the expansion, it will only be run once.

At least, I'm pretty sure that's how macros work…

@swizzard: macros are expanded at compile time

unless you're doing weird stuff with eval (that you shouldn't be), macros won't expand at runtime

makes sense, i guess

honestly i still don’t really ‘get’ macros, especially when compared to e.g. HOFs

i think i’ve only ever written one in the year or so i’ve been using clojure

it helps to think of macros as compile-time functions that generate functions

@colin.yates: it probably helps because that's what they are

or think of them as the ‘text-expander’ utilities that you get on OSX

macros are literally implemented as functions run by the compiler rather than the runtime

@bronza . They aren’t often described like that though

no they are not

well s/generate functions/generate code/ I guess would be more accurate

yeah, presumably the defwhatever-type macros you get in e.g. korma don’t necessarily generate functions

@swizzard: when learning lisps, I found that it helped me quite a lot to understand macros by understanding that there's nothing magical about `

yep - code rather than functions - my bad

and that's it's only used in macros because it's convenient, rather than because it's needed

so to take the usual example of

(defmacro unless [cond & body] `(if (not ~cond) body))

my first and (probably) last macro: https://groups.google.com/d/msg/clojure/jQoXknHxu44/aABKUfVtBwAJ

if you think of it as a function (defn unless [cond & body] (list* 'if (list 'not cond) body))

and realize that the difference between this function and the defmacro is just: 1- we don't need to use "` ", it's not special syntax it's just a convenience to avoid nesting quotes and calls to list

2- it just takes implicitely quoted args and the result is implicitely evaluated

than macros really are as simple as functions

yeah

i think part of the problem is that i haven’t had enough use cases where the pain of writing a working macro is less than just using HOFs or whatever

most experienced clojure devs would tell you that if you can use a HOF, you shouldnt' write a macro for that

yeah

like i said, i’ve only written the one, and it was just ->-but-takes-a-seq

thanks for the insights!

@bronsa: +1

Is it just me or does defonce not accept a docstring?

@jaen I believe you have to provide it as normal metadata: (defonce ^{:doc “Hello”} foo (+ 1 2))

That's kinda... arbitrary. But at least that's a workaround, thanks.

there's a ticket for this - turns out to be surprisingly complicated to fix

Interesting, why would it be so? No clear agreement on semantics?

@jaen: http://dev.clojure.org/jira/browse/CLJ-1148 is the ticket

you can follow the comment/patch trail there

can someone explain what happens to a record, when it gets evaluated?

(defrecord Kikka [])
(= (->Kikka) (->Kikka)) ; true
(= (->Kikka) (eval (->Kikka))) ; false

there were changes in this area in 1.8 - I think you will now see true for both

https://github.com/clojure/clojure/commit/1d5237f9d7db0bc5f6e929330108d016ac7bf76c in particular

alexmiller: thanks! just tested with 1.8.0-RC1 and both give true.

I think for a records with at least one element, you would have seen true for both even in 1.7

this was the special case of records or types with no fields (which is not common)

So the weird thing about using test.check to verify how rewrite-clj does things is that I’m generating random clojure programs.

what about this one (with 1.8.0-RC1):

(= pos? (eval pos?)) ; true
(= [pos?] (eval [pos?])) ; false

relates to the real life case I’m trying to figure out: (= schema.core/Keyword (eval schema.core/Keyword)) ; false. Keyword is a record with clojure.core/keyword? in it. Evaluating the record loses it’s equality.

I am looking for a way to kill the repetition in this file: https://gist.github.com/eraserhd/bcdc5a04a54f96e95ab2

I don’t even want the functions with stars appended.

All I can think of so far is a macro that takes a map and generates all the things, but this seems pretty ugly.

You might be interested in this library that helps create test.check generators: https://github.com/holguinj/jen

@noonian Does this help with recursive generators?

@eraserhd: yep, there is an example in the readme. YMMV as I haven’t used it myself. But I know the author and watched him give a talk on the library. Its motivation was making it easier to create complex generators.

Hrmm…

It appears that all it does is call gen/recursive anyhow. I think what I want is letrec for generators.

ah, sorry for the noise then 😛

@noonian: np, I mean I came here for ideas

well, I was thinking you could pull the node into an arg and just have one higher order function that you call with the different nodes in container node

to get rid of the asterisk functions

Ah.

I’m going to try.