#clojure log - Oct 06 2012

The Joy of Clojure
Main Clojure site
Google Group
List of all logged dates

0:02 aperiodic: arrdem: what, gg=G no good?

0:07 Sgeo: How often is let with mutable over lambda done in Clojure?

0:30 Is there a way to suggest changes to Clojuredoc?

0:31 Pretty much to hunt down all uses of #{} where there should be a note about it vs. contains?

0:42 Is it supposed to be interesting that it's easy to nest ->> inside -> but not the other way around?

1:08 duck1123: is anyone here using lolg to do their cljs logging, and if so, have you noticed the newline after every message?

1:09 although I'm thinking it might be straight up goog.debug.Console that's doing it

1:19 arrdem: aperiodic: never seen that one before...

1:19 aperiodic: arrdem: what, =? it's pretty handy

1:20 arrdem: can you explain that?

1:20 I've never seen it

1:21 and it has sudenly selected everything

1:21 0.o

1:21 aperiodic: that doesn't seem right

1:21 gg moves you to top of file

1:21 = is the formatting command

1:22 G moves you to the bottom

1:22 arrdem: k... according to SO = only fixes indentation

1:22 ah

1:22 aperiodic: oh, i thought that's what you meant

1:22 arrdem: I haz idea

1:25 bingo

1:25 hanfway there, thanks!

1:25 aperiodic: sure thing

1:26 arrdem: crap..

1:26 Imma just eval and pprint this crap

1:49 Sgeo: I have it.

1:49 I know what my ideal syntax for using monads looks like, and how it works.

1:54 It would probably be best to use jimduey's protocol monads

1:58 And relies on delimc

2:55 Roxxi: Hey, I got a riddle for anyone who's still awake... https://www.refheap.com/paste/5569

2:55 I've spent 2 hours suffering at the hands of this...

2:59 ivan: Roxxi: you must use the return value from assoc!

2:59 Roxxi: you can't just bang on the transient like that

3:00 Roxxi: Oh?

3:00 crap.

3:00 lol

3:00 I thought that was the whole point?

3:02 ivan: http://clojure.org/transients see two paragraphs before Example

3:03 Roxxi: I'm abusive, what can I say?

3:03 Thank you *so* much ivan.

3:04 ivan: I did not believe it when I read it in Clojure Programming. that ! is very confusing.

3:05 Roxxi: Heh. For whatever reason, that didn't stick with me either. I just thought it was the same as calling mutative methods.

3:08 Soooo what's a good syntatic element to use to iterate through a sequence, but so that I can use the return value each time... reduce?

3:33 brainproxy: having little luck getting clojure.pprint/write to enforce a right margin

3:34 what I want is for "submaps" of the map to get likewise pretty-printed

3:34 maybe right-margin isn't what I'm looking for ...

6:37 john2x: is it possible to use a function with arity 2 with `map`, but pass a "fixed" argument for the second parameter?

6:41 ah `repeat` does the trick for the second parameter..

7:04 _ulises: john2x: alternatively you could use an anonymous fn and fix the parameter yourself in there?

7:21 jaju: Hello nice folks!

7:21 I need some help with my IDE setup - emacs, nrepl, lein.

8:51 casion: does anyone know of any places that discuss clojure.core.reducers? I need a bit of a boost in understanding their usefulness and application

9:00 fredyr: casion: have you looked at the screencast about the library?

9:00 or presentation really

9:00 casion: fredyr: no, I have't found that yet

9:00 or didn't notice it :)

9:00 fredyr: hang on, i'll see if i can find the link

9:01 oh here it is

9:01 http://www.infoq.com/presentations/Clojure-Reducers

9:01 casion: awesome, thank you

9:02 fredyr: yw

9:12 wingy: if i want to put a new element in a sequence should i use conj or cons?

9:12 they both give the same results

9:12 (conj old-sequence 4)

9:12 (cons 4 old-sequence)

9:16 TimMc: wingy: It depends on which abstraction you are using: Sequence or polymorphic collections.

9:19 If you know it's a sequence, use cons. If you want conj's behavior for a known collection type, use conj. If you have a collection of known or unknown type but definitely just want a sequence back with the new element on the front, use cons.

9:23 wingy: so it depends on what i want back as well

9:23 makes sense

9:45 antoineB: hello

9:46 gfredericks: ~hello

9:46 clojurebot: BUENOS DING DONG DIDDLY DIOS, fRaUline gfredericks

9:47 antoineB: bonjour

9:47 i would like to know how to do "window.test = 12;" in clojure script?

9:48 (. set! (js/window test) 12) doesn't work

9:51 wingy: wouldn't the ability to call (.equals-ignore-case str1 str2) instead of (.equalsIgnoreCase str1 str2) be sweeter?

9:52 gfredericks: antoineB: (set! (.-test js/window) 12)

9:52 wingy: there'd be a bit of translation ambiguity there, no?

9:53 wingy: yeah

9:53 antoineB: gfredericks: where did you find the doc?

9:54 gfredericks: antoineB: oh I'm just familiar with cljs

9:54 antoineB: http://himera.herokuapp.com/index.html has it on the front page

9:55 casion: why is it .-property?

9:55 antoineB: ok thanks

9:55 casion: I've never touched cljs, just curious

9:56 gfredericks: casion: have to distinguish from (.property foo) which calls property as a function

9:56 (.property foo) => foo.property()

9:56 casion: ahh, I see

9:56 gfredericks: (.-property foo) => foo.property

9:57 antoineB: it doesn't use the clojure syntax

9:57 gfredericks: clojure doesn't have a syntax for that

9:57 or maybe they added the .-foo stuff recently

9:57 I'm not certain whether the JVM has such cases

9:57 antoineB: (. set! (js/window test) 12)

9:57 it is the clojure syntax

9:58 * gfredericks isn't familiar with that

9:59 casion: java uses getters/setters

9:59 so variable access is a method call (usually)

9:59 antoineB: oh i make a mistake the syntax is : (set! (. instance property) value)

10:00 getters/setters is a convention not a "have to"

10:01 casion: that's why I said 'usually'

10:03 gfredericks: is "foo.bar() = 12" a legal thing to do in JS?

10:03 I can't imagine what it would mean

10:03 antoineB: no

10:04 gfredericks: oh well we still need a syntax for accessing properties

10:04 in clj do we just use (. instance property)?

10:06 antoineB: (.-property instance)

10:08 gfredericks: well that matches cljs

10:08 casion: I've never seen .-property in clj?

10:10 http://clojure.org/java_interop#set

10:11 I sure hope that's correct, since the code I'm working on now does it that way and it appears to work ;)

10:16 wingy: how do i get the current timestamp for clj?

10:16 so i can save it in datomic

10:19 casion: ,(.toString (java.util.Date.)

10:19 clojurebot: #<ExecutionException java.util.concurrent.ExecutionException: java.lang.RuntimeException: EOF while reading>

10:19 casion: meh, that should work

10:20 ,(.toString (java.util.Date.))

10:20 clojurebot: "Sat Oct 06 14:13:20 UTC 2012"

10:23 wingy: ,(java.util.Date.)

10:23 clojurebot: #<ExecutionException java.util.concurrent.ExecutionException: java.lang.RuntimeException: java.lang.ExceptionInInitializerError>

10:24 wingy: casion: in the repl it returns Sat Oct 06 16:16:49 CEST 2012||

10:24 how can i return the timestamp for the db?

10:24 casion: iirc date always returns utc

10:25 wingy: so the repl is converting it to human friendly output?

10:25 clojurebot: clojure.repl in swank is not useful for two reasons: 0) everything it provides has an enhanced version in slime and 1) having to re-refer it every time you changed namespaces would be annoying

10:42 wingy: casion: http://joda-time.sourceforge.net/index.html seems nice

10:44 and https://github.com/seancorfield/clj-time

11:01 antoineB: ,(for [[x y] '(1 2 3 4 5 6) (+ x y))

11:01 clojurebot: #<ExecutionException java.util.concurrent.ExecutionException: java.lang.RuntimeException: Unmatched delimiter: )>

11:01 antoineB: ,(for [[x y] '(1 2 3 4 5 6)] (+ x y))

11:01 clojurebot: #<ExecutionException java.util.concurrent.ExecutionException: java.lang.UnsupportedOperationException: nth not supported on this type: Long>

11:03 gfredericks: antoineB: you're destructuring a number

11:03 ,(let [[x y] 1] x)

11:03 clojurebot: #<UnsupportedOperationException java.lang.UnsupportedOperationException: nth not supported on this type: Long>

11:03 antoineB: yes just understand it

11:04 i can use partial for what i want

11:04 other idea?

11:04 i mean partition not partial

11:04 gfredericks: antoineB: I couldn't tell what you were actually trying to do

11:05 antoineB: have x = 1 and y = 2 for the first loop, x = 3 y = 4 for the second etc...

11:05 (for [[x y] '(1 2 3 4 5 6) (+ x y))

11:05 gfredericks: ah; yeah partition 2 would work fine

11:07 Sgeo: partition?

11:07 clojurebot: partition is probably not what you want; see partition-all.

11:07 dlitvak: clojurebot: you dont know that!

11:07 clojurebot: No entiendo

11:07 dlitvak: :P

11:08 gfredericks: clojurebot: partition |might| be what you want. But also there is partition-all.

11:08 clojurebot: Ack. Ack.

11:08 antoineB: i assume to be even

11:08 gfredericks: then you are good

11:09 partition-all doesn't make sense in a lot of contexts

11:09 sexpGirl: I've got a quick question regarding my "workflow" if I want to use Emacs / nrepl / nrepl.el and also do some unit testing. I'd like to have both a nrepl.el Emacs buffer opened and be able to run unit tests. How should I do it? Should I run some terminal next to my Emacs and launch the tests from there? If so, which command should I use to run the tests?

11:09 dlitvak: sexpGirl: lein test

11:10 sexpGirl: considering that you started your project using leiningen

11:10 sexpGirl: dlitvak: so when running the tests I'd have two JVMs running? One for the REPL and one for the test?

11:10 dlitvak: yup, yup, Leiningen ; )

11:10 dlitvak: sexpGirl: you can also run the tests from the repl

11:10 but im a bit unsure on how that was done

11:11 sexpGirl: dlitvak: ok, I'll go read about "lein test" and then try that

11:31 technomancy: sexpGirl: `lein test` isn't good for a interactive workflow; it's just a sanity check

11:31 you should use clojure.test/run-tests from the repl during development

11:31 antoineB: do you often have name clash in clojure? (eg: (defn abc [name] (name :test))

11:31 )

11:31 technomancy: spinning up a new lein just to see if your change broke something is way too slow

11:32 gfredericks: antoineB: you mean accidentally shadowing clojure.core/name? It happens to me sometimes yes.

11:33 antoineB: yes

11:33 technomancy: antoineB: it basically only happens with name

11:33 antoineB: is there some technics to avoid it? (now i just rename the args)

11:34 technomancy: type inference =)

11:35 antoineB: there is no type inference in clojure

11:35 (and maybe more in clojurescript)

11:35 technomancy: well there is, it's just not done yet

11:36 antoineB: what?

11:36 clojurebot: what is bla

11:36 technomancy: clojurebot: google typed clojure

11:36 clojurebot: First, out of 13700 results is:

11:36 frenchy64/typed-clojure · GitHub

11:36 https://github.com/frenchy64/typed-clojure

11:36 dnolen: antoineB: I'm not sure that it's true that Clojure doesn't do a little bit of type inference here and there.

11:37 antoineB: CLJS definitely has some where it needs it

11:37 technomancy: clojurebot: forget what |is| bla

11:37 clojurebot: I forgot that what is bla

11:40 technomancy: antoineB: anyway the real answer is you just learn what the error message for that problem looks like so that when it occasionally happens you can spot it pretty much instantly

11:40 antoineB: technomancy: i write it in the "stuff to check when error"

11:41 technomancy: in practice it happens fairly rarely

11:41 antoineB: about frenchy64 i don't understand what it could be used for

11:45 chouser: I had the same kind of problem in python the other day, but didn't recognize the error message. Took several minutes to find it.

12:14 antoineB: is there a convention for mutable method? maybe "!"

12:15 akhudek: antoineB: that's right, ending in "!" is the convention

12:16 http://dev.clojure.org/display/design/Library+Coding+Standards

12:22 mindbender1: emezeske: is there a way of programmatically telling cljsbuild to check if there any modified files and compile if any

12:24 antoineB: mindbender1: i remmember seeing that

12:25 akhudek: cljsbuild auto

12:25 or wait, I suppose you don't mean the command line

12:25 mindbender1: antoineB: I need to make that call from within my code.

12:25 yes

12:25 akhudek: take a look at the cljsbuild plugin source

12:26 mindbender1: I'm there trying to figure it out

12:26 akhudek: line 128

12:26 https://github.com/emezeske/lein-cljsbuild/blob/master/plugin/src/leiningen/cljsbuild.clj

12:27 It's private though :-/

12:27 but you can probably call cljsbuild

12:27 clojurebot: I don't understand.

12:28 akhudek: and just specify auto as a subtask

12:28 mindbender1: akhudek: I'll try that and see what I get get. Thanks for the link

12:29 akhudek: (leiningen.cljsbuild/cljsbuild project "auto")

12:29 mindbender1: akhudek: maybe for project I only pass relevant info in a map

12:30 akhudek: at minimum it needs the build options, but probably other things too

12:31 mindbender1: yeah

12:36 akhudek: (leiningen.core.project/init-project (leiningen.core.project/read)) can get me the full project

12:37 akhudek: nice!

13:34 adrian_: any tips for abstracting over mutable datastructures? like implementing DOM uis?

13:35 need to rearrange dom trees in response to events

13:35 maybe just suck it up and define lots of ! functions?

13:35 akhudek: adrian_: maybe check out enfocus?

13:38 adrian_: hmm I was avoiding it because I didn't like the idea of using css selectors to find dom nodes I'd generated

13:40 casion: adrian_: I'm doing something similar in clj and I'm using enlive. I'm not famliar with enfocus but if it's comparable then it's rather simple to do

13:42 I did end up writing 'a few' functions to wrap common selectors, but it's resulted in very easy to read code

13:42 adrian_: yeah I think enfocus is based on enlive

13:42 Frozenlo`: You can also check domina (enfocus is built on top)

13:43 adrian_: using domina

13:43 this might be more what I'm after:

13:43 http://www.haskell.org/haskellwiki/Functional_Reactive_Programming

13:48 akhudek: adrian_: There was a lot of discussion about that on the dev list not to long ago, but I'm not aware of any commonly accepted solution yet. There are a few libraries trying to implement bits of reactive programming though. e.g. https://github.com/ohpauleez/shoreleave pub/sub

13:49 also http://dev.clojure.org/display/design/Reactive+Programming

13:50 adrian_: cool, thanks

13:57 lynaghk: ping: dnolen

13:58 b

13:58 dnolen: lynaghk: pong

13:59 lynaghk: dnolen: So I've run into another situtation where I want to do fancy map matching

13:59 beyond my super hackery of core.match as an event bus.

14:00 Basically, I want to do nested map matching and then rewriting using a set of rules.

14:01 Something like, "if an associative object has these (potentially nested) field/value pairs", I want to grab these bits into logic vars and splat them out over here.

14:01 I took a look at the unification stuff in core.logic, but it was too strict: e.g., {:a ?x} wouldn't "match" with {:a 2 :b 4}

14:02 dnolen: lynaghk: yeah still pondering how to support that in core.logic, likely a new map like data type that isn't so strict about unification.

14:03 lynaghk: Anyway, my question is whether 1) loosening this up is fairly straightforward and 2) if you think rules like that would be maintainable in the large

14:03 ah. Yeah, I'm hoping to do a very general kind of matching that includes record types and fields within the records.

14:04 dnolen: lynaghk: it is pretty straightforward - make a new type that unifies with maps, handle both (== u v) && (== v u) - the unification on this type doesn't care that the count doesn't match, just that the keys presents in the new type are prsent in the map.

14:04 lynaghk: there's actually a ticket for this in JIRA

14:05 lynaghk: dnolen: here's a bit more context about what I'm actually trying to do: https://gist.github.com/971172f891a8de633203

14:06 dnolen: ah, cool. I'll look into that, thanks.

14:08 dnolen: lynaghk: this sounds very interesting.

14:09 lynaghk: I'd happily take a patch for the ticket - http://dev.clojure.org/jira/browse/LOGIC-49

14:09 lynaghk: dnolen: I think it'd be a lot of fun, yeah. The biggest problem I'm grappling with now is explicit (which I like as a programmer) vs. smart implicit (which is what users will want...until it turns into Clippy)

14:09 dnolen: lynaghk: http://github.com/clojure/core.logic/blob/master/src/main/clojure/clojure/core/logic.clj#L1502

14:10 lynaghk: map unification is quite simple - it should be easy to see how it should be changed to do what you want for this new type.

14:10 clojurebot: c'est bon!

14:10 lynaghk: We should be able to extend to anything that is associative, no?

14:10 dnolen: lynaghk: you need a new type say SubMap (horrible name)

14:11 lynaghk: I think you should stick with maps, because vectors are associative too.

14:11 IUnifyWithSubMap

14:11 lynaghk: dnolen: why not try and get vectors in on the fun?

14:12 dnolen: lynaghk: I need to think about it some more - but I guess it should be ok

14:13 lynaghk: I would be concerned with protocol resolution ambiguities - but I think since we have a concrete type SubMap or some such - you'll probably be ok.

14:13 lynaghk: dnolen: yeah, definitely. I'm still very new at all this logic programming---I'm at the "boy, it'd be nice if everything was declarative and pattern-matchy" but haven't actually done serious work with that mindset so I dunno about hidden issues.

14:14 dnolen: lynaghk: one thing that's a bit weird is that is the type of the reified value.

14:15 lynaghk: so it's might be a bit strange that a map like thing unifies with a vector and you get a map like thing out of unification - not a sequential thing.

14:15 lynaghk: kind hard to say if that would be problem in practice w/o tring it out.

14:17 lynaghk: dnolen: yeah, I'll start kicking it around if I'm convinced that rules-rewriting is a sane thing to do in the first place = )

14:18 dnolen: lynaghk: my gut instinct is that it should only work with maps.

14:18 lynaghk: you can easily try it out locally by extending to vector

14:18 lynaghk: if you don't run into trouble we could consider loosening the behavior.

14:19 lynaghk: The use case I'm thinking of for my pattern matching is to make a rule that matches, e.g., a scale that starts at 0 (i.e., a record with one field that is a vector whose first element is 0)

14:19 dnolen: lynaghk: the rules rewriting sounds kinda awesome to me :)

14:19 puchacz: hi, has anybody put clojure inside postgres process using http://pgfoundry.org/projects/pljava/ ?

14:19 lynaghk: dnolen: until there are 200 and you don't know what is being invoked to change your spec...

14:20 dnolen: lynaghk: yes, I think I see what you're going for. But if you want to try for a patch, got for the map behavior and loosen it locally to vector in your project.

14:21 lynaghk: after some time if it seems OK to you I'd be open to considering it further.

14:22 lynaghk: dnolen: okay, rad. I'll keep you posted.

14:22 gotta run, thanks for the pointers

14:23 dnolen: lynaghk: np

14:45 Jacob_Goodson: Hey everyone

14:46 I need some help with understanding macros

14:46 => (defmacro wrapper [x] `'[(~x :seq)]) #'user/wrapper => (for [x [1 2 3]] (wrapper x)) ([(x :seq)] [(x :seq)] [(x :seq)])

14:47 hmm

14:47 terrible format

14:47 Can someone help me?

14:48 dnolen: seems like you want - (defmacro wrapper [x] `[(list ~x :seq)])

14:50 Jacob_Goodson: I just want to be able to put code in between other code, the seq is there to test if it works

14:50 ill try what you gave me

14:50 thank you

14:50 dnolen: ,'[(x :seq)]

14:50 clojurebot: [(x :seq)]

14:51 dnolen: Jacob_Goodson: you're macro is quoting the following expression, so that's what you're going to get

14:51 ,(let [x 1] '[(x :seq)])

14:51 clojurebot: [(x :seq)]

14:51 dnolen: ,(let [x 1] [(x :seq)])

14:51 clojurebot: #<ClassCastException java.lang.ClassCastException: java.lang.Long cannot be cast to clojure.lang.IFn>

14:51 dnolen: ,(let [x 1] [(list x :seq)])

14:51 clojurebot: [(1 :seq)]

14:53 Jacob_Goodson: thank you very much for your help(all who helped)

14:53 why does list work?

14:54 dnolen: Jacob_Goodson: quoting an expression will produce a list with symbols unevaluated. So it does *two* things.

14:55 by just using list, symbols will be evaluted to their values.

14:55 Jacob_Goodson: hmm

14:55 I thought list just returned a list lol

14:56 seems very useful in a macro

14:57 dnolen thank you once again

15:01 dnolen: Jacob_Goodson: a list does just return a list

15:01 Jacob_Goodson: but you were writing '(... stuff ...) which also returns a list

15:02 (= '(1 2) (list 1 2))

15:02 ,(= '(1 2) (list 1 2))

15:02 clojurebot: true

15:02 dnolen: ,(let [x 2] (= '(1 x) (list 1 x)))

15:02 clojurebot: false

15:02 Jacob_Goodson: so it jsut forces evaluation of a symbol?

15:02 when used the way we did?

15:03 dnolen: Jacob_Goodson: it's quoting that prevents evaluation of a symbol.

15:05 ,'[x 1]

15:05 clojurebot: [x 1]

15:05 dnolen: ,'{x 1}

15:05 clojurebot: {x 1}

15:08 dgrnbrg: dnolen: I just made a breakthrough with core.logic tracing

15:08 I have successfully traced zebrao

15:08 dnolen: dgrnbrg: whoa!

15:09 dgrnbrg: like you produced the search tree for it?

15:09 dgrnbrg: yeah

15:09 I made a few changes to handle recursion and unification

15:09 dnolen: dgrnbrg: insane :)

15:10 dgrnbrg: that's really awesome - would love to see that work get folded in!

15:10 dgrnbrg: I thread the seen set through all the goals, so that you don't need to retrace an already-seen goal (although the mplus thunks still are a problem)

15:10 jamii: Well this is weird - https://gist.github.com/3845808

15:10 dgrnbrg: want to see the output of zebrao?

15:10 dnolen: dgrnbrg: hells yeah!

15:10 dgrnbrg: I'd love to know if there's something i could do differently to recover more/better info

15:11 dnolen: dgrnbrg: btw, I'm up for grabbing a drink & chat on Monday, sorry for the slow reply.

15:11 dgrnbrg: no problem--sounds great :)

15:12 https://gist.github.com/3845814

15:12 I made a new type of exp called a :nofollow

15:12 that's things like project, unify, etc

15:12 things that you don't want to call into

15:13 dnolen: would you like the zebrao code alongside, or do you know it well by now ;)

15:14 dnolen: dgrnbrg: that makes sense. sure!

15:14 dgrnbrg: You see, the one thing that i'm still overtracing is the mplus's streams

15:16 do you have any idea for how I could give the mplus's continuation thunk a new "seen" set?

15:16 jamii: Ok, turns out I'm not crazy - https://gist.github.com/3845808

15:17 dgrnbrg: since currently each branch has no idea about the other branches, since I think that their substitutions are fixed to be identical when the conde splits their clauses

15:18 dnolen: dgrnbrg: ah yes ... I'm assuming you're using something based on my approach where seen is an atom or something?

15:18 dgrnbrg: actually, it's all purely functional

15:19 seen is an mvalue

15:19 so in bind I descend to the next goal while carrying along the seen set

15:19 that's why you have [:goal :seen] after the first time you see a goal in every bindchain

15:20 but you don't have that for the mpluses

15:20 dnolen: dgrnbrg: nice, so at mplus can't you just go through them independently?

15:20 dgrnbrg: so i'm looking at the implementation of conde

15:21 which does (mplus* ~@(bind-conde-clauses a clauses))

15:21 which is going to look like (mplus* (bind* a clauses1) (bind* a clauses2) (bind* a clauses3))

15:22 hang on, i've got to disable spelling correction on my mac

15:22 this is killing me to type

15:24 dnolen: dgrnbrg: ah right ...

15:24 dgrnbrg: so this is something you've written from scratch yourself then?

15:24 dgrnbrg: right, so the issue is that the implementation of mplus in the TraceSubstitutions monad is (mplus [this f] (if (seen (class f) (tracer (conj _tree [mplus :seen] seen)) (let [new-subst (unwrap (f))] (merge-substs this new-subst))

15:24 it's heavily based on the code you wrote in the "interp" branch of core.logic

15:25 but i had to change some things to play nicely with unification and mplus

15:25 to trace fewer paths, since the code you wrote enumerated all paths and could hit stackoverflow in certain cases, depending on the use of mplus

15:26 the reason that happened is that it didn't carry the "seen" set to the subtree, only to the parent

15:26 but you can explore both directions

15:26 when i did that, I realized that I had to change where I added the :conde to the search tree, since it messed up the tree rendering code

15:26 so I moved the detection of mplus branches into the mplus method

15:26 dnolen: dgrnbrg: ah right ... to be honest I'm a bit confused at how interp.clj works - I did that very quickly as a proof of concept :)

15:27 dgrnbrg: give me 20 minutes to annotate my code and push it to github

15:27 dnolen: dgrnbrg: excellent.

15:31 doomlord: is 'lighttable' any good

15:31 dnolen: doomlord: it's pretty sweet for playing around with Clojure

15:31 doomlord: parts of it look like visual studios' "code definition window" on steroids

15:32 dnolen: doomlord: there's a lot of work that needs to be done before it's good dev environment. it's promising that the team uses it to build it.

15:32 doomlord: ah "dogfood"

15:32 dnolen: doomlord: even in it's current form it's a pretty impressive ClojureScript app.

15:41 dgrnbrg: dnolen: here's my shot at documenting what I did: https://github.com/dgrnbrg/core.logic/blob/dgrnbrg-interp/src/main/clojure/clojure/core/logic/interp.clj

15:41 there's a bit of code in core.clj too, but that's just some interfaces with tracing metadata

15:42 it also has a lot of desugared core.logic :)

15:43 you can see in the results that it traces the mpluses in parallel, so you get duplicate tracing of goals that appear in multiple branches of the same mplus

15:44 it'd be reasonable to clean that up when replanning for parallelization, but it'd be even cooler if that cleanup step could be skipped

15:44 it's possible that the automatic interleaving precludes that, though, since the substitutions are currently closed over, and it's be a huge refactoring to change that

15:44 I suppose also the tracing could be done with an atom, which would also solve things

15:46 jamii: This makes me so happy:

15:46 (deftest self-describing

15:46 (= pattern-ast (vec (parse pattern-syntax)))

15:46 (= seq-pattern-ast (vec (parse seq-pattern-syntax))))

15:49 dnolen: dgrnbrg: great! I'll take a closer look and try this out later. digging a bit nto constraint refactoring at the moment.

15:49 dgrnbrg: dnolen: cool, good luck! I'm going to be thinking about what a planner would look like to integrate this w/ jim duey's fj branch

15:50 and fixing some other projects of mine up :)

15:50 dnolen: dgrnbrg: Will said some interesting things about Ciao's approach.

15:51 dgrnbrg: is there a record of that anywhere I could check out?

15:51 dnolen: dgrnbrg: in Ciao they use complexity annotations - perhaps that + the trace information is enough to infer a good plan?

15:51 dgrnbrg: I was thinking that i'd just add a simple "parallel-scoring" mechanism to the substitution monad

15:52 some simple heuristic like unification costs 1 and everything else is free might be sufficient

15:52 and fork after you hit cost 20, reseting the cost back to 0 post-fork

15:53 with a system for better annotations to further improve the results?

15:53 that seems easy to implement, and potentially very effective

15:53 wingy: is there a way to make this more DRY (if (get options :back-path) (get options :back-path) "/")

15:54 dnolen: dgrnbrg: yes I think a simple heuristic approach would be good just collect some basic results. and then perhaps see where annotations might help where that doesn't work.

15:54 dgrnbrg: I don't have a specific Ciao paper handy ... and I don't see anythign obvious on their site, they've published a lot

15:55 dgrnbrg: do you have any interesting and expensive core.logic programs (> 500ms) that I could look at? And that just use the basics?

15:55 dnolen: dgrnbrg: Will cites this one in his dissertation - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=

15:56 dgrnbrg: zebra is actually a pretty good one to try this one I think.

15:56 dgrnbrg: dnolen: that only takes 15ms on my machine :(

15:56 it'll only be meaningful to parallelize something that takes 500ms+

15:56 dnolen: dgrnbrg: it maybe tough to beat the optimal ordering, but reorder the goals, it can be as slow as 500ms.

15:57 dgrnbrg: on your machine I think it'll be slower than that.

15:57 ~550ms on a 4 core i7 if you have a bad ordering if I remember correctly.

15:57 clojurebot: swank.core/break is simple debugging breakpoints with swank-clojure: http://hugoduncan.org/post/2010/swank_clojure_gets_a_break_with_the_local_environment.xhtml

15:58 dnolen: dgrnbrg: I'm also keen to see if this work can be applied to the constraint stuff ... parrallel sudoku would be fun.

15:58 jamii: I'm trying (and failing) to use the new record reader syntax - https://gist.github.com/3845928

15:58 dgrnbrg: dnolen: I reversed the goals, and it's doing terrible

15:58 dnolen: dgrnbrg: great! ;)

15:59 dgrnbrg: actually, it's too terrible

15:59 it's been several seconds

15:59 jamii: I have clojure 1.4.0 in my project.clj. Is there any way to double-check that I'm using the right version from inside 'lein repl'?

15:59 dnolen: dgrnbrg: I think if you just put membero goals before firsto nexto righto, you should be good.

16:00 doomlord: does clojure do type infering in its compiler ? .. does the greater use of immutability makes this easier than other lisps

16:01 dnolen: doomlord: I think most of the type inference in the compiler is around interop - CLJS does some to avoid checking for Clojure's idea of truthy values instead of JS's

16:01 dgrnbrg: dnolen: it is staying fast :(

16:01 why couldn't you write worse code?

16:02 i think the unification in the front is too intelligent

16:03 if i put the bare unification after 4 memberos, i get 70ms/run, but if i put it after the 5th membero, it seems to take a very, very long time

16:03 possibly diverged

16:05 dnolen: dgrnbrg: maybe but the search is pretty huge if you don't do it right.

16:11 dgrnbrg: you're using run* right?

16:11 dgrnbrg: dnolen: i am trying run1

16:11 is that breaking it?

16:11 it still hasnt finished running

16:11 and somehow it's using 700% cpu

16:12 wtf…i think my machine's going to overheat (if i dissappear, you'll know why)

16:12 dnolen: dgrnbrg: haha, the perils of zebra

16:12 AimHere: 700% CPU? Have they invented virtual processor time to go with virtual memory?

16:13 dgrnbrg: it's a convention for multicore systems :) 700% cpu means it's using 7 cores fully

16:13 antares_: AimHere: 700% = 7 cores 100% utilized

16:13 AimHere: You guys know how to make technology seem far less miraculous than it should

16:15 ivan: oh how boring it is to not be really confused

16:15 dnolen: jamii: hmm I'm not sure why that doesn't work ... the record literal syntax

16:16 dgrnbrg: well, this ordering sucks

16:17 i'm going to try something else that's a bit faster

16:17 dnolen: dgrnbrg: http://gist.github.com/3845966, ~500ms for this order on my machine using run*

16:17 dgrnbrg: in this case we do want run*, we want proof there is only one answer

16:17 run 1 doesn't give us that.

16:20 dgrnbrg: dnolen: awesome--that's giving me around 500ms

16:20 jamii: dnolen: It works in files, perhaps the lein2 repl is just doing something wrong

16:21 dnolen: possibly breaking things into seperate forms by {} () [] which breaks on #user.Foo{}

16:21 dnolen: yep - https://gist.github.com/3845982

16:21 Will file a bug for that

16:22 dnolen: jamii: ah yeah I was trying with lein repl myself.

16:27 jamii: https://github.com/technomancy/leiningen/issues/806

16:33 dgrnbrg: dnolen: which part of town would you like monday? i work around soho

16:57 qerub: (read-string "`[1]") ; => (clojure.core/apply clojure.core/vector (clojure.core/seq (clojure.core/concat (clojure.core/list 1))))

16:57 Isn't that a really strange result?

17:01 devn: hello all

17:01 ivan: qerub: I think it has to do that in case you decide to unquote or splice-unquote something

17:02 there was a ClojureWest talk on macros

17:02 devn: qerub: which part of it do you think is strange

17:10 qerub: ivan: ok, thanks. I'll take a look at that talk.

17:10 devn: Partly that it returns vector+seq+concat+list instead of just vector, but mainly that the syntax-quote is handled as early as read-time.

17:10 devn: Compare with Scheme: (first (read (open-input-string "`[1]"))) => 'quasiquote

17:11 At first I found it strange that the symbols had been namespace-bound, but then I realized it does that to avoid unexpected variable captures in macros.

17:12 Is there a article on the design decisions of Clojure's macro system somewhere?

17:14 ivan: This one? http://www.infoq.com/presentations/Clojure-Macros

17:14 ivan: yep

17:14 though it might lack the design decisions you're looking for

17:15 qerub: Looks interesting anyway.

17:18 antoineB: something i never be sure, is macro done at compilation, i mean alway done at compilation?

17:18 devn: damn qerub left

17:18 antoineB: i expect a "no" answer

17:18 devn: er: "damn, qerub left."

17:18 qerub: devn: ?

17:19 devn: oh sorry, my autocomplete wasn't working for you for some reason

17:19 qerub: quasiquote in clojure has been discussed in the past on the list IIRC

17:19 chouser: antoineB: understanding that compilation can happen during runtime, the answer is "yes" -- macros are normally only expanded during compilation

17:19 devn: qerub: https://groups.google.com/forum/?fromgroups#!topic/clojure/bQxffjeQkFU

17:20 that might be of some value to you

17:21 qerub: devn: Thank you!

17:21 antoineB: chouser: when i say compilation, i mean "c" like compilation

17:21 devn: qerub: it's not a whole story, but maybe it will give you some more to work from

17:23 antoineB: chouser: so in fact view macro as function that don't evalute param and evaluate return is the good way

17:24 chouser: Hm, I think that'll trip you up at some point.

17:24 holo: hi

17:24 chouser: it's a function that's run at compile time and its return value is spliced into the code being compiled.

17:24 antoineB: so it takes param rearrange them, and "compile" the result

17:24 devn: qerub: just dug this up as well:

17:25 qerub: https://github.com/hiredman/syntax-quote

17:25 qerub: referenced from: http://blog.8thlight.com/colin-jones/2012/05/22/quoting-without-confusion.html

17:25 qerub: devn: Sweet! I'm trying to get a grasp of the different tradeoffs when designing macro systems and my focus is currently on Clojure's, all pointers appreciated.

17:25 devn: you've drank the koolaid? err i mean! read Let Over Lambda? ;)

17:26 s/drank/drunk

17:28 qerub: Macro koolaid…? Sure, I've had my share. I haven't actually read all of LOL yet. Thanks for the reminder to continue.

17:29 devn: heh, im just kidding. it's an interesting book, but the author just *loves* macros and takes every opportunity to tell you they're the most wonderful thing in the world

17:30 there's a bit more nuanced discussion of them in Clojure I think, see: http://blip.tv/clojure/michael-fogus-the-macronomicon-5970233 (which is a response/continuation of: http://blip.tv/clojure/christophe-grand-not-dsl-macros-4540700)

17:30 qerub: ^

17:30 qerub: Powerful abstractions are always cool. :)

17:30 holo: is this path correct for bash in heroku to find lein? "PATH=.lein/bin:/usr/local/bin:/usr/bin:/bin". my app is crashing because bash cannot find lein. don't remember doing anything stupid. probably i did it anyway

17:31 devn: holo: you're talking about on deploy?

17:31 holo: devn, yes "heroku logs" show it

17:32 devn: holo: ive actually never needed to tell heroku where lein is because it just knows, but ive been doing most of my work on EC2 as of late. what are you trying to do with lein?

17:33 qerub: devn: I had missed fogus' response. Thanks for the links.

17:33 devn: holo: also, what stack?

17:34 qerub: not a problem! since you're interested in the nitty gritty, doing some careful hunting on the mailing list is a great idea, but also, there's this!

17:34 http://www.google.com/cse?cx=partner-pub-1237864095616304%3Ae7qm3gycp2b&ie=UTF-8&q=quasiquote&sa=Go&siteurl=clojure-log.n01se.net%2F&ref=www.google.com%2Fsearch%3Fq%3Dclojure%2Birc%26oq%3Dclojure%2Birc%26sugexp%3Dchrome%2Cmod%3D7%26sourceid%3Dchrome%26ie%3DUTF-8&ss=1000j186994j10#gsc.tab=0&gsc.q=quasiquote&gsc.page=1

17:34 antoineB: is there an interest in using immutability in clojurescript?

17:34 holo: devn, cedar

17:34 devn: what does your project.clj look like?

17:35 qerub: yeesh, that url is hideous... http://bit.ly/QTrSiu

17:35 holo: devn, min-lein-version "2.0.0"

17:35 devn, want full file?

17:35 devn: holo: hm, that's the main question i had actually, but yeah, if you dont mind gisting it

17:35 qerub: devn: :)

17:35 devn: or refheaping it

17:35 holo: sure

17:37 devn, https://gist.github.com/3846200

17:41 devn, oh zeus.. i had PATH: PATH=... in my heroku config

17:41 devn: :)

17:41 holo: devn, fixing it sorted it out of course. i'm sorry to make you waste time

17:42 devn: i was going to say: PATH=./lein/bin:... should be sufficient

17:42 holo: it's not trouble. happy to help

17:42 s/not/no

17:42 holo: devn, i appreciate your help. thanks

17:42 devn: np

17:43 holo: have you seen https://github.com/heroku/heroku-buildpack-clojure ?

17:46 holo: it sounds like you got it figured out, but my experience is that with heroku there's inevitably some vooodoo required at some point, and since technomancy is at heroku and has been doing clojure for a long while, that repo is probably a good resource

17:46 antares_: devn: hey. If I remember correctly you were part of the documentation debate a few weeks ago, right?

17:47 * devn denies everything

17:47 antares_: but we have LOGS!!

17:47 devn: i barely remember what happened yesterday, so forgive me, but i'm drawing a blank. it's entirely possible i popped in for a second

17:47 antares_: sounds like you want to talk about something: hit me

17:48 antares_: devn: just wanted to say that we are done with plumbing for http://clojure-doc.org and now just need to migrating existing tutorials and other content, and write new one

17:48 dnolen: dgrnbrg: hey I'll be close to the Times.

17:48 * adiabatic wonders how to improve his 4clojure workflow

17:48 dgrnbrg: dnolen: i don't know where that is :/

17:49 devn: antares_: i think i have something that may be of interest to you

17:49 jamii: Woot - https://github.com/jamii/strucjure

17:50 And under 400 LOC

17:50 holo: devn, am in not i'm using that buildpack by default? at this point i cannot complain about heroku. it's really a great platform service. but the hard times are still to come

17:50 * devn golf claps

17:50 jamii: Admittedly one of those lines is 3k characters

17:50 dnolen: dgrnbrg: times square

17:51 antares_: devn: yeah? well, do tell

17:51 devn: antares_: now you lit a fire under me. are you going to be around for a few minutes?

17:51 antares_: devn: you have 20 in-depth guides about everything Clojure hidden in a private git repo?

17:51 dgrnbrg: dnolen: is that where the name comes from?

17:51 devn: antares_: actually, sort of...

17:51 * dgrnbrg not making connections

17:51 antares_: devn: sure

17:52 holo: devn, s/am in not i'm/am i not

17:55 devn: holo: i think the build pack stuff is new-ish, but im not sure, either way it cant hurt to know about that if you get into slightly more non-standard territory

17:57 antares_: erm, almost there, i keep blowing the heap for some reason? :X

17:58 oh, nevermind, found it, was testing query performance

18:00 holo: devn, that's cool. thanks

18:00 devn: antares_: https://bc7p.showoff.io/

18:01 this is an old idea that im redoing because i did it all wrong the first time, and it's still wrong at the moment, but anyway, type "let" into the input text box and search

18:01 antares_: you there? (this url is only good for 5min)

18:02 antares_: devn: looks like hoogle or hayoo? pretty cool

18:02 devn: yes, I am playing with it

18:02 devn: it needs work. it's using postgres right now, but i think it should be primarily a faceted solr search

18:03 antares_: devn: gee. You are old school :)

18:04 devn: seriously though, I think that faceted search can work but if you go with it, ES is much nicer to work with.

18:05 adiabatic: I'd like to be able to copy and paste the (= …) lines in 4clojure into something that either emacs+nREPL or 'lein test' can handle. I'm trying the latter way, and I'm doing (deftest a-test (map is [(= …) (= …)])) in core_test.clj. Thing is, when I run `lein test`, it errors out with a RuntimeException: Can't take value of a macro. Should I look into dosomething functions/macros/special forms to get what

18:05 I want?

18:05 devn: people keep telling me that, but one thing that i get for free with solr is ive already built a couple of solr-backed rails sites which use an already built search discovery tool that does a lot of work i dont want to do

18:06 like accreting constraints, bookmarks, etc.

18:06 zackmmms: Hey, are reader macros still off limits in clojure?

18:06 devn: zackmmms: yes

18:06 ordnungswidrig: adiabatic: "is" is a macro thus cannot be applied to a value with map

18:06 adiabatic: ordnungswidrig: I figured that's what the error message was saying

18:07 ordnungswidrig: however: (map #(is [(= ..) (=…)]) …) or like that should do

18:07 adiabatic: or you are looking for "are"

18:07 devn: antares_: anyway, i want to put this out there as sort of a card catalogue of simple examples, and provide an API so you could augment your examples or something

18:07 antares_: devn: you are scaring me. Rails? There is http://clojureelasticsearch.info ;)

18:07 devn: but it's pretty cool. I think it may be less useful in a dynamically typed language but all such experiments lead in the right direction.

18:08 aperiodic: ordnungswidrig: careful, you'll need to realize the sequence that the map produces at some point

18:08 devn: antares_: well, it gets more interesting, all of the stuff you're seeing there has been run through clojail. what im not currently doing is any kind of analysis, but i could be.

18:08 adiabatic: ordnungswidrig: I probably am!

18:09 * adiabatic tries it

18:09 ordnungswidrig: adiabatic: then dorun is more likely what you want

18:09 clojurebot: Roger.

18:09 zackmmms: dven: So even with an extensible reader, there is no way I could write `{parameters}` and have it be valid clojure code

18:10 antares_: devn: what kind of, for example?

18:10 devn: antares_: actually, datomic could be really interesting too, people have built the cons cell on top of datomic. there are some holes still though, like the fact that id want to use a different analyzer than the one datomic provides

18:12 antares_: like when i safely read the sexp and capture it's output and/or value, it creates a map, right? {:input "(+ 1 1)", :out nil, :value "2"}

18:12 but it could also flatten it and get all of the symbols used and pop those into :symbols #{+ and or}

18:12 antares_: interesting

18:14 devn: i mean it's not magic or anything, but since you're already reading the damn thing, you may as well use that opportunity to play around with it. hell, you could try to automatically rewrite the expression using kibit or clojure-refactoring to alternatively provide a thread-first and thread-last version of the same expression

18:16 antares_: anyway, i guess i'd like to know if you see any value in tying what im building together with what you're working on, sort of a value add for clojure-doc?

18:17 antares_: devn: yes and no. Let me explain.

18:17 devn: we want clojure-doc.org to be more than just a collection of tutorials and guides. And we want to cover more than just the language (stuff like how JVM ecosystem works or what Leiningen is is equally important).

18:17 but we won't cover the API reference part, for example

18:18 because there are already resources that do that reasonably well

18:18 clojuredocs.org will be adapted for 1.4 and ported to be 100% clojure over time

18:18 as for this thing, once it works well, I think we can dedicate an article to it and certainly link to it

18:19 maybe even give you a subdomain, why not

18:19 but clojure-doc.org is primarily a content site

18:19 devn: does that answer your question?

18:28 wingy: how do i get the request body/querystring in compojure?

18:30 aperiodic: wingy: https://github.com/weavejester/compojure/wiki/Destructuring-Syntax

18:31 dnolen: dgrnbrg: gotta run here, but I'm happy to meet somewhere halfway.

18:31 dgrnbrg: dnolen: sounds good-- sync over email?

18:32 dnolen: dgrnbrg: yes, talk soon.

18:33 wingy: aperiodic: thx

18:40 azkesz: hi guys, what is the order in which the bindings are executed inside the let form ? and is that guaranteed?

18:40 antares_: azkesz: anecdotal evidence suggests top-to-bottom and yes, I believe it is always the same

18:41 ,(let [a 1 b (inc a)] b)

18:41 Frozenlock: azkesz: 'let' in clojure is like 'let*' in lisp.

18:41 clojurebot: 2

18:41 azkesz: i don't know lisp though :)

18:41 Frozenlock: :P

18:41 antares_: Frozenlock: there is even let* in clojure.core

18:41 aperiodic: ,(let [a 1 _ (prn a) a 2 _ (prn a)] a)

18:41 clojurebot: 1

18:41 2

18:41 2

18:41 eggsby: azkesz: that just means 'let sequential', so each one happens after the prior

18:41 azkesz: ,(let [b (inc a) a 1] b)

18:41 clojurebot: #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: a in this context, compiling:(NO_SOURCE_PATH:0)>

18:42 azkesz: understood guys, thanks so much

18:42 Frozenlock: antares_: What does it do?

18:42 azkesz: after all the bindings are inside a vector right? []

18:42 antares_: Frozenlock: that's a helper for clojure.core/let

18:43 Frozenlock: my point is that there probably has been some inspiration from Common Lisp and such

18:43 azkesz: right but that may or may not be the reason

18:43 chouser: let is actually a macro. let* is the special operator. I think its like let without destructuring

18:43 antares_: azkesz: you just want that kind of guarantee for conveience

18:43 *convenience

18:46 azkesz: true but if it were a set... any ordering would be lost i figure, but even as vector the order could've been from last to first, I guess I only meant it's vector instead of set so the order at least is not lost if it's indeed ever to be used

18:47 well I noticed this code which seems to require that let keep the order else it would've been flawed

18:47 * Sgeo still has no idea what reset! does in delimc

18:48 Sgeo: Oh, reset! as in the thing for atoms

18:48 wingy: https://github.com/weavejester/valip seems not very active

18:48 https://github.com/michaelklishin/validateur is better choice?

18:48 need to validate form inputs

18:49 azkesz: judging by the author, probably :)

18:50 antares_: wingy: valip is definitely not bad. validateur in 1.x has some limitations we can't overcome in a compatible way but if you don't need i18n, it will do what you need.

18:51 wingy: also, validateur is pretty trivial to extend with custom validators. They are just Clojure functions. Not sure about valip.

18:51 wingy: antares_: ill look into both

18:52 antares_: Sgeo: reset! basically set the new value for an atom, completely ignoring the current one

18:52 Sgeo: antares_, I was getting confused because of the use of reset in delimc, for something totally different

18:54 antares_: Sgeo: delimc?

18:54 Sgeo: https://github.com/swannodette/delimc

18:54 Delimited continuations for Clojure

18:54 adiabatic: fiddlesticks. Looked up the interface for split-at to do 4clojure #49 and they gave away the answer.

18:55 antares_: Sgeo: sounds too hardcore for my brain

18:55 Sgeo: antares_, it is a little for me, but there's a clear connection between these and monads which I think will make monads easier to use

18:56 I think what I want to write will look something like:

18:57 (reset [(+ (bind-cont [1 2 3]) (bind-cont [4 5 6]))]

18:58 Which should be the same as (for [a [1 2 3] b [4 5 6]] (+ a b))

18:58 Well, same conceptually

18:58 Although different return types I guess

18:59 The concept is simple: A function that binds a monadic value to the continuation

18:59 azkesz: ,(for [a [1 2 3] b [4 5 6]] (+ a b))

18:59 clojurebot: (5 6 7 6 7 ...)

19:00 Sgeo: Hmm, maybe reset-return...

19:00 Would get rid of that []

19:01 * Sgeo pings dnolen

19:01 Sgeo: Basically, I want to put dnolen and jimduey in a particle collider and collide them.

19:03 * Sgeo will call it delimonads

19:06 wingy: http://clojurewerkz.org/articles/about.html

19:06 looks like a good project

19:07 antares_: wingy: thank you

19:08 wingy: antares_: who are you?

19:08 Michael or Alex?

19:08 adiabatic: Lambda in a gear-looking thingy. Though it was Half-Life related, at first…

19:08 antares_: wingy: Michael

19:09 wingy: antares_: are you using those projects in your company?

19:09 you got a lot of dbs covered

19:10 antares_: wingy: I use some, Alex uses some. We originally did not use Riak but later replaced mongodb with it.

19:10 besides Romulan, no clojurewerkz project is purely experimental

19:10 wingy: antares_: just curious .. how was it better than mongo?

19:10 antares_: that's why we have the docs

19:11 wingy: yeah good that you are focusing on the doc quality

19:11 antares_: wingy: it can handle very high write volumes with predictable latencies. As the load increases, you just add nodes.

19:11 wingy: seems like a missing feature in clj land

19:11 casion: antagon: how do you create your docs?

19:11 antares_: also, it is ridiculously easy to operate

19:11 casion: err, wrong tab complete

19:11 antares_: how do you create your docs

19:11 antares_: casion: github.com/clojurewerkz/docslate. Doc repos are under github.com/clojurewerkz

19:12 wingy: antares_: you are using neo4j as well? its a nice lib

19:12 db i mean

19:12 antares_: wingy: we do, although I believe there is now a better option in the graph DB space

19:12 wingy: antares_: which one?

19:12 antares_: but neo4j is the most mature, that's true

19:12 titan

19:12 https://github.com/thinkaurelius/titan

19:13 devn: antares_: speaking of subdomains, you sure you dont want a subdomain under getclojure.org? :)

19:14 antares_: devn: as soon as I am sure I get clojure :)

19:14 casion: antares_: you don't genreate stubs from the code?

19:14 antares_: devn: what is getclojure.org?

19:14 casion: generate*

19:14 devn: antares_: and yeah, that answers my question -- im just trying to think of how we can roll clojure-doc, clojuredocs, and something like im building all into one tight package

19:14 antares_: the future home of what i showed you earlier

19:14 antares_: casion: API references are generated, see http://clojurerabbitmq.info nav bar, for example.

19:14 wingy: antares_: i used to use neo4j before datomic .. why do you think titan is better than neo4j?

19:14 Sgeo: devn, what did you show antares_

19:14 Apage43: antares_: I looked at titan.. but it seems to be a kind of naive implementation.. I expect it will never be as fast as neo4j

19:15 antares_: Apage43: are you serious?

19:15 Apage43: well, have you used neo4j for real?

19:15 devn: Sgeo: https://m61j.showoff.io/

19:15 Apage43: i've used neither at scale

19:15 just played with them

19:15 devn: Sgeo: type "let" into input -- it's undergoing a rewrite right now

19:15 Apage43: so i am coming from nowhere

19:15 antares_: neo4j is nice but 1. It's not a speed daemon and because it is transactional, it probably never will be. 2. It has no backups or HA in the OSS edition. The latter is why I will probably never used it again.

19:16 titan with cassandra and hbase has replication and distribution built-in

19:16 casion: this is apparently over my head

19:16 wingy: yeah https://groups.google.com/forum/#!msg/aureliusgraphs/vkQkzjN8fo0/Sz_pLjWr7tcJ

19:16 antares_: Apage43: and Titan developers test it on data sets like full orkut dump

19:16 billions of relationships

19:16 Apage43: It's just the "distribution" part where I feel like it gets to be a hard problem.

19:16 antares_: it certainly has reasonable performance even with huge volumes

19:17 Apage43: they outsource the problem to hbase and cassandra

19:17 which, in my opinion, solved it pretty well

19:17 this is the most interesting part, you get transactions thanks to titan even on top of eventually consistent data stores

19:17 that is pretty remarkable and gives you a lot of good from both sides

19:18 Apage43: neo4j distribution is pretty complex, especially the id generation and syncing part with concurrent transactions :)

19:18 Apage43: so, it's fair to be skeptical about titan because it is young, but neo4j is far from perfect, especially on the ops side

19:19 Apage43: ah, well, I figured neo4j solved distribution by not having it =P

19:19 antares_: Apage43: the library :) there is REST server on top of it now :)

19:19 Apage43: titan can use berkeley DB Java edition to work basically like neo4j embedded

19:19 Apage43: also, keep in mind that neo4j is GPL and titan is apache 2.0

19:19 oceans of difference

19:19 wingy: i thought about neo4j or datomic but settled with datomic since it is closer to clojure and also it has time travel built in :)

19:20 antares_: do you have thoughts on datomic?

19:20 antares_: wingy: haven't used it. I think its value proposition is significant but I don't have any short term need in it. And they definitely don't need a Clojure client from us :)

19:20 Apage43: Oh yeah, the only thing I'd really like neo4j better than other things for is raw perf, which it can only do because it can't scale out

19:21 so if you really do have billions of things it's probably hard to use it anyway.

19:21 wingy: Apage43: my friend said that sicne Neo4j can't scale out to multiple machines he wont use it

19:21 antares_: wingy: if you like clojurewerkz and the docs, feel free to spread the word on twitter or clojure ML. Not to stroke our egos, to raise the awareness of now much a difference solid documentation makes.

19:22 wingy: what happens when your machine's capacity runs out with Neo4j? then you are screwed?

19:22 lpvb: Why is ClojureCLR not gaining any interest?

19:22 Apage43: wingy: right. Titan does that. and it does the silliest thing that will work, which is it just shoves the graph nodes in a distributed k-v store like cassandra or HBase

19:22 antares_: wingy: with Neo4J Server, you are not. But it's not about capacity, it's the backups and HA parts in the Community Edition that are problematic.

19:22 Urthwhyte: F# and entrenched interests probably?

19:22 casion: antares_: is there any intent on documenting how you document things?

19:22 wingy: antares_: unfortunately im not very popular in Twitter :) .. I will be one day though I hope :)

19:22 antares_: casion: hm, there honestly isn't much to it. We can write a blog post.

19:22 wingy: but i usually spread words in irc channels and google groups

19:23 devn: lpvb: lots of reasons, some historical, some related to number of people actively working on it, F#, and probably more

19:23 antares_: casion: the toolchain is very straightforward, it uses Jekyll and gists (or Pygments, in which case you will need to install it), everything is in a git repo and written in Markdown.

19:23 casion: antares_: straightforward if I knew what any of those things were :|

19:23 devn: casion: google :)

19:23 wingy: Apage43: like Datomic .. it is using other dbs as storage

19:23 antares_: wingy: you can instantly become popular on the clojure ML :)

19:24 wingy: pretty clever .. like dbs are using fs

19:24 holo: Raynes, your conch version number doc disagrees between github and clojars

19:24 casion: devn: if I have to research it, I would say it's not exactly straightfoward :)

19:24 wingy: they are one level on top of those dbs

19:24 antares_: wingy: just post something saying that docs make a huge difference and it should be a top short term priority for Clojure/core and such.

19:24 casion: not necessarily complicated, but not straightfoward

19:24 antares_: wingy: of course, if you don't want to, no problem with that :)

19:24 devn: casion: seriously, if you googled them you could grok the whole thing by definition

19:25 antares_: holo: submit a PR? Raynes is pretty ace at merging other people's work

19:25 Apage43: wingy: yeah, datomic can do a lot more clever stuff around caching though

19:25 wingy: antares_: whats clojure ML?

19:25 devn: markdown: turns plaintext to HTML, jekyll: static site generator, gists: gist.github.com (a pastebin), pygments: syntax highlighting

19:25 antares_: casion: ok. I think we will write a blog post next weekend. I have something pretty cool to work on and announce next week.

19:25 wingy: Clojure mailing list, the google group

19:26 wingy: https://groups.google.com/group/clojure

19:26 holo: antares_ it's just a small doc mistake, is it really necessary? ._.'

19:26 devn: antares_: so are you going to build a static site generator for clojure-doc?

19:26 casion: devn: that was far more useful than the last 15 minutes of googling I've done

19:27 devn: casion: instead of google i should have maybe said github

19:27 antares_: holo: of course, people will be using an outdated version otherwise

19:27 devn: quicker to skim the readmes of those projects

19:27 antares_: devn: we have it already

19:27 casion: devn: it's pretty well to assume I've lived in a cave for the last 15 years

19:27 devn: antares_: where's the repo?

19:27 antares_: devn: we use jekyll with a template repo all clojurewerkz.org projects use

19:27 devn: antares_: oh, right, but you said earlier it will be 100% clojure? no?

19:27 antares_: devn: https://github.com/clojuredocs/cds

19:27 devn: clojuredocs.org will be

19:28 the API reference site

19:28 not CDS

19:28 devn: not clojurewerkz or clojure-doc?

19:28 antares_: the domain is a bit confusing given there's already a clojuredocs

19:28 antares_: devn: clojure-doc (CDS, clojure doc site). clojurewerkz.org is not affiliated with CDS>

19:28 devn: yeah but we don't have control over clojuredocs.ogr

19:28 and there is no way to quickly split it into two parts

19:28 devn: is zkim MIA?

19:29 wingy: antares_: again, really love your initiative with the umbrella project .. when i get money ill poor somewhat in clj community making libs great

19:29 antares_: devn: we could not reach him so far. I hope someone in SF will help me with that soon.

19:29 wingy: pour

19:29 antares_: devn: the main reason is that we simply don't want to wait and potentially destabilize clojuredocs.org

19:29 but we need to publish our progress somewhere

19:30 we are now at a point where all there is left to do is to write, edit, publish, rinse, repeat

19:30 wingy: good doc .. good defacto libs .. simple and easy to create more projects

19:30 devn: weird, zkim's last commit is a month ago and his twitter moved from napple to _zkim, which doesnt exist

19:30 antares_: with a few people involved. We will also have guidelines in place for other folks who want to contribute.

19:30 Like a list of low-hanging fruits, document structure, etc

19:31 devn: antares_: im already on board, buddy

19:31 just give me a list of crap to do

19:31 antares_: devn: as far as I know zkim does not currently use Clojure at his company. But I am sure we will reach him somehow.

19:32 devn: we have a list of initial guides that need to be written. I can show it to you right now.

19:32 devn: hit me

19:33 wingy: antares_: you could almost start a company around those projects

19:33 antares_: wingy: probably not :) at least, no short term interest

19:34 wingy: im sure there are demands on high quality libs/support

19:34 no?

19:34 clojurebot: no is tufflax: there was a question somewhere in there, the answer

19:34 lpvb: I'm getting tired of vim not having great language integration. Does emacs have clojure autocompletion and jump to declarations?

19:34 technomancy: not enough to support a company

19:34 wingy: are you sure? libs are really valuable to other companies

19:35 in that case the clj market is not very high

19:35 technomancy: correct

19:35 devn: man, the CA discussion is happening again on the clojure ML

19:35 * devn sighs a long sigh

19:35 wingy: technomancy: its that bad huh

19:36 i wonder why though .. clj is way cool

19:36 casion: lpvb: yes

19:36 wingy: perhaps datomic can boost it up and make marketing for it

19:36 devn: i work for a consulting company and we have people who are asking us for clojure specifically

19:36 antares_: wingy: there are enough companies using Clojure but not enough yet to build a complete company around just libraries :)

19:36 wingy: are you sure?

19:36 i mean libs are everything :)

19:36 antares_: wingy: the same is true for even pretty popular languages. So companies like Typesafe do all kinds of things.

19:37 wingy: nope. Working products are everything.

19:37 devn: im curious how well chas has done on clojure atlas

19:37 antares_: but it takes more than libs to get there

19:37 devn: clojure atlas should probably get some love somewhere

19:37 i know it's for pay, but it's really useful

19:37 antares_: devn: are you getting my PMs?

19:37 ivan: it does not cost $200 to send a letter from Russia

19:37 devn: antares_: yeah, sorry, just saw them

19:37 antares_: ivan: yes it does

19:37 ivan: via fedex

19:38 and if you use Russian postal services, we all know how reliable it is

19:38 ivan: normal postal mail is reliable enough in my experience, and you can always send two

19:39 maybe your local post office really hates you?

19:39 casion: my experience with most russian postal has been very good

19:39 wingy: "you want your devs to build products or libs"

19:39 antares_: ivan: yeah it just takes 4 months

19:40 ivan: so if I want to contribute a small doc improvement, I will have to wait for 4 months. that's ridiculous.

19:41 wingy: you are partially right, there is value in good libs. But to have a successful company behind a language you will have to do way more than just develop libraries. Which is fine, I just don't think it's the goal for clojurewerkz right now.

19:41 wingy: yepp .. just saw something really valuable in your project

19:42 quality :)

19:42 i personally would love to pay you for maintaining stuff

19:42 if i want that .. a lot of people want .. i can't be alone

19:42 thats like the golden rule .. one person wants something means big market

19:43 or lets rephrase .. one person wants to pay for your thing means big market

19:43 antares_: wingy: maybe we will set up a donations site but I believe it will be better if both myself and Alex keep working on clojurewerkz.org as part of working on something else :)

19:43 and gain more contributors (Neocons, Monger and Welle already have several)

19:43 wingy: yepp .. was just sharing my thoughts :)

19:44 akhudek: is there a clojurescript + compojure lein new template somewhere?

19:46 wingy: akhudek: start setting up compojure first .. then clojurescript with lein-cljsbuild .. its dead simple

19:46 you dont need no template

19:46 akhudek: I was just being lazy, I already have projects using the combo. ;-)

19:46 wingy: ah

20:14 adiabatic: I can use the value of (type "blah") as a map key, right?

20:14 danlarkin: why not try it

20:15 ,{(type "blah") 42}

20:15 clojurebot: {java.lang.String 42}

20:15 adiabatic: good, that means I'm not going off in the wrong direction

20:15 thanks!

20:16 ,(= (type "blah") (type "blaaaaad"))

20:16 clojurebot: true

20:16 adiabatic: ,(== (type "blah") (type "blaaaaad"))

20:16 clojurebot: #<ClassCastException java.lang.ClassCastException: java.lang.Class cannot be cast to java.lang.Number>

20:22 Sgeo: I am contemplating a function that I want to call bind-cont

20:22 But, considering that it would likely get used frequently, is bc a good name?

20:22 or BC?

20:23 adiabatic: You can always make a shorter alias later, right?

20:29 zoldar: antares_: I will do the pull request but one question, because I'm a bit of a noob when it comes to git workflow - how to get my fork's master in sync with original repo? :)

20:29 antares_: zoldar: if you don't have a fork?

20:30 zoldar: in that case, just fork it. If you already have a fork, it should be as simple as pulling from Neocons master and pushing to your repo.

20:30 zoldar: antares_: no no, I already have a fork, I just wonder if this is something done from gui, or is it just explicit pull on git level

20:30 okat

20:30 okay*

20:30 antares_: zoldar: git pull git://github.com/michaelklishin/neocons.git master should get you there

20:31 Sgeo: $mail dnolen I need to be able to write an fn that only makes sense in the context of a reset, but is not itself in a reset

20:31 lazybot: Message saved.

20:31 Sgeo: Hmm, then again, maybe (reset (defn blah)) would work sufficeintly well

20:41 holo: did anyone notice that when running scripts using "foreman run <command> <args>", if <args> contains a single flag -x/--xxx, the number of arguments printed by echo $# will be 0?

20:43 wish is really bad..

20:43 *which

20:47 technomancy: holo: for doing clojure dev on heroku, you really only need foreman to check the syntax of your procfile

20:47 it's really not necessary for running your app in dev

20:48 foreman is not that useful with leiningen since lein has good alias support

20:52 holo: technomancy, i need foreman to read .env

20:53 Sgeo: Ok, this seems way way way way way too easy

20:54 And it doesn't work

20:56 It may be delimc brokenness

20:57 As in a "doesn't know how to use Clojure" way, not a "Some weird concept has gone wonky" way

20:58 https://github.com/swannodette/delimc/blob/master/src/delimc/core.clj#L131

20:58 Can someone explain to me the purpose of that ~'make-funcallable?

20:58 Would make-funcallable not work there for some reason?

20:59 Could ~'make-funcallable cause issues when it comes to namespacing?

20:59 holo: technomancy, what do you mean with "good alias support"?

20:59 clojurebot: technomancy: you're right I should explain the reason. The reason is I would like to be able to call aether with an extra parameter for introducing my own :transfer-listener.

21:00 holo: hun?

21:00 clojurebot: i think chunking is to blame

21:00 Sgeo: Yep, that construct is used elsewhere and causes issues

21:00 I think I understand why it would be used though

21:01 holo: clojurebot is not really a bot?

21:01 zoldar: antares_: done, hope I haven't screwed anything up

21:01 aperiodic: oh man, clojurebot just passed the turing test!

21:01 ~botsnack

21:02 clojurebot: botsnack is scoobysnack

21:02 aperiodic: clojurebot: forget botsnack is scoobysnack

21:02 clojurebot: 'Sea, mhuise.

21:02 Sgeo: ~botsnack

21:02 clojurebot: Thanks! Can I have chocolate next time

21:03 antares_: zoldar: pretty sure you haven't :) testing it locally…

21:04 doctordaebak_: dropping by to let everyone know about a hardware project that some in the clojure community might be quite interested in

21:05 kickstarter to create a 16-core computer for $99

21:05 http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone

21:06 hoping those here who want to see it funded, will also pass the word on to any who are interested in multithreaded and parallel applications

21:07 antares_: zoldar: merged. One minor thing: please move the test to relationships_test. I will release a snapshot either way, of course.

21:08 zoldar: antares_: at first I intended to do so, and even started adding a test for nodes but then I've spotted the test in cypher_test and followed the pattern

21:08 antares_: zoldar: the idea is that we will move more and more stuff to use cypher internally but tests are organized by features, regardless of the implementation

21:08 Sgeo: I want to slap dnolen

21:08 antares_: zoldar: yeah, that's for direct use of Cypher

21:10 zoldar: antares_: ok, I will keep it in my mind

21:12 Sgeo: While developing with the aid of the REPL, how much of a risk is there of removing a function but still accidentally relying on it, and not noticing because the function is still defined in the REPL?

21:12 antares_: zoldar: thank you for contributing! The most exciting part of working on OSS is seeing other people find your projects useful enough to get involved :)

21:12 Sgeo: it is more common to hit an old protocol implementation. It is not very common, though.

21:13 but you can hit various issues with hot code reloading, it is a very hard problem. I don't think anybody but Erlang solved it even distantly well.

21:18 technomancy: holo: just run export $(cat .env) first

21:18 holo: or use environ

21:19 Sgeo: Just verifying: I'm allowed to say that dnolen made a mistake, right?

21:19 technomancy: it's not required to put things in .env; it's just a convention

21:20 not a particularly good one on the JVM given that the environment is immutable without restarting

21:22 Sgeo: it isn't that uncommon to accidentally depend upon a function that isn't there anymore unfortunately

21:23 Sgeo: clojure.lang.Compiler$CompilerException: java.lang.RuntimeException: Can't specify more than 20 params, compiling:(c:\Users\Sgeo\DEV\Clojure\delimonads\src\delimonads\core.clj:5)

21:23 technomancy: holo: by good alias support I mean the fact that you can add an alias for a complete command, like :aliases {"go" ["run" "-m" "my.launcher" "--fast"]}

21:29 antares_: Sgeo: JVM methods cannot take more than 20 params

21:30 Sgeo: I have no idea how I'm giving anything more than 20 params

21:32 zoldar: &((def n [:a :b :c]) (def r [22 33]) (butlast (interleave n (conj r nil))))

21:32 lazybot: java.lang.SecurityException: You tripped the alarm! def is bad!

21:33 zoldar: &(let [n [:a :b :c]) r [22 33]] (butlast (interleave n (conj r nil))))

21:33 lazybot: java.lang.RuntimeException: Unmatched delimiter: )

21:34 Sgeo: antares_, I don't see what in the code could cause that :/

21:34 antares_: Sgeo: is the source up somewhere?

21:34 zoldar: &(let [n [:a :b :c] r [22 33]] (butlast (interleave n (conj r nil))))

21:35 lazybot: ⇒ (:a 22 :b 33 :c)

21:35 zoldar: right... is there some more elegant/succint way to do that?

21:35 Sgeo: https://gist.github.com/3846768

21:36 Oh hey my docstring's wrong

21:37 gfredericks: ,(doc interleave)

21:37 clojurebot: "([c1 c2] [c1 c2 & colls]); Returns a lazy seq of the first item in each coll, then the second etc."

21:38 Sgeo: At any rate, I know how to fix delimc's problem with namespacing

21:39 gfredericks: &((fn f [xs ys] (when (seq xs) (lazy-seq (cons (first xs) (f ys (rest xs)))))) [:a :b :c] [22 33])

21:39 lazybot: ⇒ (:a 22 :b 33 :c)

21:40 Sgeo: Tell me if I'm wrong, but if you :use a namespace, can you still use the vars within that namespace in a fully qualified way?

21:40 gfredericks: I think you can use anything fully qualified, and are responsible for it being loaded

21:41 so if you :use it then you've satisfied the being-loaded part

21:41 Sgeo: Thanks

21:41 So I just need to make every special form-ish thing defined by delimc be fully-qualified

21:42 I should probably do that for everything not just delimc.core

21:43 How do I restart or at least close the REPL in nREPL.el?

21:44 zoldar: gfredericks: doesn't (seq ...) actually realize its' argument (provided it's a lazyseq) - that would in part defeat the purpose of using lazy-seq.. ?

21:46 gfredericks: zoldar: just the first element

21:46 zoldar: I don't believe it defeats the purpose of lazy-seq

21:51 zoldar: Sgeo: Sgeo there's nrepl-restart and nrepl-quit in upcoming 0.1.5 version, but I suppose you would have to fetch it directly from github

21:51 gfredericks: I suppose that I'm mixing something up, nevermind...

21:51 gfredericks: it's tricksy

21:55 holo: technomancy, nice. can lein also parse env files in format ENV=value?

21:56 Sgeo: The more I think about it, the more I prefer how Common Lisp handles symbols over Clojure

21:56 At least, to the extent I understand the different

21:56 difference

21:57 gfredericks: how does common lisp handle symbols?

21:59 rbarraud: Very nicely ;-)

21:59 PCL FTW

22:01 Sgeo: As far as I understand, the namespace is always part of the symbol, when a symbol is read in by the reader, the namespace is known

22:02 gfredericks: does that mean that the reader expands 'conj to 'clojure.core/conj or that the programmer types in 'clojure.core/conj?

22:03 Sgeo: The former

22:03 ,*ns*

22:03 clojurebot: #<Namespace sandbox>

22:03 Sgeo: ,(= 'sandbox/foo 'foo)

22:03 clojurebot: false

22:03 Sgeo: In Common Lisp, the equivalent would be true, iirc`

22:06 * (intern "FOOBAR")

22:06 FOOBAR


22:06 * (eq 'cl-user::foobar 'foobar)

22:06 T

22:07 gfredericks: so that makes the qualification behavior in ` unnecessary?

22:07 Sgeo: If I understand everything correctly, yes

22:08 gfredericks: are there other advantages?

22:08 holo: technomancy, forget my question. just noticed the envs are already in a format compatible with bash. thanks for the tips. i'm going to move this to lein

22:09 Sgeo: gfredericks, for starters, dnolen's library would be working

22:09 >.>

22:09 gfredericks: which lib?

22:09 Sgeo: delimc

22:18 rbarraud: 20params sounds pathological to /me

22:19 casion: &(let [n [:a :b :c] r [22 33]] (replace r (take 5 (interleave n (range 3)))))

22:19 lazybot: ⇒ (:a 22 :b 33 :c)

22:19 Sgeo: rbarraud, well, I for one am not doing it deliberately

22:20 casion: if you throw counts for n and n+r, it would work with arbitrary sized seqs

22:20 Sgeo: Can Clojure fns accept more than 20 parameters?

22:20 amalloy: of course

22:20 casion: and it's faster than the aforementioned solutions

22:20 Sgeo: Ok, good

22:21 Because apply means that people seem to write fns that take arbitrary amount of parameters (e.g. concat) and with apply it actually becomes reasonable to feed in a lot of parameters for various possibly useful reasons

22:21 octagon: casion: what does the ampersand do there in front of the expression?

22:21 Sgeo: octagon, tells lazybot to do it

22:21 and , is for clojurebot

22:21 octagon: Sgeo: oh ok

22:22 gfredericks: and $ is for lazybot. And ~ is for clojurebot. And ## is for lazybot. And ...

22:22 casion: zoldar: ping if you didn't see the code I posted above

22:23 zoldar: casion: seen it, thanks

22:24 octagon: it's interesting that ((fn (vector 'x) (inc x)) 1) gives an error

22:24 gfredericks: fn is not a function

22:24 octagon: seems like that should not happen, but i can see why it fails

22:24 yes

22:24 but still

22:25 casion: &((fn [x] (vector 'x) (inc x)) 1)

22:25 lazybot: ⇒ 2

22:25 gfredericks: in that case you'd have to quote the x in a vector as well

22:25 casion: seems to work fine

22:25 gfredericks: casion: he means using (vector 'x) instead of [x] as the arg list

22:25 casion: ohh

22:26 Sgeo: octagon, you might like Tcl

22:26 octagon: the point is that you can't feed in a list of forms

22:27 you need to use the reader syntax for vector

22:27 i don't see fn* exposed in my repl

22:27 gfredericks: fn* is a special form

22:27 ,((fn* [x] x) :hehe)

22:27 clojurebot: :hehe

22:28 gfredericks: you can use it, you just can't reference it

22:28 octagon: ,((fn (vector 'x) x) 1)

22:28 clojurebot: #<ExecutionException java.util.concurrent.ExecutionException: java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.lang.Symbol>

22:28 octagon: ,((fn* (vector 'x) x) 1)

22:28 clojurebot: #<CompilerException java.lang.ClassCastException: clojure.lang.Symbol cannot be cast to clojure.lang.IPersistentVector, compiling:(NO_SOURCE_PATH:0)>

22:29 octagon: one of the cool things about lisps is that you can normally convert a list of things into an expression that can be evaluated

22:29 i don't see a way to do that here

22:30 gfredericks: macros and eval

22:30 octagon: unsatisfying

22:30 lancepantz: anyone know how to get an nio.ByteBuffer from a BufferedInputStream?

22:30 octagon: you'd have to use the macro to give you the reader syntax [x]

22:31 gfredericks: ,(eval (list 'fn (vector 'x) (list '+ 15 'x)))

22:31 clojurebot: #<Exception java.lang.Exception: SANBOX DENIED>

22:31 gfredericks: octagon: I'm not sure what you mean by that

22:31 Sgeo: octagon, check out Tcl, it works exactly how you want in this regard

22:32 (but probably not in many other regards, in particular, not very functional, no quasiquoting, stringiness, and mutable things require names)

22:32 octagon: Sgeo: tcl really? i wasn't that into it in 1999

22:32 Sgeo: But the point is that there are no special forms in Tcl.

22:33 If something's expecting a block of code, or of some syntax, you can dynamically create it

22:34 octagon: Sgeo: well you don't need to resort to tcl for that

22:34 Sgeo: most lisps work that way

22:34 Sgeo: erm, not really? At least, not in a way that doing it is generally supported/accepted/idiomatic in the community

22:35 octagon: why is the argument list a vector and not a list, like in scheme, for instance

22:35 gfredericks: clojure uses vectors in a few different syntactic places because it gives a bit more visual variety to the code

22:35 easier to read

22:36 octagon: gfredericks: that's not a reason

22:36 gfredericks: oh darn

22:36 Sgeo: Common Lisp, Clojure, and Scheme have macros, which accept literal code, but no clean way to generate code to pass in to a macro, without using the (unidiomatic in those Lisps but idiomatic in Tcl) eval

22:36 gfredericks: I'll have to work on that.

22:36 octagon: that's something you add to your emacs

22:36 there must be a technical reason

22:37 * Sgeo belatedly sees that octagon is not actually interested in what I thought he was interested in

22:37 casion: I've seen it explained elsewhere that it's the reason gfredericks stated

22:38 for that matter, it was an IRC log with rhickey speaking

22:38 gfredericks: octagon: I appreciate that the code has decent readability without having to lean on a particular text editor

22:39 octagon: well i'm surprised, then. i kind of thought that kind of syntax semicolon vs newline stuff was debunked

22:39 casion: debunked?

22:39 scottj: octagon: another reason that Rich has mentioned is that he doesn't like that lists are overloaded and used for both function calls and grouping. He prefers those be separate.

22:40 octagon: scottj: how so?

22:40 Sgeo: Well, Lisps tend to need correct newlines and indentation to be readable :/

22:40 (let [a 1 b 2] ...) the [a 1 b 2] is not a function call or a macro usage

22:41 ,(doc vec)

22:41 octagon: Sgeo: okay, now that is an actual reason

22:41 Sgeo: ,(doc vector)

22:41 clojurebot: "([coll]); Creates a new vector containing the contents of coll."

22:41 "([] [a] [a b] [a b c] [a b c d] ...); Creates a new vector containing the args."

22:41 gfredericks: list forms tend to accentuate the first element

22:41 octagon: Sgeo: so without the vector you'd need nested lists

22:42 as in scheme

22:42 gfredericks: I think all syntactic vectors in clojure could just as well be lists

22:42 Sgeo: Not a matter of nested lists, just a matter of stylistically preferring vectors over lists in circumstances where the first element isn't especially important

22:42 octagon: gfredericks: ugh, that's silly. in racket you can use [] in place of () if you like how that looks better

22:43 scottj: octagon: rich talks about this in his simple made easy talks btw

22:43 octagon: gfredericks: racket treats all [] as parens for that reason, so you can make it more "readable"

22:45 gfredericks: that makes the code across the community less cohesive

22:45 octagon: Sgeo: i can see the usefulness of using a vector instead of a list for the reason you mentioned; that it isn't an expression that will be evaluated

22:46 Sgeo: but being able to express a program as a list of things is so fundamental and useful

22:46 Sgeo: I should note that as far as I know it's a stylistic thing. Relying on it in a code-walker is probably a bad idea

22:47 gfredericks: octagon: what is an example of something concrete that you can't easily do in clojure because of this?

22:47 Sgeo: How is (let [a 1 b 2] (+ a b)) not a list of a symbol, vector, and list?

22:47 Hence, a list of things?

22:47 octagon: gfredericks: well i have this thing, it's data that is listlike

22:48 gfredericks: and i wrote some stuff in clj that takes these lists and defines the underlying symbols that appear in this data

22:48 gfredericks: the idea is that you evaluate the data in this special environment, and it produces a result

22:49 gfredericks: this works because any data that is listlike can be evaled, and as long as the symbols that appear in the list data exist in the environemtn, it can be evaluated

22:50 that seems like a pretty awesome property

22:50 but the fact that you must introduce vectors to the list structure is unfortunate for my application

22:50 Sgeo: I assume you're macroexpanding-all?

22:50 n/m, I don't get it

22:51 casion: I thought he asked what your application was were this was an issue?

22:51 octagon: well the project is here:https://github.com/micha/hlisp-starter

22:51 it reads the DOM body of the html page and evaluates it as a cljs program

22:52 gfredericks: yeah I'm having a hard time figuring out what the blocker is. You can (eval (list 'let (vec (apply concat name-value-map)) my-code))

22:52 that doesn't apply to cljs though

22:53 octagon: what i can't do is <defn><vector><x/></vector><x/></defn>

22:53 rather <defn><vector><quote><x/></quote><x/></defn>

22:54 gfredericks: oh I see; so you're working (for some reason) in a domain where you only have one type of sequence

22:54 octagon: no, i could introduce syntax for vectors, of course

22:54 which i will do

22:55 but i was just curious about why this non-uniformity was adopted, and i got some good references, so thanks for that

22:56 Sgeo: I still don't like the existence of namespace-unqualified symbols

22:57 (require [some-ns :as some-alias]) (= 'some-ns/foo 'some-alias/foo) should be true imo

22:57 * Sgeo suddenly sees a difficulty

22:58 Sgeo: In the Common Lisp version, I had to intern the symbol. Now, I think I see why

22:58 Hmm

22:58 ,(= #'ns #'ns)

22:58 clojurebot: true

22:58 Sgeo: Hmm!

22:58 gfredericks: ?

22:59 Sgeo: I know how to fix delimc.

22:59 Resolve the symbols, should resolve to a var, right?

22:59 ,(resolve 'ns)

22:59 clojurebot: #'clojure.core/ns

22:59 gfredericks: yep

22:59 Sgeo: These vars are in fact exactly what I want when I speak of Common Lisp symbols.

23:00 gfredericks: that sounds like an important realization :)

23:00 Sgeo: This needs to be mentioned somewhere

23:01 gfredericks: there was a talk at clojure/west entitled "namespaces, symbols, vars, oh my!" I think

23:01 surely a video o the interwebs by now

23:01 Sgeo: I hate videos. Any transcripts?

23:01 gfredericks: ha; no idea

23:01 but you probably understand most of it b ynow

23:11 I say that but just remembered I haven't actually seen the talk myself

23:13 Sgeo: gfredericks, watching now

23:14 gfredericks: Sgeo: let me know how it is :)

23:17 Sgeo: gfredericks, the slides are somewhat unreadable, as in, not high enough resolution

23:29 gfredericks: drat.

Logging service provided by n01se.net