#clojure log - Apr 22 2010

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

0:16 rfg: What's the easiest way to get an infinite sequence of [0 1 2 3 0 1 2 3 ...]?

0:18 hiredman: ,(doc cycle)

0:18 clojurebot: "([coll]); Returns a lazy (infinite!) sequence of repetitions of the items in coll."

0:18 rfg: Danke

1:24 johnmn3: hi

1:24 herdrick: you know what'd be nice? to make memoized functions implement IPersistentMap

1:24 that'd be super handy

1:24 and would make symmetric the 'maps are functions' property

1:26 Chousuke: I think it might actually be confusing

1:26 herdrick: how?

1:26 clojurebot: with style and grace

1:26 herdrick: words of wisdom, clojurebot

1:27 Chousuke: "maps are functions" makes sense and is useful, but "functions are maps (only if memoized)" just isn't :/

1:28 hoeck: herdrick: regardless wether its useful or not, with deftype, its only a few lines away :)

1:29 Chousuke: though I suppose it would facilitate a "lazy map" kind of thing

1:29 herdrick: it makes plenty of sense, and it's something i want right now

1:29 hoeck: really?

1:30 hmmm - i'll check out deftype

1:30 thanks

1:30 !

1:30 hoeck: herdrick: yes, wait a little, I'm pasting sth in a few minutes

1:30 Chousuke: note that deftype is 1.2

1:33 herdrick: Chousuke - thanks

1:33 ouch

1:33 i don't have 1.2 and my last clojure upgrade hurt

1:34 Chousuke: you can have two versions installed

1:36 johnmn3: was reading the irc logs, seeing rhickey's comments about creating some kind of mechanism in clojure that can mimic some of OOP's modeling... could someone give an example of how that would look in code, using 'bank accounts' for example?

1:37 like... would the clojure example require dereferencing of the fields or would that be taken care of by the object?

1:42 if dereferencing was taken care of internally, would the clojure example look essentially like the java example (hiding the "where to put the value problem")? And what would be the limitations of the clojure version compared to the java one?

1:48 hoeck: herdrick: http://gist.github.com/374865

2:00 herdrick: hoeck: thanks!

2:01 can't try it until i get the newest clojure

2:01 :(

2:01 have been avoiding that

2:03 hoeck: herdrick: you can download clojure.jar from build.clojure.org, and just try it on the repl, without upgrading you other clojure environment

2:03 slyphon: so, is it possible to hand a function a var's name (like *foo*) and then have that function deref the var? kind of like what a macro would do in terms of delaying evaluation?

2:03 herdrick: hoeck: ok.

2:03 or make a new swank-clojure-project using that jar, eh?

2:03 that'd be a snap

2:04 huh

2:04 hoeck: if it works for you, of course :)

2:05 herdrick: yeah

2:27 wooby: hoeck, that's sweet

2:33 hoeck: wooby: deftype is really nice, you can combine all the clojure interfaces to create new datastructures

2:34 wooby: hoeck, i've read a little on it but didn't realize the implications

2:35 hoeck: for example, if you implement seq and cons on your new datatype, you can map over and conj onto it as if it were a "native" clojure datatype

2:37 you could do this all the time, maps and seqs were defined by interfaces all the time, but deftype makes it really succinct and efficient to implement them, in contrast to genclass or proxy

2:39 wooby: is defprotocol related to this capability somehow?

2:39 * wooby would enjoy a blog post covering 1.2 changes with code examples

2:45 wooby: hm, i think ill try and write it myself :)

3:15 TheBusby: is there a simple way to check to see if something can be treated like a seq? I know (seq? [1 2]) is false...

3:15 Licenser_: morning

3:16 TheBusby: try coll?

3:16 ,(coll? [1 2])

3:16 clojurebot: true

3:16 Licenser_: ,(coll? '(1 2))

3:16 clojurebot: true

3:16 TheBusby: Licenser_: perfect, thank you!

3:16 * Licenser_ bows

3:17 _ato: ,(coll? "wheeeeeee")

3:17 clojurebot: false

3:17 _ato: ,(seq "wheeeeee")

3:17 clojurebot: (\w \h \e \e \e \e \e \e)

3:17 _ato: ,(seqable? "wheeee")

3:17 clojurebot: true

3:17 Licenser_: and again _ato is smarter then me :P

3:18 _ato: though coll? maybe more appropriate if you

3:18 re really only interested in clojure collections

3:18 TheBusby: yep, '() [] and {} for the moment

3:18 Licenser_: (coll? {1 2})

3:18 ,(coll? {1 2})

3:18 clojurebot: true

3:18 TheBusby: strings are have their own special level of hell reserved for them

3:19 Licenser_: ^^

3:20 tomoj: just need a language where the words are integers

3:20 Licenser_: 1

3:21 _ato: hellish java string fact of the day: if you take a small subsequence of a large string and then release all references the large string so it can be garbage collected, prepare to be surprised by an OutOfMemoryError

3:21 Licenser_: *g*

3:22 TheBusby: java strings have their fun, but everything after pure ASCII strings in C seems to have their own pain

3:22 _ato: I think it'd be interesting to try implementing strings using persistent vectors, it'd allow structure sharing but still avoid that problem

3:22 and would probably make manipulating them faster

3:23 TheBusby: I blame human language though, that's the difficult part

3:23 Chousuke: ICU is one of the few string libraries that is probably doing it right :P

3:23 _ato: or maybe they'd just be slower, dunno. Interesting to try

3:23 TheBusby: usually speed is the big problem

3:23 gregh: any problem in compute science can be solved with an other level of indirection

3:23 computer

3:24 Chousuke: gregh: including the problem of overabstraction?

3:24 TheBusby: haha

3:24 gregh: except that one, it creates other problems. :)

3:31 anyway, one could create a string class that broke up the backend representation when a substring was taken, so that freeing the large string would allow the majority to be garbage collected

3:31 you'd have to have a pretty good reason to go to all that work, though.

3:32 _ato: true that

3:33 I occasionally run into the problem when parsing lots of big data and tallying things up in hash tables, based on some smaller part of the larger text

3:33 just need to remember to explicitly copy the string before sticking it in your hash map

3:58 Licenser_: meep meep

4:02 TheBusby: er, what's the trick where you can call a function before you define it? It's not bind...

4:02 _ato: declare

4:02 TheBusby: _ato: thank you!

4:04 esj: I do declare Mr. Beauregard... !

4:05 mercy, i need more coffee.... I'm turning into a ircbot

4:21 Licenser_: _ato: do you still remember the bug in clj-sandbox that let you remap a function wiht var and such?

4:22 you called .alterRootVar on some var and then got a eval with another name when I remember correctly

4:22 _ato: hmm

4:24 Licenser_: i think clj-sandbox fixed the issue now but I'd like to make a test to be sure :P

4:25 _ato: Licenser_: see #clojure-casual

4:53 MadWombat: Hello all

4:55 If I have a defstruct with 3 keys and 3 lists of items, how can I make a bunch of structs picking values from each collections?

4:57 AWizzArd: Hi MadWombat. Could you please describe what you want in other words? I am not sure that I understand what you want.

4:57 bsteuber: MadWombat: you pick values with (:your-key your-struct) - not sure if this is what you need either :)

4:58 Chousuke: (map (partial struct yourstruct) list1 list2 list3)

5:29 AWizzArd: What do you use to parse .csv files? Does Contrib have something for that?

5:31 zmila: certainly, exists some java implementation of csv-parsers

5:32 esj: AWizzArd: I think there is such a thing in Incantre

5:43 stilkov: AWizzArd: check out http://github.com/davidsantiago/clojure-csv

5:46 AWizzArd: oki

5:49 stilkov: how does one submit bug reports for the clojure documentation?

5:50 Chousuke: normally through the ticket system I think

5:50 there's a support tab in assembla for non-members

5:52 stilkov: thanks

6:26 npoektop: what does mean :memory here (defmethod read-session :memory [repository id] ...)?

6:27 bsteuber: npoektop: depends on the definition of read-session

6:28 http://clojure.org/multimethods

6:28 npoektop: bsteuber, thank you

7:03 _exterm: Is the ASM used by the clojure compiler just a repackaged org.objectweb.asm? if so, which version is it based on?

7:20 eglyph: hello, I'm following the tutorial on enlive. scrape3 isn't working. it uses html/selector which is not in enlive-html. am I missing something?

7:34 _exterm: Hey people, I'm trying to do some modifications to the Compilre to target a different Bytecode format. Is the ASM used by the clojure compiler just a repackaged org.objectweb.asm? if so, which version is it based on?

7:37 hoeck: _exterm: yes, its org.objectweb.asm, but I don't know which version

7:38 eglyph: which tutorial?

7:39 eglyph: hoeck: the one from here: http://github.com/swannodette/enlive-tutorial/

7:40 everything is fine until the moment i have to load scrape3.clj. i've browsed the source and couldn't find 'selector' function

7:42 hoeck: eglyph: and what about net.cgrand.enlive-html/selector ?

7:43 eglyph: is there such a function in enlive-html?

7:44 enlive-html is :required as html, so the full name is html/selector. this is the exact thing the slime barks at.

7:44 hoeck: I don't know, I have never used enlive :P

7:44 eglyph: as i've already said there is no such function in enlive-html

7:44 hoeck: ah, okay, I first thought that there was a typo somewhere in scrape3 or so

7:45 eglyph: it seems the idea is to use partial application if i'm not mistaken

7:59 kzar: So I'm probably just being stupid here but I'm a bit confused by the clojure.http.client package. If I call the request function with the GET method and add some parameters the parameters seem to be given as post ones. (Even though the :method of the response is GET) . If I add the parameters manually to the URL they are sent as GET parameters though

8:01 AntonyBlakey: eglyph: in enlive_html.clj line 343 I see (defmacro selector ...) so it's definitely defined. This is enlive 1.0.0-SNAPSHOT

8:04 eglyph: AntonyBlakey: I'm looking into git version and there is no such string. the tutorial is installed with lein. so probably enlive evolved and the tutorial is not updated.

8:04 AntonyBlakey: Check your project.clj to see what version lein specifies.

8:05 I did the tutorial with no problems.

8:05 eglyph: enlive "1.0.0-SNAPSHOT"

8:06 i've just looked into enlive jar for the source. it's not there. wtf?

8:06 _ato: kzar: I think when you pass a map to it it always puts it in the request body, not the query string. If you want to put something on the query string add it to the URL (you can probably use url-encode on the map to format it)

8:07 AntonyBlakey: eglyph: It is in mine (along with many compiled files)

8:08 eglyph: AntonyBlakey: my jar consists two .clj files. this is weird.

8:11 AntonyBlakey: You have a later version than I do.

8:13 Did you perchance checkout out the git head and build it?

8:14 Licenser_: hmm has anyone worked with clojure to write gui applications?

8:14 AntonyBlakey: berlin brown has a blog post on it.

8:15 http://berlinbrowndev.blogspot.com/2009/04/practical-clojure-with-swt-junit-and.html

8:15 sexpbot: "Berlin Brown and Software Development: Practical Clojure with SWT, JUnit and Spring"

8:15 eglyph: AntonyBlakey: no, I just cloned the tutorial repo and followed all the steps listed. lein deps && lein repl

8:15 AntonyBlakey: and this for swing: http://stuartsierra.com/2010/01/03/doto-swing-with-clojure

8:15 sexpbot: "doto Swing with Clojure - Digital Digressions by Stuart Sierra"

8:15 hoeck: eglyph: it looks like the selector macro was there for "compiling" selectors

8:16 and now selectors are just plain functions, and you don't have to compile them

8:16 eglyph: hoeck: so the tutorial won't work until fixed, right?

8:16 hoeck: eglyph: can you try to just remove the html/selector calls from the scrape3 source?

8:17 AntonyBlakey: eglyph: try deleting the enlive jar from your lib directory and your ~/.m2/ repository

8:17 hoeck: eglyph: yes, but it looks like its a very small fix

8:17 AntonyBlakey: But I don't think that 1.0.0-SNAPSHOT matches git head.

8:17 hoeck: eglyph: but beware, I'm only doing a wild guess while looking at the commits

8:18 vu3rdd: _exterm: The clojure package in debian is built without the in-built asm library, as asm is already present as a deb package

8:18 and it uses version 3.2 of libasm3

8:20 Licenser_: thanks fort the links

8:20 kzar: _ato: OK fair enough

8:22 eglyph: hoeck: right, selector has to be removed from the code

8:22 hoeck: eglyph: that was the commit I'm referring to: http://github.com/cgrand/enlive/commit/6d0919ef2be0f3ea1ca69335d8098ec9e3442f8f#L0R369

8:26 eglyph: i wonder why it is not working

8:55 jfields: are there any conventions around naming in clojure? obviously dashes are preferred to camel case or underscores. Anything else? For example, Stu Halloway once mentioned to me that foo->bar was idiomatic for converting a foo to a bar. Are there other things like that?

8:55 kzar: _ato: Hey you're right the url-encode function works really well, I love how you can just give it a map. It was actually really easy to make a little web-request function that behaved like I expected. Thanks

8:56 Chousuke: jfields: ? for predicates

8:57 hoeck: jfields: also: http://www.assembla.com/wiki/show/clojure/Clojure_Library_Coding_Standards

8:57 sexpbot: "Clojure Library Coding Standards | Clojure | Assembla"

8:57 Chousuke: jfields: also parameter names: coll for collection, x, y, n, f, g for functions etc.

8:57 hm, that's unclear. obviously not x, y, n for functions :P

8:58 also shadowing core names is okay if the scope of shadowing is short-lived

8:59 jfields: Chousuke, hoeck: cool, thanks

8:59 Chousuke: eg. (defn reverse-map [map] (into {} (for [[k v] map] [v k]))) is okay

9:01 jfields: I've been using using name:age for maps (e.g. (def name:age {:jay 30 :bob 24}), but I was considering switching to name=>age and saving : for another option

9:05 hoeck: jfields: colons in symbols are not a good idea

9:06 jfields: hoeck: why's that?

9:07 hoeck: it looks bad, as they are usually reserved for keywords

9:09 an idiomatic name would be name-age

9:10 jfields: hoeck: which looks good in that case, but when it's first-name-last-name you loose a bit of information. first-name=>last-name is more expressive, to me.

9:10 s/loose/lose/

9:10 Chousuke: I think it's better to name the map to indicate its use

9:11 though if such a naming scheme is not applicable, name=>age is pretty good

9:11 or name->age

9:11 hoeck: right

9:13 jfields: chousuke: name->age, really? doesn't that collide with the idea that -> is usually used for converting from one data structure to another?

9:13 Chousuke: nah.

9:13 AWizzArd: Could be a map :)

9:14 jfields: cool. I'm just learning this stuff, so thanks for the ideas.

9:14 Chousuke: the type->another is mostly borrowed from scheme

9:15 but in clojure most things for now are simply maps and vectors and strings and other mundane data types. there isn't much need for conversion

9:15 jfields: chousuke: sadly, most of my clojure work is tied very closely to an existing java codebase. which has me doing ->'s all the time.

9:16 Chousuke: heh.

9:16 AWizzArd: “(->” does not look similar to “(something->”

9:17 Chousuke: AWizzArd: I don't think anyone said that.

9:32 dnolen: hmm, so is it possible to augment final Java classes now with extend ?

9:36 chouser: yes

9:38 kzar: Is there a way to use clojure.xml/parse with a string or a seq of strings instead of giving it a filename to open?

9:39 * dnolen mind is blown

9:40 zmila: kzar - xml/parse can parse InputStream

9:42 chouser: ,(clojure.xml/parse (java.io.ByteArrayInputStream. (.getBytes "<foo>bar</foo>")))

9:42 clojurebot: {:tag :foo, :attrs nil, :content ["bar"]}

9:42 chouser: dnolen: yeah, that's appropriate

9:45 kzar: zmila: chouser: sweet thanks

9:45 Which html parser would you recommend that can handle badly formed html?

9:46 chouser: tagsoup

9:46 works quite nicely with clojure.xml/parse

9:46 dnolen: ok, interesting so you can't extend a protocol right ? (that is, without having to specify a list of types)

9:48 chouser: "extend" is always about filling in an implementation at an intersection between a protocol and something else.

9:48 the something else can be a record, datatype, class (even a final one), or an interface.

9:48 dnolen: chouser: so that's a no. A protocol is a bit less of dynamic thing. I can't change the protocol for everyone (a bad idea I suppose anyway), but I can extend it for a type.

9:49 chouser: oh... you mean add a new method spec to an existing protocol?

9:49 dnolen: chouser: yes

9:49 chouser: I think you can redefine the entire protocol if you want.

9:50 dnolen: chouser: sure, I meant for users of a library which defines a protocol. They want to add some methods, they have to extend all types that use that protocol, they can't jsut add a method spec.

9:54 kzar: chouser: Dumb question but how do I know what to add to my use list to get clj-tagsoup working? Usually I just add the name but with hyphens changed to periods so [clj.tagsoup] but it's not working

9:56 _ato: kzar: if it's not documented then just look at the directory structure of the library source, or read the (ns ...) line: http://github.com/nathell/clj-tagsoup/blob/master/src/pl/danieljanus/tagsoup.clj

9:57 so in this case you'd use pl.danieljanus.tagsoup

9:57 dnolen: kzar: hyphens don't become periods.

9:58 chouser: it looks like http://github.com/nathell/clj-tagsoup produces different output than clojure.xml

9:58 if you want clojure.xml output, you don't need a clojure lib, just the tagsoup java .jar

10:03 dnolen: chouser: now that I think about it, I suppose it doesn't matter really, since protocols are composable :)

10:04 Licenser: aloa

10:05 kzar: Hmm so I did try pl.danieljanus.tagsoup earlier, but no luck. At least I was onto something. It's good to know how that stuff should work even if it's not all functioning atm. I tried unziping the tagsoup jar somewhere and the structure looks like org/ccil/cowan/tagsoup so I should add [org.ccil.cowan.tagsoup] to my use list right? Well that's not working either so I'm stumped now

10:06 vu3rdd: I see something very strange with (comment .. )

10:07 if I put some clojure code there, it is fine. But if I use a strange looking identifier (for example cut and paste code from dotrace), it does not compile

10:07 ,(comment (defn x[] "true"))

10:07 clojurebot: nil

10:08 vu3rdd: ,(comment (defn x[] xyz:))

10:08 clojurebot: Invalid token: xyz:

10:08 kzar: (I have copied all of the jars from tagsoup into my classpath and restarted Clojure)

10:09 Licenser: kzar: there is already a clojure lib for tagsoup that works

10:10 kzar: Licenser: Yea I think you joined us mid convo, I was having trouble getting that loaded and it was suggested I tried the tagsoup directly

10:10 lpetit: hello all

10:11 vu3rdd: Is this behaviour with (comment) a bug?

10:13 Licenser: ah okay, kzar I used it with the lib and it worked w/o any trouble - strange

10:14 kzar: Licenser: Heh it's very probably me doing something wrong to be fair on the library

10:14 * Licenser digs up the source

10:15 Licenser: http://github.com/Licenser/lein-search/commit/7fc3030f8d5edb0099f93ea87717673d3f6fe56f#diff-7

10:16 actually http://github.com/Licenser/lein-search/blob/7fc3030f8d5edb0099f93ea87717673d3f6fe56f/src/leiningen/update.clj

10:17 kzar: OK I got it working, turns out killing Clojure wasn't enough to add the new jar files to the classpath. I needed to restart emacs

10:17 On the plus side I understand how things are laid out a lot better now

10:22 Licenser: :)

10:28 kzar: Oh sweet it's working really well with the inspect-tree gizmo, go to love that

10:29 Does anyone know of a good guide that explains the difference between map apply and reduce? I know it's dumb but the difference always eluded me

10:38 liebke: kzar: a one line summary of the difference is: (apply foo [1 2 3]) => (foo 1 2 3), where as (reduce foo [1 2 3]) => (foo (foo 1 2) 3).

10:39 bsteuber: and (map foo [1 2 3]) => [(foo 1) (foo 2) (foo 3)]

11:07 caljunior: so I was creating this ubercool gui. Sensibly designed it in Netbeans. :import into clojure. everything beautiful. now I want to (.add some-panel (JScrollPane. some-table)) but the ScrollPane is not showing up in the gui. Just an empty JPanel.

11:09 all my (.addActionListnener ) code I added in clojure is working just fine in the same gui.

11:09 Chousuke: did you try forcing a repaint of the whole thing?

11:10 caljunior: no

11:10 Chousuke: or check if the scroll pane has any visibility flags

11:10 caljunior: ok

11:10 Chousuke: There's no reason code like that shouldn't work as far as I know so I'm guessing it's for some reason deciding it doesn't need to get drawn at all

11:15 caljunior: (.repaint some-panel) no effect.

11:16 *checking visibility flags on scroll pane*

11:23 jfields: does anyone know if (append-spit) is thread safe?

11:24 Chousuke: ~def append-spit

11:24 clojurebot: It's greek to me.

11:24 jfields: it's in contrib

11:25 stuartsierra: define thread safe

11:25 Chousuke: you can always ensure it's thread safe by using an agent :P

11:25 jfields: I have two threads that want to write to the same file

11:26 stuartsierra: Since it's just calling java.io methods, I would guess it is not safe to use from 2 threads at once.

11:26 jfields: okay, thanks

11:27 Licenser: I really want to write a program with agents, then I'd call them all smith

11:27 stuartsierra: ha

11:27 Chousuke: (def agent smith)

11:27 Licenser: (def smith agent)

11:27 Chousuke: except hm, yeah

11:28 Licenser: then you could use smith instead of agent

11:34 cYmen: I wrote down how I set up vimclojure in case anybody wants to try it: https://simon.a-oben.org/pages/clojurestart.html

11:36 Licenser: I really start to dislike my typo blog :( is there any nice clojure one out there? (yes I know I could write my own)

11:36 propably would be best but I hate to do the graphics part -.-

11:38 tomoj`: "typo" blog?

11:38 Licenser: for god's sake :(

11:38 yes it's a blog in rails

11:38 tomoj`: oh, yeah

11:38 Licenser: and I start to hate it more and more

11:38 tomoj`: naturally

11:39 Licenser: it refuses to update properly and their new editor is hell - I am scared to share the link of my latest post since it is embarrasing ...

11:40 rfg: Link please.

11:40 :)

11:41 I've been writing a blog in CL.

11:42 I would've liked to write it in clojure, but it's cgi based and so the jvm start-up speed is problematic.

11:42 Licenser: rfg: it is broken

11:42 I really want to cry now

11:43 *sighs*

11:47 Chousuke: rfg: have you considered using a wholly jvm-based approach like compojure with Jetty serving the pages?

11:48 zaphar_ps: rfg: I went the static site generator route and wrote it in clojure

11:49 rfg: Chousuke: It's mainly because my host was only (until recently) allowing cgi. They've started a beta for using arbitrary technologies which should allow me to do it in clojure, but I need to look into how one goes about doing that.

11:51 ivenkys: afternoon gents - i am learning clojure (and loving it) , any suggestions for code to look at for getting the idioms right

11:51 rfg: zaphar_ps: Yeah, I've seen your site.

11:52 zaphar_ps: really?

11:52 heh

11:52 not used to people having been there :-)

11:52 it's pretty low traffic

11:52 ivenkys: a small library or so would be perfect -

11:52 rfg: Actually could people have a look at my site and tell me if it is rendering nicely. www.radiofreeganymede.com

11:52 sexpbot: "Radio Free Ganymede"

11:53 zaphar_ps: ooohhhh pretty picture :-)

11:53 looks ok to me chrome on ubuntu linux

11:54 slaxor: looks fine on firefox, windows 7

11:54 rfg: I hope to have my first clojure related post up soon.

11:55 Licenser: rfg: looks good on safari http://grab.by/3VYX

11:56 dnolen: invenkys: you could look at cgrand's repo on GitHub tho his work tends to be ultra-dense. The clojure-contrib 1.2.0 libraries might also be educational.

11:56 ivenkys: ^

11:56 ivenkys: dnolen: tkx

11:57 rfg : looks good in Firefox on Arch

11:57 lrenn: can you pass wild cards clojure.lang.Compile? In other words, is there a way to write an ant task that doesn't list an argument for each namespace?

11:58 stuartsierra: with a lot of ant XML foolery, yes

11:58 chouser: if you compile a namespace the 'require' or 'use's other namespaces, all those will be compiled too

11:58 Licenser: for gods sake :( this is so sad

11:59 lrenn: grr, thanks folks.

12:00 cemerick: lrenn: if you're locked into ant, but not opposed to using the maven-ant-tasks lib there, the clojure-maven-plugin supports wildcards for choosing which namespaces to compile.

12:03 lrenn: cemerick: thanks but I'm just fooling around with reimplementing lein in ant + ivy :) It's backwards compatible other than having to specify all your namespace :(

12:03 cemerick: heh, ok

12:29 ad4m: I've got a Clojure app that calls for clojure-1.2.0-master-SNAPSHOT.jar, but I can't seem to find that specific version in any repositories. I have found repos with 1.2.0 snapshot jars with dates in the name, do I just grab one of those and rename it to remove the date?

12:30 Chousuke: ad4m: that name is what you get if you build from git, but renaming a maven snapshot will work as well

12:32 ad4m: Ah, so is the idea that I should have that already in my local repo by way of building it from source?

12:33 Chousuke: probably

12:33 though the clojure build process doesn't actually put it in a *maven* repo. it just produces a jar

12:34 ad4m: Ok, well I don't know much about Maven either, but I'd assume there's a way to publish things to your local repo or cache or whatever it is called

12:34 Thanks

12:44 raek: how is clojure's = done for java arrays? identity? element-wise-=?

12:45 dakrone: raek: the doc is pretty clear about it

12:45 ,(doc =)

12:45 clojurebot: "([x] [x y] [x y & more]); Equality. Returns true if x equals y, false if not. Same as Java x.equals(y) except it also works for nil, and compares numbers and collections in a type-independent manner. Clojure's immutable data structures define equals() (and thus =) as a value, not an identity, comparison."

12:47 dakrone: raek: does that make sense? so they are equal if the elements are equal

12:49 stuartsierra: ,(= (list 1 2 3) (seq [1 2 3]))

12:49 clojurebot: true

12:50 dakrone: ,(= '(1 2 3) [1 2 3])

12:50 clojurebot: true

12:50 dakrone: even though one's a list and one's a vector

12:50 Chousuke: arrays have identity comparison.

12:50 stuartsierra: ,(= (list [:a 1] [:b 2]) {:a 1 :b 2})

12:50 clojurebot: false

12:51 Chousuke: ,(seq {:a 1 :b 2})

12:51 clojurebot: ([:a 1] [:b 2])

12:51 Chousuke: ,(= (into-array [1 2 3]) (into-array [1 2 3]))

12:51 clojurebot: false

12:52 Chousuke: (this is not a bug :))

12:55 slyphon: what's the deal with the "trailing attr-map" style?

12:56 * slyphon 's been looking for an example somewhere

12:56 Chousuke: what do you mean?

12:57 tchalvak: Hey, it worked!

12:57 slyphon: what does it look like?

12:57 you can do (defn blah "docstr" {:private true} [] )

12:58 or even (defn #^{:private true} blah "docstr" [] (println "yay!"))

12:58 there's a form where the metadata comes at the end, though

12:59 no?

12:59 * slyphon wonders if he imagined that

12:59 Chousuke: hm

12:59 tchalvak: So I'm trying to get started with clojure, and I'm running into problems including jars, can I do this, to test that I have the right paths, from within the repl? And if so, what's the actual syntax?

12:59 Chousuke: maybe with multiple arities?

12:59 (doc defn)

12:59 clojurebot: "([name doc-string? attr-map? [params*] body] [name doc-string? attr-map? ([params*] body) + attr-map?]); Same as (def name (fn [params* ] exprs*)) or (def name (fn ([params* ] exprs*)+)) with any doc-string or attrs added to the var metadata"

12:59 slyphon: maybe

12:59 stuartsierra: (defn foo ([x] ...) {:meta ...})

13:00 slyphon: ahhh

13:00 ok, i couldn't figure out where the meta went in that form

13:00 stuartsierra: thanks

13:00 stuartsierra: test that; I think it's right

13:00 * slyphon tries

13:01 Chousuke: tchalvak: do what?

13:01 slyphon: stuartsierra: yep, that's it, thanks!

13:02 stuartsierra: you're welcome

13:02 tchalvak: Get a jar library loaded, e.g. if I'm running the repl from a folder, and I have a blog/blog.jar file, can I include or require it from within the repl?

13:02 stuartsierra: nope

13:05 tchalvak: Ah.

13:05 raek: is there a simple way to make a hex string from a java byte array?

13:06 bsteuber: tchalvak: but you can check your existing classpath with (System/getProperty "java.class.path")

13:08 tchalvak: bsteuber: *nods* Ok, I guess that'll work for seeing whether the stuff that I'm trying to load is actually present, guess I'll have to load & reload the repl and test different configs. Thanks.

13:08 bsteuber: tchalvak: yes, I think so - or you can use something like leiningen or maven that sets up the right classpath for you

13:09 dakrone: raek: something like this perhaps: (map #(Integer/toHexString (int %)) [\a \b \c\ \d])

13:09 ,(map #(Integer/toHexString (int %)) [\a \b \c\ \d])

13:09 clojurebot: ("61" "62" "63" "20" "64")

13:09 dakrone: you'll need to convert your byte array to a seq first

13:10 ,(map #(Integer/toHexString (int %)) (.getBytes "asdf"))

13:10 clojurebot: ("61" "73" "64" "66")

13:10 dakrone: or not

13:11 bsteuber: tchalvak: but actually google gives some results for "java modify classpath runtime" - maybe you can experiment with this..

13:12 would be nice to do a clojure wrapper around this

13:12 stuartsierra: Clojure has add-classpath, deprecated because it doesn't work in many environments

13:12 tchalvak: bsteuber: Not to dump lots of text, but: user=> (System/getProperty "java.class.path")

13:12 "/home/tchalvak/sites/threadhack/www/crow-blog:/home/tchalvak/sites/threadhack/ClojureX/lib/clojure.jar:/home/tchalvak/sites/threadhack/ClojureX/lib/clojure-contrib.jar:compojure/compojure.jar:rhino1_7R2/js.jar:blog/admin.clj:blog/config.clj:blog/db.clj:blog/layout.clj:blog/markdown.clj:blog/pages.clj:blog/rss.clj:blog/server.clj:blog/test.clj:blog/tokyocabinet.clj:blog/util.clj:::/home/tchalvak/sites/threadhack/ClojureX/lib/jlin

13:12 So I have all these files actually loaded, so that's good.

13:12 I'm using ClojureX, if y'all are familiar.

13:13 Which allows me to push some jar paths into a .clojure file on a project to project basis, and it's apparently working.

13:13 Chousuke: tchalvak: that's not right.

13:13 tchalvak: you should have the root of the blog dir in the classpath, not the blog files itself

13:14 tchalvak: Now I just have to figure out how to actually access the namespaces in the jars.

13:14 Right, I was kinda confused by that, I'll try that then.

13:14 Chousuke: so that you can have (ns blog.pages) etc.

13:15 hiredman: tchalvak: you have some clojure files (.clj) on the classpath, that is not going to work

13:15 I should say in the classpath

13:15 jars and directories go on the classpath

13:16 Chousuke: ie. if it's /foo/bar/blog/some.clj, add /foo/bar/ to the classpath

13:16 tchalvak: Chousuke: that applies to both .jar files and .clj files?

13:16 Chousuke: jar files are added directly

13:16 they represent a directory

13:16 tchalvak: Chousuke: Ok, I'll edit based on that then.

13:17 Chousuke: they contain a /foo/bar/whatever.clj hierarchy

13:17 the point is that in java, and in clojure, a namespace segment corresponds to a directory

13:17 except the last one, which is the file/class

13:18 so if you have some directory or jar in a classpath, the directories under it are going to represent namespaces

13:18 again, a jar is essentially just the dir hierarchy packaged into a zip file. you could extract it and it'd work just the same :)

13:23 bsteuber: stuartsierra: thx for the information

13:24 stuartsierra: np

13:34 tchalvak: Chousuke: 'lright, I guess I'm part of the way there, I'll have to go back to the book now and start at the beginning.

13:34 Chousuke: Thanks.

13:42 joshua-choi: I've been reading http://www.assembla.com/wiki/show/clojure/Clojure_Compiler_in_Clojure.

13:42 What does "written in Clojure" mean?

13:42 As in, clojure.core and other libraries that use no direct Java and no eval?

13:42 chouser: no, that page is just about the compiler

13:43 slyphon: wow, STM FTW

13:43 joshua-choi: chouser: So it doesn't have to be in "pure" Clojure then?

13:43 chouser: which is currently clojure/lang/Complier.java

13:43 slyphon: talk about "a good idea"

13:43 joshua-choi: Well, that's written in Java.

13:43 chouser: joshua-choi: oh, I see. yeah, I think it will strive to be pretty pure

13:44 cemerick: so, I wonder who's going to knock out a far shorter/simpler version of this... http://wiki.github.com/FaKod/Circuit-Breaker-for-Scala/

13:44 joshua-choi: chouser: Does "pure" mean end-dependent on clojure.core and special forms only?

13:44 chouser: that is, the reader gives it clojure structures, and the compiler analysis phase will produce nested clojure records.

13:45 joshua-choi: Clojure records?

13:45 chouser: the step after that is to produce jvm bytecode, and of course that will use java and be java-specific.

13:45 yeah, defrecord instances.

13:45 joshua-choi: The reader would spit out raw lists, vectors, maps, and sets. What would need to be translated into records from those?

13:46 chouser: right, that's what the analysis phase of comiplation does now -- takes the raw lists, vectors, etc. and produces a tree of Java objects.

13:46 the new code would be in clojure instead, do roughly the same kind of analysis and produce a tree of records.

13:46 joshua-choi: What would need to turn into objects? Functions?

13:47 Oh, I see

13:47 chouser: the analysis phase includes all macroexpansion, type inference, and the necessary gathering of data about primtives, local scopes, local-clearing (to prevent head-holding), etc.

13:48 joshua-choi: Would it matter if the reader was in pure Clojure too? Or could the current Java-implemented reader suffice for the compiler?

13:49 chouser: The reader can be ported to clojure independently. The interface between the reader and compiler is very well understood, so you should be able to mix-n-match implementations.

13:49 a couple people have written more or less complete Clojure readers in Clojure now.

13:50 joshua-choi: Really? I wrote one too, but I didn't know that there were others.

13:50 chouser: Chousuke and hiredman each wrote one

13:51 that's not including things very like clojure readers that the various syntax highlighters may have.

13:51 joshua-choi: Do you know if their sources are on GitHub?

13:52 chouser: I think one of the early steps for ccinc will be to make the interface between the compiler's analysis phase and emit phase as stable and accessible as the reader->compiler interface. That will allow compilers and emitters to be developed independently as well.

13:54 joshua-choi: The difference, I think, is that the reader-compiler phase uses standard Clojure data structures, while the compiler current emits weird Java objects that I don't get. :)

13:55 I don't dare to figure them out.

13:55 chouser: right, that's why I want that to change early, into nice printable records

13:55 I'm actually kinda itching to get started on that, but I have this pesky book to finish first...

13:55 joshua-choi: It's not that urgent, though, anyway, right?

13:55 hiredman: joshua-choi: http://github.com/hiredman/clojure/blob/readerII/src/clj/clojure/reader.clj

13:56 chouser: hiredman: thanks. I was getting close...

13:56 oh, I'm in readerIII ...maybe not so close.

13:56 stuartsierra: I started sketching out plans for a reader that was entirely table-based.

13:57 joshua-choi: hiredman: Whoo, that is amazing, and dense. Why did you put your helper methods in a letfn?

13:58 hiredman: joshua-choi: to avoid var lookups in the generated bytecode

13:59 joshua-choi: hiredman: Does it make it that much slower? I've always wondered if I should use let instead of def for my parser libraries...

14:00 hiredman: http://groups.google.com/group/clojure-dev/browse_frm/thread/b20ce4b2c226ffcf/26860813d1c39ba1?lnk=gst&q=reader#26860813d1c39ba1

14:00 it's not a speed thing it is a bootstrapping thing

14:00 joshua-choi: Ah

14:04 Chousuke: my reader is out of date now though :P

14:04 hiredman: ditto(?)

14:05 chouser: what's changed in the reader??

14:05 Chousuke: meta warnings? :

14:05 hiredman: ^ deprication at least

14:06 chouser: oh

14:06 Chousuke: I think I might still be missing some obscure features

14:06 and to be honest this thing has basically no design :P I wrote it as I thought it

14:07 http://github.com/Chousuke/clojure/blob/clojure-reader/src/clj/clojure/lang/reader.clj

14:07 hiredman: http://www.thelastcitadel.com/images/syntax.png

14:08 * chouser blinks, closes that tab.

14:08 hiredman: trace of the method for reading syntax quote stuff

14:08 joshua-choi: Did you generate that?

14:08 Chousuke: I toyed with writing a restartable reader but after I had a restartable string reader I stopped; I wish Clojure had continuations

14:09 oh, and yeah, syntax-quote is horror

14:09 hiredman: joshua-choi: yes

14:09 Chousuke: I tried implementing it as a macro at first but gave up after a lot of frustration

14:10 joshua-choi: Chousuke: When I wrote mine, I cheated and just used `syntax-quote.

14:10 ad4m: Is there a way to add annotations to a class created with proxy?

14:11 hiredman: well, generated as in walked through the code trying to figure out what it did

14:11 joshua-choi: ad4m: I don't think so. It is unfortunate.

14:11 hiredman: Ah, I meant if you used some program to do it for you.

14:11 ad4m: What about gen-class?

14:12 Chousuke: I don't think Clojure has any support for annotations

14:12 joshua-choi: It's a commonly requested feature. But comparatively low in priority, I think.

14:13 ad4m: Is there any sort of work around?

14:13 chouser: can you use the Java reflection API to adjust annotations?

14:14 ad4m: I don't know. I don't know that much about Java.

14:14 joshua-choi: Do you have a Java library that requires them?

14:15 In any case, I found http://tutorials.jenkov.com/java-reflection/annotations.html from Google

14:15 sexpbot: "Java Reflection: Annotations"

14:15 ad4m: Yes, I'm using a library named Netty

14:16 technomancy: every time people ask about annotations, it's for netty. I wonder if you could write one Java wrapper around Netty that would let it be used without annotations and be done with it.

14:16 would probably be less work than adding annotations to clojure

14:17 pareto principle, etc, etc.

14:17 hiredman: ad4m: have you seen http://github.com/texodus/saturnine?

14:18 ad4m: hiredman: Nope, but the one line description makes it sound interesting. I'll check it out.

14:19 danlarkin: I love this: "clojure.contrib.logging integration (is this a feature?)"

14:21 ad4m: Thanks for the various ideas

14:42 zakwilson: I am frequently impressed by the well-reasoned explanations of design decisions posted to clojure.org.

14:52 Is fnparse the only game in town for a Clojure parser library?

14:54 replaca: zakwilson: the only one I know about, but some folks have also used antlr

15:21 jfields: how do I change {:a {:b 1}} to {:a {:b 2}}

15:21 chouser: ,(update-in {:a {:b 1}} [:a :b] inc)

15:21 clojurebot: {:a {:b 2}}

15:21 Licenser: jfields:

15:21 what chouser sayd

15:21 jfields: cool, thanks all

15:22 ,(update-in {:a {:b 1} :c {:d 1} [:a :b] inc)

15:22 clojurebot: Unmatched delimiter: )

15:22 jfields: ,(update-in {:a {:b 1} :c {:d 1}} [:a :b] inc)

15:22 clojurebot: {:a {:b 2}, :c {:d 1}}

15:26 Raynes: "If the main difference between Scala and Clojure is syntax then the main difference between an F-22 Raptor and Boeing 777 is paint color." - James Iry on Twitter.

15:27 chouser: Which demands an answer to the question, which is which?

15:29 LauJensen: Incredible that the discussion is still going. Remind me to be careful what I blog about if I'm ever Head of Development anywhere, people pay attention :)

15:31 slyphon: doh, does 'binding' create a recursion point?

15:31 chouser: nope

15:31 slyphon: hrm

15:31 * chouser double-checks

15:32 slyphon: chouser: https://gist.github.com/f1044271c41c0880bee6

15:32 error: java.lang.IllegalArgumentException: Mismatched argument count to recur, expected: 0 args, got: 1 (jms_internal.clj:177)

15:32 i thought that recur would re-enter from the top of the fn

15:33 chouser: hm, that's a reasonable assumption. :-)

15:33 slyphon: ok! :D

15:33 hooray for not being crazy

15:33 ohhhh

15:33 dnolen: "@jamesiry and Clojure is the nimble single-seat F-22, Scala the enterprise passenger-bearing beast, right? ;)" - David Nolen on Twitter

15:33 slyphon: jta/with-user-transaction is a macro

15:34 * slyphon sighs

15:34 * slyphon thought he was gonna be uber-l337 here

15:34 chouser: hm, binding acts funny in a recur loop.

15:34 slyphon: ah

15:34 no TCO for you!

15:34 * slyphon puts on his "Soup Nazi" voice

15:35 Raynes: I wish that MMEmail guy would do a full on JavaMail Clojure wrapper or similar.

15:35 chouser: slyphon: you're expecting *session* to retain it's newly-bound value when you recur there?

15:35 slyphon: chouser: yeah, that was the idea

15:35 chouser: I think it might actually behave that way currently, though I don't know that I'd rely on that.

15:35 slyphon: i don't *really* need to recur there

15:36 i can just call myself again, it's not like it's a recursive algorigthm

15:36 er

15:36 algorithm

15:37 chouser: oh my.

15:37 slyphon_away: chouser: ?

15:38 chouser: binding acts really weird with recur

15:38 slyphon_away: that's good to know

15:38 chouser: like 'set!' almost

15:38 slyphon_away: ouch!

15:39 chouser: oh, maybe not. weird

15:39 hiredman: recur is gnarly, really tco would make things much simpler

15:40 chouser: ,(loop [a 1] (prn *print-level*) (when (< a 3) (binding [*print-level* 10] (recur (inc a)))))

15:40 clojurebot: 30 10 10

15:40 chouser: ,*print-level*

15:40 clojurebot: 30

15:41 chouser: actually ... that's exactly right. 'recur' inside binding acts like a recursive call, with the original value restored on final exit.

15:41 ,(loop [a 1] (prn *print-level*) (when (< a 5) (binding [*print-level* a] (recur (inc a)))))

15:41 clojurebot: 30 1 2 3 4

15:45 Chousuke: chouser: I think that's a good demonstration of how dynamic scope is not "visually" intuitive like lexical scope :)

15:46 even though it often appears to be

15:46 stuartsierra: I've come to the conclusion that dynamic scope is a bug.

15:47 chouser: heh

15:47 Chousuke: kind of like goto is a bug

15:48 chouser: but there are so many clever tricks it can do

15:50 rfg: It makes some java interop stuff much more bearable but one can certainly feel dirty using it.

15:51 chouser: it's not really dirty until you start using 'set!' to return multiple values across multiple stack frames.

15:55 Chousuke: I did that in my implementation of read-time syntax-quote

15:56 chouser: clojure.lang.Compiler does it in several places.

15:56 * chouser tries to glare disapprovingly at rhickey, but it doesn't stick.

15:57 Chousuke: I certainly hope the cinc compiler will avoid any dynamic binding

15:57 chouser: yeah, and mutable fields.

15:58 Chousuke: it's a bit difficult to make compiler information available to "outsiders" if it's all context-dependent and has dynamic scope :/

15:58 chouser: well, outsiders don't need to see it until it's done mutating ...but, yeah.

15:58 it shouldn't do that.

16:03 slyphon: does the CLR do TCO?

16:04 they just rolled out F#

16:10 anyone know what handles the channel logging/publishing in here?

16:10 kotarak: slyphon: ask chouser

16:13 * hiredman started writing a compiler (not a cinc) and is using binding and set! all over the place

16:14 chouser: slyphon: my IRC client logs everything, and then I have a clj program that converts it to html and rsyncs to the web server.

16:14 slyphon: chouser: HA!

16:14 awesome

16:14 *way* more low-tech than i expected

16:14 cemerick: The channel logs have a bus factor of 1, though. :-P

16:15 hiredman: 59M #clojure.log

16:16 dcnstrct: word on the street is that Oracle will be charging for it's JRE very soon. =^o If this were to happen would it bother you folks ?

16:16 cemerick: dcnstrct: linky?

16:16 slyphon: dcnstrct: i highly suspect that's FUD

16:16 hiredman: is that the word on the street?

16:17 I think thats the word on the "what's new" on reddit from a few days ago that didn't get enough votes to make the front page

16:17 cemerick: I doubt they're that stupid.

16:17 slyphon: anyway, isn't gcj like *nearly* there?

16:17 dcnstrct: http://www.sun.com/software/star/odf_plugin/

16:17 sexpbot: "Sun ODF Plugin for Microsoft Office"

16:17 slyphon: dcnstrct: yes, that's totally different

16:17 hiredman: gcj is garbage

16:18 well, it might be good for something

16:18 dcnstrct: according to someone @ oracle's tech support many of their previously free offerings are going to be behind a paywall

16:18 including the jre

16:18 hiredman: but it is not a general purpose jvm

16:18 dcnstrct: but this is just second hand info

16:19 slyphon: hiredman: i was being "arch"

16:19 dcnstrct: my question though is would it matter much to you guys ? are you perfectly happy with OpenJDK ?

16:20 cemerick: dcnstrct: I think we'd have to actually see something concrete supporting the premise before thinking about much.

16:20 slyphon: i've used the mac version of openjdk, it did what i needed it to do

16:20 not w/ clojure, but with jruby

16:20 dcnstrct: cemerick, it's just a conspiracy theory @ the moment, but it would explain why James Gosling resigned in protest.

16:21 slyphon: really? i think he might not have liked the culture @ oracle

16:21 i mean, he'd been employed at Sun with *that* sr. team for like how many years?

16:21 hiredman: danlarkin says if you use the openjdk you'll run into bugs, but I have used it, and haven't to my knowledge run into such

16:22 Licenser: hmm ruby makes me sad this days :(

16:22 * slyphon wishes they'd just figure out how to release memory back to the OS

16:23 hiredman: pffft

16:23 who cares, just release a jvm with tailcalls

16:23 or fixnums

16:23 well, I think I'd prefer tailcalls

16:24 Licenser: I want a jvm that just works

16:24 apple should make one

16:24 with a nice apple logo and a full metal frame

16:24 hiredman: Licenser: I have one of those

16:24 dcnstrct: yes and then they should mandate that users can only use it with the java programming language

16:24 Licenser: dcnstrct: nah I doubt it

16:25 apple prefares objective C

16:25 dcnstrct: oh ya gp

16:25 Licenser: they'll kill java and make ojective C or ruby the main language on the jvm

16:25 hiredman: as it happens the compiler I am working on is source to source and generates java source code

16:26 Licenser: hmm I strongly doubt that oracle will make the jre payware

16:26 it would kill it entirely

16:26 dcnstrct: unless their intention is to kill it

16:26 rfg: Anyone want to hazard a guess when dalvik will be "fast enough" for clojure?

16:26 *if/when

16:26 Licenser: dcnstrct: if I'm not mistaken their projects use java quite a lot right? At least the middleware sector

16:27 what I fear more is that they go to neglect java

16:27 dakrone: rfg: I thought the concern was more in dalvik's garbage collector than its speed

16:27 cemerick: Their entire app business is JVM-based. Why talk about conspiracy theories?

16:27 hiredman: rfg: needs a good gc and jit, but as far as I know google's priority is battery life

16:28 stuartsierra: I thought the problem with dalvik was startup time for Clojure RT

16:28 dcnstrct: they could still run their app business while charging for manditory support for jvm downloads

16:28 Licenser: cemerick: yea but it does not mean that they will keep the 'wishes' of the comunity in mind anything like sun

16:28 dcnstrct: but half the world would kill them for that :P

16:28 dcnstrct: hahaha

16:28 Licenser: and people would stop using the jvm as platform

16:28 which would hurt the jvm

16:29 dcnstrct: well the openjdk stuff would just get stronger

16:29 and oracle would have to spend less time thinking about the jvm

16:29 Licenser: or .net and the M thingy

16:29 dcnstrct: because it would be "pushed into the open source" more

16:30 they did the same thing to solaris.. it's free but with a mandatory $90 support contract

16:30 people still use solaris

16:31 slyphon: my whole shop is solaris

16:31 Licenser: wow they now charge for solaris? Since when

16:32 downloaded it just like 2 weeks ago

16:32 slyphon: only if you're running it on non-sun hardware

16:32 iirc

16:32 Licenser: ah okay I use it on a v440

16:32 dcnstrct: see.. they're rolling out this new policy on all kinds of stuff that used to be free

16:32 week after week

16:33 slyphon: dcnstrct: are you just trying to drive Oracle's stock price down?

16:33 dcnstrct: LOL

16:33 busted

16:33 mfex: hi all, how do I map a function over each value in a hash-map?

16:33 dcnstrct: sell damn you all

16:33 sell

16:33 hiredman: clojurebot: ticker ORCL

16:34 clojurebot: ORCL; -0.05

16:34 dcnstrct: w00t!

16:35 cemerick: ,(map inc (map val {:a 0 :b 1}))

16:35 clojurebot: (1 2)

16:35 Licenser: dcnstrct: I loved sun as a company and I'm very sad they got bought too, and oracle isn#t the same ;(

16:35 cemerick: mfex: ^^

16:35 or...

16:35 ,(map (comp inc val) {:a 0 :b 1})

16:35 clojurebot: (1 2)

16:36 dcnstrct: licenser, care to speculate about what sun's big mistake was ? why did they end up selling ?

16:36 mfex: cemerick, I want to new values bound to the same keys in a map again

16:36 cemerick: ah

16:36 Licenser: dcnstrct: because oracle had way more money

16:36 cemerick: I think there might be a map-map fn to do just that in contrib somewhere. Otherwise, you can do...

16:36 dcnstrct: $_$

16:36 sexpbot: Command not found. No entiendo lo que est?s diciendo.

16:36 mfex: cemerick, so (?? inc {:a 1 :b 2}) -> {:a 2 :b 3}

16:36 Licenser: also I think sun's buiggest mistake was that they were always briliant with new ideas but never got to an end.

16:37 Also MySql was a big compeditor for oracle

16:37 cemerick: ,(into {} (for [[k v] {:a 0 :b 1}] [k (inc v)]))

16:37 clojurebot: {:a 1, :b 2}

16:37 cemerick: mfex: ^^

16:37 hiredman: I find it hard to fathom how mysql competes with anything

16:37 mfex: cemerick, thanks

16:38 Licenser: hiredman: it might not be geniuse but it is widely used

16:38 cemerick: mfex: there is a map-map fn floating around somewhere though, that simplifies that a little

16:38 hiredman: ,fmap

16:38 clojurebot: #<MultiFn clojure.lang.MultiFn@3e6962>

16:38 dcnstrct: sqlite beats it for most of the use cases I care about

16:39 hiredman: ,`fmap

16:39 clojurebot: clojure.contrib.generic.functor/fmap

16:39 hiredman: ,(fmap inc {:a 1 :b 2})

16:39 clojurebot: {:a 2, :b 3}

16:39 cemerick: yeah, there you go :-)

16:39 hiredman: you know contrib well for someone that doesn't use it. ;-)

16:40 mfex: even better, thanks

16:42 hiredman: cemerick: well, you know, you keep an eye out, snag anything interesting that floats by

16:43 lancepantz: i saw a thread in the google group that mentioned flightcaster.com as the largest site with clojure in production

16:43 does anyone know otherwise?

16:43 cemerick: Sounds plausible.

16:43 Do they publicize figures of any kind?

16:43 technomancy: probably depends on how you measure it

16:43 LOC, cluster size, team size, etc.

16:45 slyphon: is there something in contrib that does bean but can set values too?

16:45 lancepantz: i'm not sure, but their alexa rank is 330,252

16:45 cemerick: them's some hardcore metrics! ;-)

16:45 lancepantz: makes me nervous

16:46 slyphon: "How many spyware infected nop-heads can you attract to *your* site?"

16:50 dnolen: interesting JavaScript V8 looks like it's faster than Python 3 or Ruby 1.9. Good target for Clojure?

16:51 * slyphon boggles

16:52 * dcnstrct plays with node.js + compojure

16:52 dcnstrct: it sure would be cool to have that be 100% clojure

16:52 go ahead and get started on it would ja ?

16:53 hiredman: clojurebot: scriptjure?

16:53 clojurebot: Gabh mo leithscéal?

16:53 hiredman: clojurebot: google scriptjure

16:53 clojurebot: First, out of 57 results is:

16:53 arohner&#39;s scriptjure at master - GitHub

16:53 http://github.com/arohner/scriptjure

16:54 dcnstrct: if you called it scriptjure people who write frameworks on top of it would use a biblical naming scheme

16:54 and that could get pretty ridiculous fast

16:55 hiredman: adam begat cain and cain begat seth(?) and ...

16:55 pjstadig: hehe

16:55 rfg: Ecclejiastes

16:56 pjstadig: pjalms

16:56 dcnstrct: jejus

16:58 * dcnstrct tries to get his head around what this does exactly.. but fails: http://code.google.com/p/nodb/

16:58 dcnstrct: is it plain text search for in-memory arrays ?

17:00 ahh it's that plus transactions and compression.

17:03 stuartsierra: Anybody here read the RSpec book from Pragmatic ?

17:04 Raynes: stuartsierra: No, but it's on my to-read list right after War and Peace and the Common Lisp spec.

17:05 stuartsierra: umm, not exactly a ringing endorsement

17:06 dcnstrct: I like test/spec a lot better

17:07 it does all the main stuff you'd need and it's 1/50th the code size

17:07 technomancy: my thoughts on rspec: http://technomancy.us/118

17:07 sexpbot: "in which a hypothetical death occurs in order that a real one may be avoided - Technomancy"

17:08 slyphon: technomancy: :D

17:09 rspec is proof of what happens when "the void stares back into you"

17:09 chouser: 40KLOC of ruby!?

17:09 slyphon: chouser: it ain't pretty

17:11 nurv: well, i worked for a comp that had about 1.2MLOC of lisp

17:11 not CL

17:11 chouser: nurv: elisp?

17:11 ;-)

17:12 nurv: no no

17:12 an in-house dialect

17:12 chouser: ah

17:12 slyphon: yeeeeeeeeeeeeeeeeeeuuuuwwwww!

17:12 nurv: it "mostly" worked

17:12 chouser: heh

17:12 zakwilson: Was it the source code for the OS and all included software on a Lisp machine?

17:12 nurv: that is, until it didn't

17:13 nope, although that would be great

17:13 zakwilson: Did the people who wrote most of the code actually understand Lisp?

17:13 nurv: it ran on hp-ux and some ibm 390s machines

17:13 cemerick: technomancy: LOL @ bludgeon :-D

17:14 nurv: zakwilson, it depends on the project they were working

17:14 zakwilson: bludgeon is pretty funny. It needs more possible outputs though.

17:14 nurv: there were some great coders and some awful ones too

17:14 but at least it was challenging an fun

17:15 *and

17:15 zakwilson: Perhaps *what* you could bludgeon to death with a printout of the source code. (e.g. a fly, a mouse, a whale)

17:15 technomancy: =)

17:15 zakwilson: I'd enjoy working at a company that wrote lots of Lisp.

17:15 hiredman: lisp has been many things to many people

17:16 nurv: btw, rebooting...bbl

17:16 hiredman: ~google the evolution of lisp

17:16 clojurebot: First, out of 24500 results is:

17:16 The Evolution of Lisp

17:16 http://citeseerx.ist.psu.edu/viewdoc/download?doi=

17:16 hiredman: bleh

17:16 there are free pdfs if you google

17:22 stuartsierra: technomancy: bludgeon install fails for me

17:23 technomancy: =(

17:23 yeah, looks like there's some bitrot

17:23 stuartsierra: I like the concept though

17:24 technomancy: the code is mostly there to make the joke better

17:24 mmarczyk: BLUDGEON_PAGES = 500 # TODO: is this accurate?

17:24 lol :-)

17:24 um, hi

17:25 stuartsierra: 500 pages is only about 2.5 inches thick

17:25 dcnstrct: are we talking about text on both sides ?

17:25 technomancy: I was going for a standard ream like for a photocopier

17:25 if you pack it tightly enough it could be deadly... maybe. further research is required.

17:25 stuartsierra: not sufficiently deadly, in my opinion

17:26 'cause I've tried bludgeoning the photocopier and never made a dent

17:26 technomancy: patches welcome =)

17:28 stuartsierra: well, I just bought the RSpec eBook, will report how it goes

17:28 technomancy: this is also assuming you have the advantage of surprise

17:29 stuartsierra: ah

17:29 mmarczyk: technomancy: I've read your last blog entry on submissions, very enlightening for me

17:30 guess I'll be re-sending my clojure-mode patch

17:30 so, just wanted to ask, would a git-prepared patch file be preferable to a pull request or does it not matter?

17:31 technomancy: mmarczyk: I have a slight preference for pull requests

17:31 mmarczyk: alright

17:31 I'll make a fresh fork, put my stuff on a topic branch, then send one

17:31 technomancy: for the record, I like all kinds of submissions

17:31 so don't stress over it. =)

17:32 mmarczyk: :-)

17:32 that's about the coloured repl, btw

17:32 there's also a swank bit to that

17:32 Licenser: if someone is interested I wrote togehter a little article about how to make libs nicely multy version compatible: http://blog.licenser.net/2010/04/22/on-clojure-libs-and-versions

17:32 sexpbot: "lice! : On clojure libs and versions."

17:32 technomancy: sweet

17:32 Licenser: also I fixed my blog after a day of work

17:32 doh

17:32 mmarczyk: so... would a defadvice be ok inside swank

17:33 or would that be better as an "extra recommendation" in the readme

17:33 (or do I not make sense now...)

17:33 hm... guess swank uses an advice on M-x slime anyway

17:33 MadWombat: hello

17:34 Licenser: hi MadWombat

17:34 MadWombat: How do you use nth-of-type or first-of-type in enlive?

17:34 technomancy: mmarczyk: sorry, don't have time to check the context on that. it's probably ok though.

17:34 mmarczyk: sure

17:37 slyphon: hrm

17:48 dnolen: MadWombat: you need an extra set of square brackets, [[:div (nth-of-type 2)]]

17:52 raek: Licenser: *bookmarked*

17:53 honestly, hadn't thought about that stuff

17:56 kotarak: Licenser: *daumenhoch* for the version range tip. This should be promoted more!

17:56 Licenser: thanks :) I am glad if it helpes

17:59 weöö bed time for me, night people!

18:07 livingston: so if I'm working with a swank project with lein and emacs...

18:08 if I want to work with a lot of source code, what's the preferred way to load it all in / manage that?

18:10 _ato: livingston: I use ido-mode for quickly switching between buffers, if that's what you mean

18:11 others use things like speedbar to show a tree of files

18:11 livingston: no i just mean loading all my source, when the project open although my classpath is set nothing is loaded

18:11 dnolen: livingston: I used desktop-save and desktop-read

18:12 that saves all your open buffers to a file that can be read later

18:12 no so hot for multi project setups but works well enough for me.

18:12 livingston: dnolen: oh that sounds useful (but that's another problem)

18:13 dnolen: oh, you mean loading all of your source and sending it to the repl?

18:13 _ato: livingston: if you mean loading into clojure not emacs, just have some main namespace which loads all the others?

18:13 livingston: yeah getting the code into clojure

18:14 in lisp there was a thing called asdf which managed the pacakage and file dependencies and would load and compile etc.

18:14 technomancy: livingston: in CL, you mean.

18:14 sorry, pet peeve

18:15 livingston: i would think lein or something would help me with that too, but I haven't found examples of anything like it.

18:15 technomancy: if you (use 'top.level.ns.that.requires.all.others :reload-all) it would do it

18:15 provided you have such a top-level ns

18:15 livingston: technomancy: prior to about 2 years ago lisp and common lisp were fairly interchangable (what else would you be using?), but yes I agree

18:16 technomancy: no love for MACLISP? come on guys...

18:16 =)

18:16 livingston: oh, really? that's all I have to do?

18:16 * _ato is confused about what else there would be to do

18:17 _ato: does CL make things really hard or something?

18:17 livingston: that's pretty sweet, but wait how will it know what order to load all my files in?

18:17 * technomancy would venture a guess that there's an order of magnitude more elisp in active use than CL fwiw

18:18 livingston: _ato: no it doesn't, things like ASDF make managing CL pretty straightforward

18:18 nurv: MACLISP... there is an implementation for it to the JVM too.

18:18 Didn't play much with it though.

18:19 _ato: livingston: the order you specify... if namespace A use/requires B, which in turn use/requires C and then D then that's then they'll be loaded in the order C D B A

18:19 (assuming they all do so at the top of their code anyway)

18:19 livingston: that makes sense, although... what if I have the same ns in two different files? or is that really bad practice?

18:20 slyphon: gah

18:20 nurv: technomancy|away maybe you'll enjoy this: http://github.com/hz7k-nzw/lapin

18:20 _brian2_: noob question, i am trying to use exception handler (from dnolen's enlive tutorial) http://clojure.pastebin.com/QMZeeZnq but I get "Don't know how to create ISeq from: java.lang.Boolean" when i run it

18:20 slyphon: https://gist.github.com/898a45ef78d87bc521ce

18:21 _ato: livingston: then it's only going to pick up one of them, the one that's named to match the namespace (yes it is bad practice)

18:21 slyphon: i'm trying to dynamically call a set method here, but i can't seem to get the escaping syntax correct

18:22 _ato: you can probably get around that by using load-file or some such (kind of like a C #include) but normally you should have namespaces correspond 1 to 1 to files.

18:22 livingston: oh, ok, well, I should stop that then...? but it just seems if I have a lot of code for a ns that naturally falls in a few big chucks I could split the files, but then you're saying they should then have their own subordinate ns each?

18:23 _ato: that's how I think it's intended to work, yes.

18:23 livingston: _ato: ok that makes sense. in CL we could spread it out, but then you specified each file in ASDF and their file level dependancies, so I was expecting to have to do something like that

18:24 and I guess I could roll my own equivalent of that but why bother if I get it for free with some ns disipline

18:24 dnolen: _brian2_ it just looks like your try/catch syntax is wrong right? true is outside the expr

18:25 _brian2_: dnolen> ok thnks

18:25 _ato: yep, and in my opinion by having namespaces match files then it's easier to remember where stuff is defined

18:26 livingston: well that makes it more like java than CL but that's ok, I've conceeded that that's going to happen to me as I move forward with clojure ;)

18:28 mmarczyk: actually clojure.core is split into a couple of files

18:28 and there's a bunch of (load "...") forms in core.clj

18:28 so it's perfectly possible, if not necessarily great for readability

18:29 livingston: I guess there's no real difference from writing (load foo) vs (use foo)

18:29 and the net effect is basically the same

18:29 mmarczyk: well actually there are fairly significant differences

18:30 a separate ns can be reloaded separately

18:30 technomancy: core is not always a good example to follow

18:31 mmarczyk: well that I'm entirely in agreement with :-)

18:31 livingston: you could just load one of the files though again independently, I think. but the separate ns is probably better...

18:32 although then you need a lot of "use" code everywhere to avoid having the qualify the heck out of everything

18:34 _brian2_: dnolen> I'm not sure, I'm just substituting (html/html-resource (java.net.URL. url)) into this (try

18:34 (Class/forName class-name) true

18:34 (catch ClassNotFoundException _ false)) from the book

18:34 mmarczyk: livingston: maybe see http://onclojure.com/2010/02/17/managing-namespaces/ and http://code.google.com/p/clj-nstools/

18:34 sexpbot: "Managing namespaces &laquo; On Clojure"

18:36 livingston: mmarczyk: cool thanks for the pointers

18:37 mmarczyk: sure

18:37 _ato: livingston: yep, for imports within a project I tend to just 'use all the symbols rather than listing just the ones I want explicitly as is usually recommended, since it's the one project if you introduce something that conflicts you'll know immediately when you try to run it, so it's easy to fix. It's too much hassle and repetition otherwise in my opinion. Listing symbols is definitely a good idea for between projects thoguh.

18:57 livingston: so regarding our previous discussion about namespace and files... if I have a ns foo and that breaks down into foo.bar and foo.baz but there is no actual foo code anymore, I could just create a foo ns that requires (or uses) both?

18:58 and I would put that foo ns defining code in just a foo.clj file and that's it?

18:59 there seems to be a naming convention for ".core" though, what's the deal there?

19:00 rfg: This is strange. I have some code where if it reads (first thing) it hangs the repl, but changing it to (take 1 thing) it works.

19:01 mmarczyk: I believe that a namespace foo.bar corresponds to a bar class in the foo package

19:01 and with a "single-segment" namespace you're putting your class into the default package, or whatever that's called in Java

19:02 livingston: mmarczyk: makes sense I was just trying to figure out how to define/name the foo ns based on best practice

19:03 there seems to be a "core" naming convention that I was trying to understand

19:03 mmarczyk: well, anyway, if you want a foo.bar.baz namespace, but no foo.bar namespace

19:03 you don't need a separate .clj file for foo.bar

19:03 Raynes: mmarczyk: Thanks for that correction earlier.

19:05 livingston: I know, I want to break up my code into more files though, and based on some previous advice I think I will give each file it's own ns...

19:05 mmarczyk: Raynes: which correction? you're welcome anyway, though :-)

19:05 livingston: but then I want to create a parent ns to kinda glom it all back together

19:05 Raynes: mmarczyk: I said "vars are immutable" when that isn't what I actually meant. You corrected me.

19:06 rfg: Ah, I see what my problem was.

19:09 mmarczyk: Raynes: ah, you mean in Clojure 101

19:09 Raynes: Indeed.

19:10 mmarczyk: well, I do use the same mental picture

19:11 of shadowing, I mean

19:11 for locals, which I hope came across from my posting

19:13 incidentally, I'm rather enjoying clj101 already :-)

19:17 livingston: not sure why you'd want to do that

19:18 livingston: so that then I don't have to say :use bar baz etc.. I can just use the parent package/namespace

19:18 mmarczyk: that Google Code project of Konrad Hinsen's might help you with that

19:20 livingston: how do I get clojure to tell me what it thinks the classpath is?

19:21 mmarczyk: yeah I'll probably end up using konrad's ns+

19:21 rfg: (System/getProperty "java.class.path")

19:22 Borkdude: I also saw this somewhere: (println (seq (.getURLs (java.lang.ClassLoader/getSystemClassLoader))))


19:22 mmarczyk: (filter #(.startsWith (name %) "foo.bar") (clojure.contrib.find-namespaces/find-namespaces-on-classpath))

19:23 though I suppose that's borderline ridiculous

19:23 Borkdude: what exactly is the difference?

19:23 mmarczyk: but you could map clojure.core/use over that

19:25 livingston: thanks. I just wanted to make sure it could see my class, turned out I had a typo somewhere else...

19:28 if I use a ns foo that itself uses a ns foo.bar don't I get all the symbols in foo and foo.bar?

19:29 arohner: does anyone have a function for elegantly handling functions with optional parameters in the middle

19:29 like defn

19:29 (defn-magic foo [a b? c])

19:29 technomancy: livingston: that's what ns+ is for

19:30 livingston: arohner: why can't you put them at the end?

19:30 technomancy: but no, nothing like that is in clojure yet. I have hopes that template namespaces will get included some day.

19:30 arohner: livingston: cause I want it this way :-)

19:30 livingston: technomancy: oh ok, I though I got it all for some reason since the subordinate symbols do all exist in the parent ns

19:31 arohner: then you have to just use a & rest and pull off the end ones OR

19:31 technomancy: there's not really any such thing as a parent ns. the hierarchy is an illusion.

19:32 livingston: (defn foo [a c] (foo ([a default-for-b c]) ([a b c] ....))

19:33 technomancy: right, right of course, but I was referring to foo as the parent in the context of (ns foo (use bar)) ... I thought maybe if I used foo I should get all of foo and bar

19:33 because I though the use brought bar's symbols into the foo ns.

19:34 mmarczyk: arohner: how about (defn foo ([a c] (foo a default-b c)) ([a b c] ...))?

19:34 livingston: use isn't transitive

19:35 arohner: livingston mmarczyk: yeah, I guess that'll work. thanks

19:35 mmarczyk: in fact, it most definitely shouldn't be

19:36 livingston: i wouldn't really think you would want it to be, but I thought I read it as actually creating the symbols and adding them to the "parent" ns

19:37 mmarczyk: not really

19:37 see the refer function and the refer method on ns objects

19:37 livingston: well this certainly makes package organization more of a pain.. I think I'll definately get Konrad's ns+ now

19:50 rfg: Is anyone using 1.2 with lein?

19:51 livingston: rfg: i was, but I just broke it, I think..

19:51 rfg: I tried changing to 1.2 in project.clj but running lein deps gives errors.

19:51 livingston: what errors are you seeing?

19:52 I was running that just fine but now tried to add in ns+ from here: http://code.google.com/p/clj-nstools/

19:52 sexpbot: " clj-nstools - Project Hosting on Google Code"

19:52 livingston: but it's telling me it's missing 1) org.apache.maven:super-pom:jar:2.0

19:53 technomancy: super-pom is maven's way of saying "the current project"

19:53 so it's missing one of your project's deps; read up higher in the output

19:54 livingston: Konrad's nstools needs 1.2 master and clojure contrib master , but I'm not quite sure how to figure out what to tell lein to make sure I have what he requires

19:56 rfg: livingston: http://paste.lisp.org/display/98174

19:57 livingston: rfg: that's very much like what i'm seeing, if not exactly

19:58 so if i'm staring at a github page, how can i figure out exactly what string I need to pass lein to get a given version?

20:03 _ato: rfg: 1.2 hasn't been released; try "1.2.0-master-SNAPSHOT" instead of "1.2.0"

20:04 rfg: Cheers, will give that a go.

20:04 livingston: _ato: what contrib library goes with that?

20:04 _ato: "1.2.0-SNAPSHOT" i think

20:06 livingston: that seems to work...

20:07 (except it doesn't for konrad's code I'll email him: UnsupportedOperationException: nth not supported on this type: PersistentArrayMap)

20:07 _ato: so how am I supposed to know what tags like "1.2.0-SNAPSHOT" are available for download? / figure that out?

20:08 _ato: the way I do it is look here: http://build.clojure.org/snapshots/org/clojure/ and here http://build.clojure.org/releases/org/clojure/

20:08 sexpbot: "Index of /snapshots/org/clojure/"

20:09 livingston: OH, huh

20:10 so does lein talk directly to github? or does it go there? because I was looking at tags on github, and there's one called "par" for better support of the parallelization code

20:12 _ato: lein goes there and http://clojars.org/repo/

20:12 sexpbot: "Index of /repo/"

20:12 _ato: not github

20:14 livingston: that makes sense.

20:23 well, that's unfortunate...

20:23 MadWombat: this is weird, I have an HTML page whith table based layout, there is a single table under body, that contains all the other nested tables. Yet (select page [:body :> :table]) returns 6 separate tables. Am I misunderstanding the way :> works in enlive?

20:23 livingston: I emailed Konrad about nstools, but his autoresponder told me he's out for the week

20:24 anyone have ns+ working?

20:37 dcnstrct: http://www.lisperati.com/clojure-spels/casting.html

20:37 sexpbot: "Casting SPELs with Clojure - Home -"

20:39 dcnstrct: this is not quite as psychedellic as Ruby's beginner book.. but it looks just as useful and is perhaps even easier to read.

20:40 lancepantz: referring to why's poignant guide?

20:40 dcnstrct: absolutely

20:41 back in the day it was the cartoon foxes or the half-japanese docs, or the prag book

20:42 I like this wizard book a little better than why_'s stuff

20:42 lancepantz: naa, i like the foxes

20:42 dcnstrct: well I don't dine on swine so maybe that's why_ I'm a little bias

20:42 the chunky bacon meme got a little out of control and I never quite got it.

20:43 this lisp wizard is a boozer

20:44 admittedly that's probably worse for the community than chunky bacon..

20:47 dnolen: not sure if somebody's already mentioned this here, but KirinDave's podcast is pretty cool: http://www.redmonk.com/cote/2010/04/22/makeall002/

20:48 danlarkin: pretty glorious headshot there

20:58 joshua-choi: It's a shame that Why disappeared from the Internet.

21:01 lancepantz: yeah, pretty strange

21:10 livingston: what's the best way to file a bug report on the 1.2.0-master-SNAPSHOT build?

21:10 dnolen: livingston: is this something you built from source?

21:11 hiredman: 1.2.0-master-SNAPSHOT is being constantly rebuilt on hudson

21:11 livingston: there's a bug in the ns macro

21:11 no

21:11 hiredman: that is what SNAPSHOT means

21:11 livingston: oh that makes sense

21:11 dnolen: what I meant was, where did you get that from? clojars?

21:12 livingston: um, whatever lein just gave me

21:12 hiredman: that comes from hudson

21:12 build.clojure.org

21:12 dnolen: hiredman: interesting, I did not know lein pulled from hudson.

21:13 hiredman: build.clojure.org keeps a maven repo

21:13 livingston: so where do I report the bug? assembla?

21:14 hiredman: what makes you think you have a bug?

21:16 dnolen: livingston: I think it's best to bring it up on the ML or here, have people reproduce. You may need a CA to file a bug (I forget)

21:16 livingston: this generates an exception: (ns foo "doc" {:author "livingston"})

21:17 hiredman: why wouldn't that?

21:17 mmarczyk: um, why wouldn't it?

21:17 :-)

21:17 livingston: which is odd, because you would expect code like that everywhere

21:17 hiredman: have you read the docs for ns?

21:17 livingston: the docs on ns say the args are {:arglists '([name docstring? attr-map? references*])}

21:18 mmarczyk: (ns #^{:author "livingston" :doc "a cool ns to do foo efficiently"} foo)

21:19 livingston: it's not my code, it used to work that way... the ns for KonradHinsen's nstools is defined that way

21:20 hiredman: livingston: are you sure you are on the latest SNAPSHOT build?

21:20 (works for me)

21:20 dnolen: ditto

21:20 livingston: yes. is there anyway to make the repl tell me?

21:20 hiredman: *clojure-version*

21:20 ,*clojure-version*

21:20 clojurebot: {:interim true, :major 1, :minor 1, :incremental 0, :qualifier "master"}

21:21 livingston: shoot - it says 1.1.0

21:21 hiredman: are you using lein repl?

21:21 livingston: damn

21:21 I'm using lein

21:21 but starting it from swank-clojure-project in emacs

21:22 dnolen: livingston: that should work, in your lib directory do you have multiple clojure jars ?

21:22 livingston: I thought that was supposed to be a safe way to avoid the fact that lein prefers 1.1.0

21:23 yes there are multiple there

21:23 dnolen: remove the 1.1.0 one

21:23 lein doesn't really do anything in the way of cleanup

21:23 livingston: i just ran lein clean and lein deps and it's gone...

21:24 dnolen: great, yet another thing I did not know about.

21:25 mmarczyk: livingston: which exception are you seeing?

21:25 livingston: damn - ok it's all good now

21:25 it was saying it couldn't call nth on a persistent map

21:29 this is the really bad thing about the bleeding edge - every once in a while you get really cut (that was a few hours I'd like to have back)

21:30 mmarczyk: well, I'm using the very latest checkout from GitHub right now

21:30 nothing seems broken

21:31 dnolen: livingston: luckily 1.2.0 is really close, lein is converging on it as well as clojure-contrib.

21:31 livingston: no it's all fine now - it was the 1.1.0 jar that was left laying around by lein

21:31 mmarczyk: ah, ok

21:31 livingston: why are you going from github?

21:31 mmarczyk: oh, by the way

21:31 I was going to ask

21:32 livingston: and when will the good parallelization stuff get mainlined?

21:32 mmarczyk: does the current lein HEAD pull in two versions each of ant and ant-launcher

21:32 dnolen: livingston: ?

21:32 livingston: mmarczyk: ask me? , yeah, shoot...

21:32 mmarczyk: for other people, or is it just me

21:33 livingston: I use my own jars for a play-around REPL

21:33 livingston: dnolen: ? what ?

21:34 mmarczyk: what do you mean, like your own modified one?

21:35 mmarczyk: not necessarily, just ones which I built here myself

21:35 for no particular reason

21:35 livingston: oh, well, why bother? unless you like to give bug reports back

21:35 mmarczyk: I already have the sources, I build them to see if anything breaks, then it's more convenient to just use those jars

21:35 then do a separate download

21:35 I like to file bug reports when I encounter bugs :-)

21:36 that happened once so far with c.c build process

21:36 anyway, that's just something I like to do

21:37 livingston: I don't have multiple ant anythings that I can see,... but lein grabs a ton of stuff I don't ask for, I only ask for clojure, clojure-contrib and swank... maybe if I asked for ant specifically i'd get doubles on things or smethign

21:37 mmarczyk: um, no, I mean the leiningen project itself

21:37 livingston: oh -

21:38 mmarczyk: nvm, I'll bring this up when technomancy appears

21:39 livingston: I can look .. to tell you the truth, I don't even know where lein puts all it's stuff?

21:43 mmarczyk: are you using the stable version or a snapshot from GH?

21:44 if the former, then I guess you need a launcher script and the rest sits in a jar in your local maven repo

21:45 livingston: me, for lein? it is the stable one there's the script to download and then it seems to like to manage itself, yeah

21:45 oh, it just runs itself right out of the maven repo?

21:46 mmarczyk: see vim $(which lein) if you're on *nix

21:48 livingston: huh, wouldn't have thought to do it that way myself, but hey less copies of stuff - which is surprising, given the lein general take on that otherwise

21:48 mmarczyk: some irony in that I suppose ;-)

21:49 btw, how's your unifier coming along?

21:50 livingston: it's good the unifier is just an extremely tiny piece of what I'm ramping up to

21:51 I was hoping to be further along with a minimalist RDF interface by now

21:53 thanks for your comments

21:53 mmarczyk: not at all, that was lots of fun

21:54 livingston: it's all going to end up open source so there will be a chunk of code come out regularly, probably

21:56 mmarczyk: cool

21:56 and what kind of project is it?

21:56 livingston: I do natural language understanding

21:57 actually my real interest is memory, but the NLP is in there too, since I'm interested in how memory can facilitate nlp

21:58 uberjar: I thought NLP is a way to hypontize girls into comming back to your apartment.

21:58 Neuro-linguistic programming

21:59 livingston: yeah it's odd, they can both share a lot of the same concepts too, like memory, etc. makes googling a pain

22:00 mmarczyk: livingston: in this case I'll definitely be reading your code

22:00 should be of interest to me

22:01 livingston: what do you do?

22:01 uberjar: livingston, how do you do parts of speech tagging ?

22:01 mmarczyk: I'm a logician, mostly

22:01 uberjar: do you have a big dictionary of words and their parts of speech ?

22:02 livingston: uberjar: no and yes...

22:02 so bottom up approachs to NLP take things through pipelines that almost always start with POS tagging

22:04 I use a top-down approach to language (because I believe NLP isn't bottom up) and can leverage other things like memory and context which is important to understanding everything

22:04 the approach I use is called DMAP (Direct Memory Access Parsing)

22:05 where patterns of text and memory references are mapped to other memory references -- basically.

22:05 so if the patterns use POS information we look it up and try to leverage it, but there's nothing about the approach that requires it

22:06 uberjar: interesting

22:06 lancepantz: uberjar: most words have more than one sense that have multiple possible parts of speech

22:06 uberjar: where do you get your lexicon file for when you do look it up ?

22:06 is that right word for it ?

22:06 livingston: (also I'm punting a little because I'm rewriting it from scratch so I don't have all the details for the new one yet - this is my second DMAP)

22:07 lancepantz: the "best" way to solve the problem is using an iterative rule based tagger, look up the brill tagger

22:07 the most complete lexicon is wordnet

22:07 livingston: lancepantz: yes there is a lot of ambiguity in language - it's everywhere

22:08 lancepantz: word sense disambiguation :)

22:08 livingston: which is why we focus on meaning and integrating with the underlying memory right away, since that's what it's all about anyway

22:08 if we get information from a POS tagger or whatever along the way, great

22:09 uberjar: there will be some form of lexicon you have to know what things mean

22:09 but that's kind of what the patterns are too - "if you see this input it means this"

22:10 lancepantz: i don't thinkg meaning doesn't really has any implications in nlp, what is important is quantifying conceptual similarity

22:10 that's all that meaning really is

22:10 i worked alot on measuring word similarity at a previous job

22:11 livingston: talk to the biologists I work with ... things mean things to them ;)

22:11 uberjar: the only NLP problem I've ever been up against was back in my spamming daze when I needed to build an autoresponder that could try to extract first names out of incomming messages. I ended up just trying to match every captialized word against a first name wordlist.

22:11 livingston: it's not just about words though - it's the whole utterance in context that meaning can be derrived from

22:11 even who's saying it can change the meaning

22:11 uberjar: then I would try 3 different POS taggin libraries to rule out words.. when too many made the cut

22:12 the problem was the larger my first name wordlist the more false positives

22:13 hamza: gents, is there a func anywhere in contrib that can tell me if a val is with in a given range?

22:14 mistra: hi

22:14 Clojure and Java

22:14 qbg: ,(<= 2 5 7)

22:14 clojurebot: true

22:14 livingston: aren't < and > n-ary?

22:14 lancepantz: livingston: in my case it was always trying to measure the similarity of one collection of words to another

22:14 granted, i was working on search though, so the problem may be different from yours

22:14 livingston: yeah ok, that's what I thought (re: < and >)

22:15 lancepantz: I'm actually trying to understand the text - extract information from it

22:15 slyphon: argh

22:15 lancepantz: can you share an example?

22:15 mistra: it compiles to jvm bytecode

22:15 slyphon: who here is good at macroexpansion?

22:16 * slyphon throws down the gauntlet

22:16 uberjar: ,(println "I am")

22:16 clojurebot: I am

22:16 slyphon: uberjar: :)

22:16 mistra: it compiles to jvm bytecode

22:16 livingston: lancepantz: biologist publish 2,000 journal-level quality articles a *day* they can't keep up

22:16 mistra: watchign the videos on youtube

22:16 search clojure

22:17 livingston: lancepantz: we are trying to say extract what proteins interact with what other proteins, or partisipate in a given funcational pathway, or...

22:17 lancepantz: interesting

22:18 mistra: no want...statement

22:18 have coffee and have video on subject.... = me happay

22:18 uberjar: mistra, yes.. think of java as a turtle.. clojure is what happens afer the turtle had mutagen for breakfast. It is still a turtle underneath but now it can do more.

22:18 livingston: slyphon: the best thing to do is start with an example of what you want to produce and then write code that cons it up like data

22:19 slyphon: livingston: well

22:20 sorry, one sec

22:20 mistra: ok uberjar

22:20 livingston: by the way clojurebot should have answered your question, that he is good at macroexpansion - better than all of us ;)

22:20 uberjar: he did!

22:21 mistra: uberjar you mean 100 times as slow as C?

22:22 uberjar: it's pretty fast

22:22 for something that takes out it's own trash

22:22 livingston: lol i totally missed that, very cute

22:23 uberjar: ;)

22:23 mistra: and processors seem to have moved along quite smartly in the last fifty years uberjar

22:26 slyphon: https://gist.github.com/29aaebed045617f5d595

22:27 uberjar: mistra, processors were getting smarter, but some programming technologies were getting dumber.. at least for a while

22:27 slyphon: i can't understand how to escape things properly so that at line 11, it treats the value in setter# as the sym

22:28 mistra: which ones however are smart? uberjar

22:28 slyphon: instead of taking the autogenerated symbol name literally

22:28 livingston: so you should always have an example of the input and the code you expect to be produced...

22:29 slyphon: i'm looking at the expansion

22:29 mistra: which ones however are smart? to go with modern quad, octal and other processors uberjar

22:29 uberjar: mistra, the ones with Lots of Irritating Superfluous Parentheses

22:29 mistra, functional languages.. ones that avoid mutable state

22:30 mistra: okay

22:30 so AmigaE is out

22:30 livingston: what you probably want ...

22:30 slyphon: livingston: i just updated it with the macroexpand

22:30 uberjar: mistra, you can do it in C.. it's just painful

22:31 livingston: is setter currently contains a string of the name and you acttually want the symbol, call symbol on that

22:31 * uberjar is too young to know what an Amiga is

22:31 slyphon: hmm

22:31 * slyphon tries

22:32 livingston: the macro expand is helpful, but you really want an example of just the plain code you expect your macro to produce

22:32 * slyphon nods

22:32 slyphon: No matching method found: symbol for class mbox.harpo.errors.BogusSetter

22:32 i think '.' is doing something evil

22:32 mistra: uberjar How old are you anyway?

22:32 livingston: then you start building a macro up that generates that exact code

22:32 uberjar: mistra, you want one that makes meta-programming easily

22:33 mistra, my first CPU was a 486

22:33 mistra: but is too smart not to know ars technica jeremy reimer..... amiga info

22:33 yes

22:33 language, irc bot that is realistic leading to one that is useful

22:33 I like blot

22:33 I like hEx

22:34 livingston: presuming the map is fixed set of stuff, i would have the macro write it out not produce the doseq bit - if the map isn't fixed, but set at runtime then there is very little reason for you to do that with a macro, just use a function

22:35 mistra: who's te other one....Z?

22:36 livingston: if you are going the macro route I would expect the output of your macro to be simply (do (. obj set-foo 8) (. obj set-bar 9)) so make your macro produce that list of stuff

22:36 * slyphon nods

22:36 uberjar: mistra, ask the wizard http://www.lisperati.com/clojure-spels/casting.html

22:36 sexpbot: "Casting SPELs with Clojure - Home -"

22:37 livingston: slyphon: I'd normally gladly stay and keep helping but I'm so hungry I'm shaking I have to go... good luck. if i'm on later, I'll check in

22:38 slyphon: cool, thanks anyway

22:38 good suggestions

22:38 livingston: later everyone and thanks for all the fish, I mean, help.

22:44 mmarczyk: slyphon: why not just make it a function?

22:44 slyphon: mmarczyk: i'm workin on it

22:51 mmarczyk: slyphon: actually I just realised why this perhaps should be a macro, d'oh

22:51 anyway

22:51 http://gist.github.com/376105

22:51 would this work?

22:51 slyphon: yeah, . is a special form

22:52 i could try that :)

22:52 mmarczyk: please do :-)

22:52 slyphon: yeah it's the ~@(map ) bit that i don't really know

22:54 doh

22:54 brb, baby just woke up

22:54 mmarczyk: :-)

22:55 * slyphon looks at what comp does

22:55 slyphon: oh, cool

22:56 I Did Not Know That

22:56 hrm

22:56 mmarczyk: :-)

22:56 ~@ splices whatever comes next into the sequence it occurs in

22:56 clojurebot: @ is splicing unquote

22:57 mmarczyk: thanks, clojurebot :-)

22:58 slyphon: yeah, it's kind of like, it makes sense when you use it to splat a bunch of forms that are handed to the macro, but it kind of confuses me when you're generating it *inside* the macro

22:58 hmm

22:58 Wrong number of args passed to: core$map

22:58 mmarczyk: argh

22:58 slyphon: yeah, i see it

22:58 mmarczyk: right, of course

22:58 I'll fix the gist too

22:58 slyphon: so you zip that?

22:58 mmarczyk: (done)

22:59 slyphon: :)

22:59 oh, right

22:59 map takes multiple args

22:59 (duh)

22:59 oof

22:59 Malformed member expression

22:59 mmarczyk: btw, where are the camelize and upcase functions coming from?

22:59 slyphon: oh

22:59 one sec

22:59 i ported them from rails, basically

23:00 mmarczyk: oh, ok

23:00 ok, I see one more thing to fix

23:01 just a sec

23:01 slyphon: http://gist.github.com/376119

23:01 that's the other bits

23:02 mmarczyk: ^^

23:02 oh

23:02 gah

23:02 mmarczyk: make that two things

23:03 done now, updated the gist

23:03 should work

23:03 (macroexpand-1 '(update-bean *bs*{:foo-bar 7 :byte-property 8}))

23:03 (clojure.core/doto *bs* (. setfoo-bar 7) (. setbyte-property 8))


23:03 (with my stand-ins for as-str, camelize, upcase)

23:03 slyphon: :)

23:03 http://gist.github.com/376119

23:04 mmarczyk: cool, thanks! :-)

23:04 slyphon: sure

23:04 now i just have to put *your* code at the bottom

23:06 mmarczyk: it works!

23:06 hooray!

23:08 mmarczyk: :-)))

23:13 slyphon: mmarczyk: ok, now to figure out what the hell this is doing :)

23:15 mmarczyk: yup, that would be the next step ;-)

23:16 oh, only I might mean it in a different way, since this has just reminded me how I don't really get the Java beans stuff

23:16 slyphon: hah

23:17 yeah, i just wanna learn why this works and the thing i did didn't

23:17 brb

23:29 mmarczyk: (added a slightly different version to the gist, moving the map outside of ~@; doesn't matter much)

Logging service provided by n01se.net