#clojure log - Oct 12 2008

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

7:47 Lau_of_DK: Can somebody enlighten me on the syntax of Cond?

7:57 hoeck: Lau_of_DK: (cond condition1 expression1 conditionN expressionN ..)

7:58 Lau_of_DK: hoeck, yes, I realised that if one condition is true, other true values are ignored. That just surprised me. I was looking for something like (cond (> 2 1) ((do 1) (do 2) (do 3)) (> 2 0) ((do 1) ... (do n)))

8:02 hoeck: then use multiple ifs, cond is really just a replacement for else-nested ifs

8:03 Lau_of_DK: k

8:03 Do we have something like (if (isin list1 list2)) that checks for replicates in either list?

8:05 hoeck: no, but we have sets, so just use (set/intersection (set list1) (set list2))

8:05 Pupeno: Is there a downloadable version of the documentationz

8:05 Z

8:06 ?

8:06 Chouser: http://clojure.googlegroups.com/web/manual.pdf

8:12 Lau_of_DK: hoeck, first Im using a sequence of LazyCons with integer in it, secondly, my Clojure doesnt have set/intersection, that something new?

8:13 Pupeno: Chouser: how is that generated, do you know?

8:17 hoeck: Lau_of_DK: its the set.clj which comes with clojure and the correct name is clojure.set/intersection, sorry

8:18 hans: Pupeno: custom css, browser printing to acrobat. search the group archive, i posted the css

8:18 Lau_of_DK: no worries, I got it

8:52 Is there an alternative to pop, which drops the first item in a list, but that does not throw an exception if its an empty list?

8:58 Chouser: "rest", depending on the data structure

8:59 Lau_of_DK: k, thanks

11:11 Fanda: hello all!

11:12 rhickey: hi

11:12 Fanda: i have couple questions

11:12 one is about contributing

11:12 how do you contribute to clojure-contrib?

11:12 rhickey: I sent you a letter (agreement)

11:13 rhickey: who is Fanda?

11:13 Fanda: Frantisek Sodomka

11:14 my usual way is to put code on my website:

11:14 http://intricatevisions.com/index.cgi?page=clojure

11:14 rhickey: ok, I've got your CA

11:14 Fanda: good :-)

11:15 my question is, how code gets into clojure-contrib

11:15 rhickey: first step would be to submit patches for contrib on the Clojure group, the contrib owners will include them if they fit in

11:15 After I get a feel for your contributions I might make you a developer on the SF contrib project

11:16 if you have a whole new lib for contrib, again post to group, I'll look at it and decide

11:16 Fanda: ok, i see

11:16 i like to contribute anything that I find useful for myself

11:16 rhickey: what did you want to contribute?

11:18 Fanda: just a second, i will find it...

11:18 http://groups.google.com/group/clojure/browse_frm/thread/d15f190f09c0eead/bb705e2e2d897349?lnk=gst&q=aspect#bb705e2e2d897349

11:19 "code sections", which were rewritten Stephen C. Gilardi as "aspects"

11:20 i would put it into some 'util' namespace

11:20 similar goes for http://groups.google.com/group/clojure/browse_frm/thread/d7e82974b2b58d0a/b8f1d7adeba0dfda?lnk=gst&q=time-expr#b8f1d7adeba0dfda

11:21 rhickey: ok, I remember, sorry I didn't chime in then. Are you familiar with CL #+ and #- ?

11:21 Fanda: no, I don't really have experience with CL

11:21 what #+ and #- do?

11:23 rhickey: http://www.lispworks.com/documentation/HyperSpec/Body/02_dhq.htm

11:23 http://www.lispworks.com/documentation/HyperSpec/Body/24_abaa.htm

11:27 better: http://www.lispworks.com/documentation/HyperSpec/Body/24_aba.htm

11:29 Fanda: yes, (aspect :debug (do-something)) would be (do #+debug (do-something))

11:30 rhickey: a big difference is #+/- are reader macros that conditionally read

11:32 Fanda_: sorry, my connection isn't too stable

11:32 my last sentence was "since # is special for Clojure, it would need to be implemented in the core, right?"

11:33 rhickey: yes, reader macros have to go in LispReader

11:34 Fanda_: it is just a suggestion... i like both versions

11:34 i am sure that I will just it when developing, since it is a common pattern

11:34 rhickey: something like #+ will end up in Clojure

11:37 Fanda_: can (do #+debug (do-something)) be somehow rewritten without 'do'?

11:37 (#+debug (do-something)) ???

11:38 rhickey: #+ turns into whitespace if false

11:39 Fanda_: so #+debug (do-something) would be fine

11:39 rhickey: would just disappear if not debug

11:40 Fanda_: yeah, sounds good

11:40 sometimes i need to define different levels of debugging - debug-calculation, debug-writing-to-file, debug-ui, ...

11:41 or logging - log-xyz, ...

11:41 ... so that's where it started :-)

11:42 rhickey: logging is different - vars help there

11:42 (binding [*log-level* :verbose] ...)

11:43 Fanda_: oh, i see

11:43 that makes sence

11:43 sense

11:43 kotarak: If I understand correctly, this is compile time. So it's like good olde #ifdef?

11:44 Fanda_: yes, exactly

11:44 rhickey: read-time

11:44 compiler never sees it

11:44 kotarak: Ok.

11:44 Fanda_: preprocessor-time :-)

11:45 kotarak: as in C. #ifdef is also not seen by the compiler.

11:46 Fanda_: .......

11:46 other thing - compiler now has *warn-on-reflection*, right?

11:46 i would put it into map

11:46 something like *compiler-settings*

11:47 and do (def *compiler-settings* {:warn-on-reflection true})

11:47 so that we can easily add more compiler options later...

11:48 kotarak: I'm not sure, I like this #+ thing....

11:49 Fanda_: syntax of it or the behavior (semantics)?

11:49 (still here)

11:49 kotarak: the behaviour

11:51 Fanda_: what don't you like? i find it very useful

11:53 kotarak: If there something, that can be done with it, which cannot be done with "normal" macros?

11:53 /If/Is/

11:53 Fanda_: it is up to Rich - if he wants it in the core or not

11:53 i am ok with macro

11:54 kotarak: I didn't say it shouldn't go to Clojure core. I don't have to use it after all. It is up to Rich.

11:57 Fanda_: ok

11:57 other thing: should boot.clj be renamed to clojure.clj??? since it is a 'clojure' namespace? :-)

11:58 rhickey: kotarak: normal macros can't disappear

11:58 Fanda_: will rename soon

11:58 Fanda_: :-)

11:58 and what about testing Clojure?

11:59 i would create some test-clojure.clj, test-set.clj, ...

11:59 to test its correctness

11:59 rhickey: Fanda_: go for it!

11:59 kotarak: rhickey: I see stuff like (if (something) #+ (x) (y)) and people asking, why does my code not work? A macro can always expand to nil...

12:00 Fanda_: rhickey: I actually might - it would be a nice exercise

12:00 rhickey: once it gets started, other people will join and it can be done very quickly

12:02 rhickey: kotarak: in source, nil is not the same as whitespace

12:05 duck1123: what does it mean if I get an InvocationTargetException when trying to construct an object?

12:07 kotarak: rhickey: I still don't get the use. It's like old #ifdef. I don't see why I would that.

12:08 rhickey: duck1123: do you have an example?

12:10 lisppaste8: duck1123 pasted "exception when creating object" at http://paste.lisp.org/display/68368

12:11 Fanda_: kotarak: look in thread http://groups.google.com/group/clojure/browse_frm/thread/ce795dd1fb646df1/d1c6c3b3f6485ece

12:11 kotarak: download http://groups.google.com/group/clojure/attach/394c5334d6d7e2a1/cfa.clj.zip?part=4

12:11 kotarak: see lines like ;(println (list r g b a));;;;DEBUG

12:12 ;(try

12:12 ; (println (list 'push cnt (.. Thread (currentThread) (getName))))

12:12 ; (println stack)

12:12 ;(catch Exception e (println (list 'exception at c cnt))))

12:13 rhickey: duck1123: you might be missing other needed jars - also try putting them in your real classpath vs add-classpath

12:13 Fanda_: kotarak: all these lines could be put into #+debug (do ...)

12:14 kotarak: #+test (do ...)

12:15 rhickey: where does symbol for #+ come from? is it from global namespace? it might cause symbol pollution...

12:15 rhickey: aspect macro has (def +aspects+ #{:debug :log0 :log1}) which looks better to me

12:17 kotarak: Fanda_: And all these lines can be put into macro which conditionally evaluates to nil at compile-time.... But have to go now. Bye.

12:17 rhickey: CL isn't perfect, but you need to fully understand it before you attempt to better it

12:18 Fanda_: bye, kotarak

12:19 rhickey: Yeah, no offense. I have no clue about CL.

12:19 rhickey: more to kotarak's point, there are cases where generating nil is insufficient

12:20 Fanda_: yes, I believe in that

12:20 rhickey: #+it-takes-another-arg that-arg

12:22 Fanda_: could you elaborate more, please?

12:24 karmazilla: is there a literal syntax for BigIntegers, like 3M is a BigDecimal?

12:24 rhickey: #+ can deal with platform differences, API versions etc. Imagine an api has added an argument in a new version. Code that needs to work with both version needs to supply that arg only if it is the new version, if not, it can't pass nil

12:26 karmazilla: no - you want to create a small BigInteger?

12:27 karmazilla: well.... yes. it just happens to be the only class that have the isProbablyPrime method, that I need for a project euler problem

12:28 Fanda_: rhickey: i am still wondering about grouping symbols for #+

12:28 rhickey: maybe just put it into namespace!

12:29 rhickey: reader-options/it-takes-another-arg

12:29 rhickey: karmazilla: (BigInteger/valueOf 123)

12:32 Fanda_: rhickey: after all - symbols for #+ will be automatically only in the namespace, where I write the code, so I will create symbols only in certain namespaces

12:32 rhickey: good enough

12:35 thanks for chatting all!

12:35 have to go

12:53 achim_p_: hi everyone

12:53 rhickey: i like the new #= reader macro and print emitting constructor calls.

12:54 i assume there's no easy way to determine whether a java class can take its toString representation as a ctor arg and make sense out of it, or is there?

12:55 kotarak: back with a problem: did anyone experience problems with gen-interface? The .class file is generated and in the classpath, but using/importing the interface throws a NoClassDefFoundError: <vector: - 1 items>.

12:55 rhickey: achim_p_: there isn't, but the default presumes that might work

12:56 duck1123: is proxy what I need to be using if I need to create a new class to pass to a function? (like this http://jena.sourceforge.net/ARP/standalone.html#sample)

12:57 danlarkin: duck1123: proxy creates an instance of a class

12:57 achim_p_: rhickey: on my machine a printed j.u.Date won't read for some reason, while it appears to work for other people - did you consider falling back to java's serialization for java classes without dedicated print method?

12:58 sth like (RT/resurrect \? \? \

12:58 it doesn't look humane at all, but might make serialization via print more generally applicable ...

12:58 rhickey: achim_p_: how does a date print for you?

12:59 achim_p_: I'd really prefer text serialization

12:59 achim_p_: like that: #=(java.util.Date. "Sun Oct 12 18:56:32 CEST 2008")

12:59 localization issues, i assume

13:00 rhickey: achim_p_: that's really lame

13:01 well, it may end up that Date needs a print-method

13:01 achim_p_: but it *might* end up that lots of classes need a print method ;)

13:02 rhickey: achim_p_: that's true

13:09 Chouser: kotarak: how are you trying to use gen-interface?

13:12 kotarak: Chouser: I did (gen-and-save-interface "some-dir" 'full.qualified.Name ['bla [] Object] ['blub [] Object]). It generated a class file.

13:12 Then using the interface somehow fails, although the file is in the classpath.

13:15 Chouser: using it in what way?

13:16 kotarak: (proxy [Name] [] ....) , or via import. Neither works (it blows the import, or later on the proxy, when using a fully qualified name)

13:16 Chouser: hm...

13:20 ok, I'm seeing that, but I don't understand -- I can use introspection on the interface

13:23 kotarak: Shall I try something on my side?

13:26 Chouser: no, I've got proxy failing for me here, I just don't understand how, since the introspection works.

13:27 kotarak: ok

13:36 achim_p_: lisppaste8: url

13:36 lisppaste8: To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.

13:37 achim_p pasted "printing dates" at http://paste.lisp.org/display/68377

13:37 Lau_of_DK: I'm working on a (supposedly) simple decimal expansion routine, that would expand something like 1/7 to 1.142857. It should expand further than that because the sequence then repeats itself, and it should not expand to anything shorter, because that would be inaccurate. Does anybody here have a few minutes to help out?

13:41 achim_p_: Lau_of_DK: (double 1/7)?

13:42 Lau_of_DK: And you can see, that expands further than I need

13:42 I need it to stop as soon as the infinite cycle has been identified

13:42 Here's a couple of ideas on how to attack it: http://www.bestinclass.dk/download/fraction.clj

13:42 sadly, neither work, thought I think option #2 is closer to working

13:44 (just updated)

14:05 Chouser: kotarak left.

14:06 achim_p_: Lau_of_DK: mmh, what's your strategy here?

14:06 i'd first try to obtain a sequence of the digits and then run some cycle-detection algo on it

14:07 Lau_of_DK: strategy is this. if x > a, then a*10 goes into recursion. if not, then the first digit of (x / a) is the next digit in my result, and (rem a x) goes into recursion. This breaks when (rem a x) gives a result, which has already been given once before

14:07 achim_p, that not possible, sometimes I will need to dig deeper than 600 digits

14:09 duck1123: aren't all iterators seq-able?

14:09 if not, how would I get a seq from one

14:10 Chouser: duck1123: (seq iterator) ?

14:11 duck1123: Don't know how to create ISeq from: StmtIteratorImpl

14:11 Chouser: what's a StmtIteratorImpl?

14:12 duck1123: don't know about the impl part, but http://jena.sourceforge.net/javadoc/com/hp/hpl/jena/rdf/model/StmtIterator.html

14:13 got it from this http://jena.sourceforge.net/javadoc/com/hp/hpl/jena/rdf/model/Model.html#listStatements()

14:18 Chouser: Hm, seq need it to be Iterable, and oddly Iterator is not?

14:18 rhickey: duck1123: (iterator-seq iter)

14:20 duck1123: that works

14:21 gnuvince_: rhickey: do the profits for the "State: you're doing it wrong" t-shirt go to you?

14:22 achim_p_: Lau_of_DK: i might be a bit slow on the uptake atm, but what's the idea behind this approach?

14:23 Lau_of_DK: No nothing special, its just imo the simplest approach to computing all the decimals of a decimal expansion, one by one

14:23 achim_p_: i'd suggest floyd's cycle detection on the sequence of digits, which you can generate as you go

14:23 rhickey: gnuvince_: what shirt?

14:24 achim_p_: it's a one pass thing, so it should be reasonably fast

14:24 Lau_of_DK: http://en.wikipedia.org/wiki/Cycle_detection

14:24 gnuvince_: http://www.zazzle.com/state_youre_doing_it_wrong_black_shirt-235150036182618712

14:24 Chouser: gnuvince_: the zazzle clojure account is mine currently, and I have so far spent every penny

14:24 ...of the $0 it has raised.

14:25 gnuvince_: ok

14:25 Lau_of_DK: achim_p_, but its still overkill. If you take the first solution I posted, the bulky one, then it actually generates the sequence justs as its supposed to. All I need to implement is something that abort the whole show, if (rem a x) equals something thats already in list 'modulus'

14:25 gnuvince_: Chouser: upvote to get more views: http://www.reddit.com/r/programming/comments/76p6s/clojure_tshirt_state_youre_doing_it_wrong/

14:25 Chouser: gnuvince_: The profits will be passed along to Rich.

14:29 anyn: hi

14:31 I'm a newb, so I'm looking for some basic help. I'm playing around with the java-gnome bindings and, when utilizing Gtk, one must first 'initialize' gtk. In the java-gnome examples, they do this:

14:31 public static void main(String[] args) { Gtk.init(args); ...

14:31 achim_p_: Lau_of_DK: modulus contains digits, right? how do you handle recurring sequences like 12131213?

14:32 anyn: I'm guessing that, in those examples (I've ran them), args is just an empty String Array

14:32 Lau_of_DK: I handle them individually. The solution would probably lie in making this work according to my will (= true (some #(= % last) (rest modulus))) and it almost works

14:33 user=> (fraction 7.0)

14:33 (1 4 2 8 5 7 1 4 2 8 5 7)

14:33 this is a good expansion, it just runs twice as long as its supposed to

14:34 the corresponding sequence of modulus is

14:34 (fraction 7.0)

14:34 (3 2 6 4 5 1 3 2 6 4)

14:34 and at the 2.nd '3', it was supposed to break, that would have given the correct result

14:34 anyn: and the trouble I'm having is in feeding an empty String Array into init

14:35 gnuvince_: Did you guys get a chance to read my first blog post on Clojure? http://www.reddit.com/r/programming/comments/76ofb/cribbage_points_counter_in_clojure/

14:37 anyn: gnuvince_: talking to me?

14:37 Chouser: anyn: (make-array String 0) is a zero-length array of String

14:38 anyn: Chouser: ah, ok.. I was trying (make-array String 1) and it crashed the gui-Repl :)

14:38 Chouser: that would be the right type, but with a null that apparently made gtk mad.

14:39 gnuvince_: anyn: I was asking in general. I'd like to get feedback for that code.

14:39 (especially the count-straight function which makes me cringe)

14:41 achim_p_: Lau_of_DK: (fraction 7.0) returns about 500 digits for me

14:41 Lau_of_DK: yes, but notice the commented out line in the beginning of the lop

14:41 (cond (or (> count 10)

14:41 (= true (some #(= % last) (rest modulus))))

14:41 if you use this as the circuit-breaker, then you'll only get a few

14:45 According to my head, this:

14:45 (= true (some #(= % last) (rest modulus)))

14:45 anyn: Chouser: that worked

14:45 Lau_of_DK: means, if the value of the var "last" is found anywhere in "modulus" except for the first value, then break, right?

14:57 Maybe I'm imaging, but doesn't clojure have something along the lines of (but-last list)

14:57 something that gives me everything, except the last item, and does not throw exceptions

14:58 fyuryu: Lau_of_DK: butlast

14:58 Lau_of_DK: great

14:58 exactly what I needed.

14:59 achim_p_, stand down, and thanks alot for looking into it, its working now

14:59 duck1123: fyuryu: is that lazy?

15:00 achim_p_: Lau_of_DK: cool. i hate number theory, but at least now i understand that why you're safe to assume that each element in the sequence of recurring digits is unique. :)

15:01 Lau_of_DK: oh, I sent you out on a mathematical hike?

15:02 achim_p_: correction: each remainder must be unique. yeah, enough hiking for today.

15:02 fyuryu: duck1123: no

15:03 Lau_of_DK: user=> (fraction 7.0)

15:03 "1.1428571"

15:03 user=> (/ 1 7.0)

15:03 0.14285714285714285

15:03 argh, correction

15:04 user=> (fraction 7.0)

15:04 "0.1428571"

15:04 user=> (/ 1 7.0)

15:04 0.14285714285714285

15:04 there, now fraction yields the infinite sequence, which is accurate

15:19 anyn: Trying to tie some events into the gui here, in java-gnome bindings. How they do it, as they describe, "you need to hook up a "signal handler" to deal with the 'clicked' signal" ... "Button.Clicked is a Java interface. You create an instance of it to do what you want when the callback happens, and thenn pass it to the Button's connect() method."

15:19 http://java-gnome.sourceforge.net/4.0/doc/examples/button/ExamplePressMe.html

15:20 So, I'm trying to figure out how to translate that into clojure

15:20 I've got something like...

15:21 (.connect b (Button. Clicked ... hmm

15:21 or would it be (.connect b (.Clicked Button.

15:22 rhickey: Button$Clicked

15:22 anyn: and then I need to implement the method, as would be done inside java's {}, after the new Button.Clicked()

15:22 rhickey: $? is that new?

15:22 rhickey: anyn: no, that's the JVM name for a nested class

15:23 then you'll need to use proxy

15:23 http://clojure.org/jvm_hosted

15:23 anyn: ok, something analogous to the celcius example

15:25 Lau_of_DK: anyn, I got this from blackdog, it makes things easier:

15:25 (defmacro onClick [obj & body]

15:25 `(. ~obj addActionListener

15:25 (proxy [ActionListener] []

15:25 (~'actionPerformed [evt#]

15:25 ~@body))))

15:28 anyn: Lau_of_DK: I'm not quite there yet. still learning defs.. but once I get the hang of it, I'll try to use that.. I'm using the java-gnome binginds too

15:29 Lau_of_DK: k

15:36 anyn: in "new Button.Clicked()" .. is Clicked() a regular method or a constructor for a class called DeleteEvent? Trying to use proxy.. would I use (.connect b (proxy [Button$Clicked] (onClicked [source] (print "clicked")))) ?

15:36 or (proxy [Button] [Clicked] (onClicked ...

15:37 or (proxy [.Clicked Button] (onClicked

15:39 rhickey: (proxy [Button$Clicked] [] (onClicked ...

15:41 anyn: ok, I'll give that a shot

15:44 * hoeck commented on gnuvince_'s clojure blogpost

15:48 anyn: Unable to resolve symbol: Button$Clicked in this context

15:49 i'm running a fairly recent version of clojure

15:53 hoeck: anyn: have you imported Button$Clicked?

15:54 anyn: or does `Button' on the repl eval to sth useful?

15:57 anyn: hmm. the java version doesn't import it.. but I'll try

15:57 rhickey: anyn: you need to import it

15:58 anyn: ah.. my bad.. I've never noticed that $ usage yet in clojure

16:03 it's working slightly

16:08 lisppaste8: karmazilla pasted "partial reduce patch?" at http://paste.lisp.org/display/68385

16:09 * karmazilla thinks that would be useful

16:11 rhickey: karmazilla: interesting, but begs the question about currying map, filter etc similarly

16:12 karmazilla: yes it does... should I?

16:12 rhickey: not right now, let me think about it

16:13 Hun: one could also add autocurrying for 2+-adic functions.. not sure if it fits in the picture though

16:13 rhickey: that sort of currying is not idiomatic Clojure - people don't expect those functions to return fns

16:14 but many of the fns are set up for partial application, and could curry

16:15 could you post on the group so we can get some feedback from more users?

16:15 karmazilla: sure

16:15 achim_p_: what's the recommended way of providing a print-method for something that already has one (Class, in this case)? switching to the clojure ns and redefining feels wrong.

16:16 rhickey: achim_p_: why do you want to do that?

16:24 achim_p_: rhickey: hm, you're right, it hardly makes sense. different print behaviour (with serialization) is probably better implemented separately

16:24 rhickey: achim_p_: you can always build your own multimethod where the default calls print-method

16:25 gnuvince_: rhickey: are you considering adding currying to clojure?

16:26 rhickey: gnuvince_: no, karmazilla was suggesting some fns like reduce could curry themselves

16:27 gnuvince_: OK

16:28 Not sure that currying is very "Lispy"

16:28 rhickey: I agree

16:29 currying is a dubious shorthand, because it requires reader of code to know the arity in order to distinguish a curry from a call

16:30 karmazilla: I suppose it is kind of foreign to any language that allow callables to have more than one signature

16:31 rhickey: it's also weak in being arg-order dependent, thus flip etc

16:31 gnuvince_: I really like Haskell, but sometimes I think that currying makes some code harder to grasp than it would be otherwise

16:31 rhickey: exactly +1 for write, -1 for reader

16:31 gnuvince_: Especially when people go crazy with currying and function composition.

16:31 rhickey: writer

16:32 I like Clojure's #(), can replace any arg or args easily, stands out that you are doing a partial application and returning a fn

16:32 and fairly concise

16:32 karmazilla: haskell takes the consept to the extreme. I'm only suggesting this for a select few functions

16:34 rhickey: karmazilla: I understand, and I think if it was very uniform in the libs, i.e. every fn that takes a fn + args can be curried given fn, then maybe

16:35 otoh I don't want people to presume there is general currying

16:40 karmazilla: man, language design is hard :)

16:40 gnuvince_: It is

16:42 anyn: when I run (.main Gtk), the repl hangs until I close the gui. I'd like to be able to interactively change the gui while it's running. How does one do that?

16:42 rhickey: anyn: can you run the GUI in another thread?

16:45 anyn: yea, would I use a ref or an agent?

16:46 rhickey: an agent

16:46 or just Thread

16:46 Clojure fns are Runnable

16:47 anyn: k

16:49 karmazilla: that reminds me, BGGA closures can do this trick with single-method interfaces. Can Clojure do that too?

16:50 Chouser: gnuvince_: "for" can walk over more than seq, in which case they act like they're nested. "powerset" could be written with a single "for"

16:50 rhickey: karmazilla: no sugar for that yet

17:03 lisppaste8: achim_p_ pasted "patch: ctor exceptions with class names" at http://paste.lisp.org/display/68389

17:03 * anyn got it working

17:04 rhickey: achim_p_: please post patches to group - I can't take patches from pastes, thanks

17:35 rev 1065 has first steps toward AOT compilation - constants now read from strings in classfiles vs. hand-off from classloader, looking for testers

18:02 karmazilla: it ain't much but my four project euler scripts all seem to work with 1065 :p

18:05 rhickey: karmazilla: thanks, I expect most normal stuff to work, but macros that emit tricky constant values might need some print-methods

18:24 achim_p_: rhickey: ok, i thought such a minor change wasn't worth bothering the mailing list, but i understand there are probably license issues. i just posted it.

18:30 mhinchey: rhickey: slime fails to start with svn 1065

18:53 achim_p_: require seems broken

18:54 when requiring a file that's empty except for a namespace declaration, an exception is thrown: java.lang.ExceptionInInitializerError (NO_SOURCE_FILE:0)

18:54 Caused by: java.lang.IllegalStateException: var: clojure/*loaded-libs* is not public

19:15 cads: hello

19:16 I just watched the clojure video I was impressed

19:18 I've been reading about concurrency and wanting to try it... also been interested in lisp

19:19 now I read about clojure and it's very interesting as a lisp with powerful concurrency primitives

19:20 but I don't know whether it will be too hard for me to learn with my schedule

19:21 there is another concurrency-strong language that I've been thinking about, and that is erlang

19:22 and I was wondering if I could get your opinions about which of the languages would be more approachable for someone with limited background in writing concurrent code

19:26 one thing I was wondering about is whether clojure being a lisp makes learning it particularly difficut

19:34 gnuvince_: cads: depends on your background.

19:50 rhickey: mhinchey: what's the error from slime with svn 1065?

19:51 mhinchey: user=> java.lang.ExceptionInInitializerError (core.clj:0)

19:51 user=> java.lang.Exception: No such var: swank/ignore-protocol-version (NO_SOURCE_FILE:5)

19:51 user=> java.lang.Exception: No such var: swank/start-server (NO_SOURCE_FILE:7)

19:53 that's probably not too helpful. I'll try to debug it more to find out.

19:53 cads: I've mostly played around in ruby haskell, in ruby where I loved the friendliness but it didn't seem like it would be simple to do concurrency, and I wanted to be able to redefine some syntax and could not, and in haskell where I loved recursion and the power of algebraic datatypes but where again the level of abstraction needed to do concurrency was too complicated for me to do on my own right away.

19:53 achim_p_: rhickey: a (require 'anything) triggers a Compiler.resolveIn(current-ns, clojure/*loaded-libs*), which fails because it's not public.

19:54 that didn't happen in r1064

19:55 i think that's the same problem mhinchey describes

19:57 mhinchey: yes, same thing

20:12 arbscht: cads: that clojure is a lisp does not make it especially difficult

20:17 spacebat_: cads: erlang is hell bent on concurrency, while other aspects of the language leave much to be desired

20:17 but its process based concurrency not threading yes?

20:18 I think clojure has much more sound fundamentals, flexibility of expression with easy and correct thread based concurrency

20:19 and someday it will probably have the multiprocess/multiserver concurrency of erlang via some library or other

20:48 rhickey: mhinchey: looking at it now, thanks

21:00 rev 1066 should help with require error

21:19 cads: thanks for the input you guys

21:20 rhickey: cads: did you get any? I was of for an hour

21:20 off

21:24 mhinchey: rhickey: thanks, slime starts now

21:25 rhickey: mhinchey: great!

21:34 cads: rhickey, yeah, I'm going to go ahead and try messing around with it

21:35 I think I'll miss the smalltalk-esque encapsulation and object oriented rapid prototyping from ruby, at least till I learn the idioms

21:36 put to get to use java's libraries for graphics and what have you... without having to touch java... seems like a big plus

21:37 rhickey: cads: you should find a lot of people from similar backgrounds both here and on the group to help out with the paradigm shifts

21:38 Chouser: ruby was my favorite language to play in before Clojure.

21:38 rhickey: Chouser: do you miss he OO?

21:38 the

21:39 Chouser: haven't yet.

21:39 I fumble a bit with design questions

21:40 as in, I knew OO well enough that most problems have "obvious" design solutions for me

21:42 I'm not there yet with Clojure -- I'm afraid I'm making poor choices of when to use a multimethod, how to structure my data, etc.

21:43 In practice, the couple times I've chosen poorly, the "refactoring" has been pretty shallow and easy to deal with.

21:46 rhickey: I had become so tired of "new class per idea", had already gotten used to using generic data structures even when in C#/Java

21:47 cads: ruby seems to make it a bit less painless than java

21:48 Chouser: hm, I was not tired of "new class per idea" in ruby. I saw that as an upgrade from the generic data structures of perl.

21:48 rhickey: heh

21:48 so isn't Clojure similar in that way?

21:49 Chouser: I loved the ruby generator/iterator stuff, but of course that's a sort of weak echo of macros.

21:50 cads: I've been wondering if that isn't that case, too

21:51 rubyists seem extremely averse to any kind of macro system

21:52 Chouser: that's an interesting point. hadn't thought of it that way. Maybe the difference is that I thought I was supposed to be using objects.

21:53 cads: even though ruby it pretty good at defining a domain specific language using its block closures, sometimes the syntax is unwieldly and I wish I could just define a construct

21:53 Chouser: perl didn't have multimethods and multi-taxonomy inheritence to help manage the generic data structures.

21:54 mhinchey: rhichey: 1066 causes this exception in my code, still trying to track down the cause: Caused by: java.lang.IllegalArgumentException: No matching ctor found for class clojure.lang.Namespace at clojure.lang.Reflector.invokeConstructor(Reflector.java:136)

21:54 rhickey: mhinchey: do you have a macro that embeds a namespace object in the expansion?

21:55 mhinchey: yes

21:55 rhickey: that's it

21:56 cads: chouser, was clojure your first introduction to a lisp?

21:56 rhickey: mhinchey: you can try defining a print-method for Namspace that prints to a #= call to Namespace/find

21:57 back in a bit

21:57 Chouser: cads: I had made a couple solid runs at CL, but never got comfortable with the parens, and felt betrayed by the lackluster libraries.

22:00 Since much of my "fun" little programs need to plug into other things -- web site, files, directories, GUI toolkits -- I felt like CL was supposed to be this great powerful thing, but it was lots of work to do anything useful.

23:00 rhickey: mhinchey: I've added a print-method for namespaces, svn rev 1067

23:00 mhinchey: thanks, I'll check it later

23:06 spacebat_: I've learned a bit of CLOS now, and I wonder if anything like that exists in clojure

23:07 or do we just have structs and generic methods, and leave the OOP to java

23:08 rhickey: Clojure has multimethods and ad hoc hierarchies and metadata

23:08 mix and match to make as much OO as the problem requires

23:08 spacebat_: type hierarchies, like structs that inherit?

23:08 anyn: gtk wants to return a zero at the end of a method. How do I do that?

23:08 rhickey: (derive :square :rect)

23:08 spacebat_: ok

23:08 that sounds sane

23:09 anyn: erm, I mean false

23:09 rhickey: user=> (derive ::square ::rect)

23:09 nil

23:09 user=> (isa? ::square ::rect)

23:09 true

23:09 anyn: "return false)

23:09 s/)/"

23:09 Chouser: anyn: (fn [] ..... (false))

23:09 gah

23:09 rhickey: so you can use ns-qualified keywords as 'types', but put them in map attributes, metadata etc

23:10 Chouser: anyn: (fn [] ..... false)

23:10 anyn: ah, ok

23:10 spacebat: rhickey: I noticed a bunch of projects like dotLisp and foil that you've involved with, are they live/active or more like stepping stones on your way to clojure?

23:10 anyn: perfect

23:11 rhickey: spacebat: you can also 'derive' Java classes from keywords, to superimpose additional taxonomies on classes you didn't write

23:11 spacebat: jFli has seen some enhancement from Nick Levine, the others are static

23:12 Clojure goes way beyond what they could do

23:12 spacebat: ok, yeah I thought if they were live and you were involved, you'd be unspeakably busy

23:12 so there's a clojure book out from pragmatic programmers next march or so?

23:13 I'll have to start reading the docs again, after playing with CL a few months I think things will make more sense

23:14 rhickey: spacebat: they work, but one thing I learned was that very very few CL people care about Java integration, yet Clojure proves many people care about Lisp + JVM and the other things Clojure addresses

23:15 spacebat: well I like CL, and I have some hopes that Armed Bear CL will get there in the end

23:16 but I know that lisp needs a single modern implementation to focus on, and clojure might just be it

23:16 s/implementation/dialect/

23:17 rhickey: I'm on a 'Next 50 Years of Lisp' panel at OOPSLA - should be interesting

23:17 spacebat: sweet

23:17 a bit more conservative than the idea of a '100 year language' like arc

23:18 I want to believe arc is significant, but I've yet to grok what's so good about it

23:18 been playing with ECL, librep and picolisp the last little while

23:19 both librep and picolisp are mentioned as pre-arc arcs by some people

23:19 Chouser: I used a window manager for a long time that was largely written in librep

23:19 spacebat: picolisp is probably the most radical, purely interpreted, functions can mutate their own forms at runtime etc

23:20 gnuvince_: rhickey: what's that panel gonna address? Multi-core programming?

23:20 spacebat: sawfish, yes I've been meaning to run that again and play

23:20 anyn: http://paste.lisp.org/display/68403

23:21 converted the celcius example to java-gnome

23:21 it's probably more verbose than it needs to be, but it works

23:21 spacebat: and picolisp has no special forms, you just bypass evaluation by declaring a function like (de myfunc X (;;in here X is a list of all arguments, and you can evaluate what you want))

23:21 rhickey: gnuvince_: It's a panel - me, Will Clinger, Kent Pitman, and Martin Simmons opining about the future of Lisp

23:22 spacebat: macros are just normal functions because everything is runtime

23:22 rhickey: Clojure represents my opinion pretty well

23:22 spacebat: I hope they video that talk

23:22 gnuvince_: I imagine

23:25 Chouser: hm, using a javascript debugger to examine clojure data structures is ... tricky.

23:52 cads: spacebat, have you read heinlein's stranger in a strange land?

23:56 spacebat: no, its just been on my list forever

23:57 and I've spent enough time on slashdot for grok to slip into my vocab

23:57 dnm: Howdy.

23:58 Chouser: dnm: hi!

Logging service provided by n01se.net