#clojure – 2010-01-30

2010-01-30

00:11	krumholt	will a compiled jar file increase executing speed compared to evaluating something in the slime repl?
00:12	chouser	krumholt: the compiled jar might load faster, but once it starts running it should be the same
00:13	krumholt	chouser, ok thanks
00:30	krumholt	i noticed that a few type in clojure start with an A like ASeq and APersistentVector. There is also ISeq which i assume stands for Interface. what does the A stand for? and is it a kind of interface?
00:31	chouser	A is for Abstract
00:32	chouser	an Interface provides no implementation at all for any method
00:32	krumholt	ok and abstract provides some implementation for all subclasses?
00:32	chouser	an Abstract class provides implementations for some methods, but not all of them, so you still can't instantiate one.
00:32	krumholt	ok thanks for clearing that up
00:33	chouser	you need a concrete class like PersistentVector in order to actually make instances. It will provide implementations for the methods the abstract class left out, plus could provide new methods of its own (though that's generally bad practice)
00:35	krumholt	i get the idea thanks. is there a convention to name protocols?
00:35	chouser	That's a great question. I don't think any has emerged yet.
00:36	chouser	I suspect they'll be capitalized like type names, but beyond that I don't know.
00:37	krumholt	ok. i think i just go with PMyprotocol for now
00:41	chouser	We'll all have to get such conventions ironed out before fogus and I get around to writing the protocols chapter. :-P
00:43	krumholt	ok one more question. can i have my own constructor for protocols? because i need 4 fields but the user really only ever needs to supply 3 of those.
00:44	chouser	fields? are you talking about deftype?
00:45	Raynes	I think he is.
00:45	Raynes	Otherwise, he's using a different language than the one I'm using.
00:45	Raynes	:o
00:45	Raynes	Speaking, or otherwise.
00:46	chouser	krumholt: Users will not generally create deftype instances directly -- instead you'll want to provide a regular function they can call, and create your instance in there.
00:46	krumholt	chouser, so i should provide a make-instance function for a deftype
01:10	chouser	krumholt: yes, if it users will be wanting to make instances of that deftype
01:35	krumholt	in a deftype i should be able to use the unqualified name for calls to new and instance?. this seems not to work for calls to instance?
01:35	krumholt	http://paste.lisp.org/display/94136
01:39	Raynes	I hereby declare that load-file is the most awesome thing ever.
01:40	Raynes	Arbitrary user configuration has never been so easy.
02:06	Raynes	This email client is going to be the most hilarious mess of code evar.
02:20	Raynes	Is there a version of load-file that takes a java.io.File?
02:25	krumholt	Raynes, (load-file (.toString yourjava.io.File)) that shuold work
02:25	Raynes	krumholt: Hah. Nice. I didn't know Files had a toString method.
02:26	Raynes	Thanks.
02:26	Raynes	Hrm. They've been "migrating to ircd-seven shortly" for the last two weeks or so. >.>
03:11	Raynes	Hrm. Why is it that lein new spits out src/projectname.clj with a namespace of projectname.core?
03:11	Raynes	Shouldn't the namespace just be projectname, considering it's top level in src?
03:15	_ato	Raynes: yes, that's a bug that was fixed a while ago. You might want to update to lein 1.0.1
03:15	Raynes	I didn't know I was using an old version.
03:16	_ato	lein -v should tell you what version you have
03:16	_ato	although I don't know when that -v option was added
03:17	Raynes	_ato: The way to upgrade would be to download the install script, and do that stuff again right?
03:17	Raynes	There isn't any update magic I don't know about, I'm hoping.
03:19	_ato	Raynes: yeah, just do the install steps again. Probably it'll eventually get a lein self-update or something, but it doesn't do that yet
03:20	Raynes	Just checking. Didn't want to fry anything. :p
04:59	Hali_303	is there something in slime that allows me to automatically insert a :use in case I start to type a form with (str-join? so that I don't have to manually type the :use at the top?
05:03	krumholt	am i correct that methods from deftype are not to be invoked by the user of a library directly. instead there should be a number of functions using the methods?
05:04	_ato	Hali_303: not for :use AFAIK. _mst did this hack for java :imports though: http://dishevelled.net/Generating-Clojure-import-lines-using-SLIME.html
05:05	Hali_303	_ato: so for example, if you'd like to use str-join, but it's early morning and you don't know what package has that, what do you do?
05:08	_ato	Hali_303: heh, if I was smart, I'd probably just leave http://richhickey.github.com/clojure-contrib/ open in another window and search it. But actually I tend to just do: grep -r 'str-join' ~/src/clojure-contrib/src
05:09	_ato	krumholt: if I understand it correctly the idea is that deftype methods are only for java interop (ie implementing a java interface)
05:09	Hali_303	_ato: I see, thanks
05:10	_ato	ah wait.. I see protocols get called "methods" as well
05:11	krumholt	_ato, i am not using any java at all
05:11	_ato	defprotocol will define the functions
05:12	_ato	(just think of them like grouped multimethods)
05:12	_ato	so to answer your question directly: yes you are correct. ;-)
05:13	krumholt	ok but except for the speed gain they are really a lot of extra work. multimethod is somewhat nicer
05:14	_ato	yes, I agree
05:16	_ato	they were originally introduced so that you can write fast data structures in Clojure. If you're not writing a data structure (or have crazy performance requirements) then you probably better off sticking with multimethods, maps and vectors
05:16	krumholt	i think i am using them wrong. i am stuck in object oriented thinking :)
05:19	_ato	yeah, it's too easy to start thinking of them as OO objects, but a lot of the time what you're really looking for when you think "object" is just a plain old map and some plain old functions.
05:21	Raynes	I use deftypes where I would use structs. For instance, in my email client, I have a MessageInfo type that holds all relevant email message information. A processing function processes an email folder and turns an array of Messages into a sequence of MessageInfos
05:21	Raynes	I've yet to have a reason to use multimethods or protocols so far.
05:23	Hali_303	how can I have rainbow parentheses in emacs?
05:23	Hali_303	I've got emacs installed, but it's still black and white
05:23	Hali_303	I mean slime installed
05:23	somnium	being able to implement IFn on deftypes is cool. They feel really first class.
05:23	Raynes	Hali_303: highlight-parentheses-mode
05:23	Hali_303	Raynes: thanks
05:24	Raynes	Hali_303: Also, show-paren-mode is awesome.
05:25	Chousuke	hmm
05:26	Chousuke	Raynes: as long as you don't overuse types. :)
05:26	Raynes	Chousuke: Structs are used for maps that have keys in common, right?
05:27	Chousuke	it's a good thing that types are interoperable with standard maps but I fear people will use them for every little thing :/
05:27	_ato	Hali_303: if you want vim-style rainbow parens, where all the parens are coloured, not just the ones the cursor is inside, then there's this one: http://dishevelled.net/elisp/rainbow-parens.el
05:27	Raynes	I want. :D
05:28	_ato	Hali_303: personally I find that way too "busy" though. _mst just made that because of all the people complaining that emacs didn't do vim-like rainbow parens, he doesn't use it either. ;-)
05:28	Hali_303	:D
05:28	Hali_303	I'll try it and see if it helps or not
05:28	Chousuke	paredit does away with most need for paren highlighting though. :)
05:29	Hali_303	hm. I'm not using that, what is that good for?
05:29	Raynes	Paredit doesn't help me with that sort of thing at all.
05:29	Raynes	If anything, it slightly hinders me.
05:29	Chousuke	Hali_303: it lets your edit s-expressions instead of text
05:29	_ato	heh.. I prefer less-visible parens. I don't look at them much
05:29	the-kenny	C-k doesn't kills the current line, but the next sexp :)
05:30	Chousuke	Hali_303: you will never have to hunt for the missing paren again!
05:30	Chousuke	the-kenny: that's what I like most
05:30	Hali_303	thanks, I'll try that then
05:30	Chousuke	makes moving code around very easy
05:31	the-kenny	Chousuke: Yeah, that's very cool
05:31	_ato	technomancy or someone did a paredit tutorial didn't they? it takes a little while to learn the hotkeys but once you do it's extremely handy, I find it really painful coding without paredit now
05:32	krumholt	ok i have a question if i just want to dispatch on type is there a way to do that with multimethods that is fast?
05:32	Chousuke	the-kenny: I also like barf and slurp eg. for wrapping things in a let (or removing one)
05:32	Chousuke	krumholt: deftype :P
05:32	Chousuke	krumholt: it's only on master right now.
05:32	Chousuke	whoops, deftype and protocols I mean :)
05:33	_ato	this might be what I'm thinking of: http://p.hagelb.org/paredit-outline
05:34	krumholt	Chousuke, ok thanks
05:34	the-kenny	Chousuke: Yes, the wrap-stuff is pretty neat
05:39	Hali_303	yet another slime question: when I do a CTL c CTL k to do compilation and there is an error, how do I jump to the line where the error is?
05:42	noidi	Hali_303, great question, I've wondered about that too, but I've been always too busy fixing my code to remember to ask about it :)
05:43	_ato	I think it might be supposed to be: C-x ` (backtick) but it doesn't work for me. I might be one of the SLIME features that swank-clojure hasn't implemented yet
05:43	Chousuke	I hadn't even thought of that :P
05:43	_ato	in my output the errors are prefixed with "Unknown location"
05:43	_ato	although it does have the line number in the actual error message
05:43	_ato	I suspect SLIME is the one saying that "Unknown location"
05:43	Chousuke	usually the error is local to what I was previously editing so having a jump function wouldn't be that much faster anyway :/
05:44	Chousuke	Hali_303: also, that's "C-c C-k" in emacs parlance :)
05:44	_ato	I just usually M-g g and type the line number in the error
05:44	_ato	it'd be nice if there was a shortcut though :(
05:44	Hali_303	Chousuke: ok :)
05:44	Hali_303	_ato: better than nothing, thanks
05:49	Hali_303	paredit is really cool :D
05:50	Hali_303	hm. however, it only insert normal parentheses automatically, not { or [
05:51	Hali_303	ok [ also works, but { does not
05:51	hiredman	my paredit does ( and [, but not {
05:51	Hali_303	yes
05:51	hiredman	right, shrug
05:51	Hali_303	hiredman: btw what code are you using to auto-load it for slime?
05:52	Hali_303	now I have to manually type M-x paredit-mode
05:52	noidi	(add-hook 'clojure-mode-hook (lambda () (paredit-mode +1)))
05:52	Hali_303	noidi: thx
05:53	noidi	or for slime (add-hook 'slime-mode-hook (lambda () (paredit-mode +1)))
05:53	noidi	the latter doesn't work for me though, for some reason
05:53	noidi	and to highlight parens every time you load paredit:
05:53	noidi	(add-hook 'paredit-mode-hook (lambda () (show-paren-mode +1)))
05:54	the-kenny	myn paredit does { [ and ( :)
05:55	noidi	the-kenny, how did you manage that?
05:55	_ato	mine does all three as well. I added this for the M-{ hotkeys: http://gist.github.com/290508
05:55	noidi	I always have to insert the closing } manually or paredit gets confused
05:55	the-kenny	noidi: I think I got the file from someone here. Maybe _ato?
05:55	noidi	ok
05:56	_ato	I'm also using paredit version 22 if that makes any difference
05:57	noidi	Hali_303, you might (or might not, depending on how much you like to look at parens) like parenface.el -- it dims the parens so they blend into the background a bit
05:57	Chousuke	my paredit works just fine as well
05:57	_ato	http://mumble.net/~campbell/emacs/paredit-beta.el
05:58	noidi	here's parenface.el with clojure support added http://gist.github.com/290509
05:58	noidi	I've probably changed the paren color, too, to match zenburn
05:59	Hali_303	after typing (str-join how do I display the docs for str-join?
05:59	noidi	ah, I'm using paredit 20 from elpa
05:59	Hali_303	I know that it shows the parameters at the bottom
05:59	Hali_303	but I'd like to see the full docs
05:59	_ato	C-d C-d d
05:59	noidi	how the %#¤& can the elpa package be 2 versions behind? o_O
06:00	Chousuke	no-one has bothered to update it :)
06:00	_ato	you can also use C-. to jump to the source code for str-join
06:00	_ato	err. I mean M-.
06:01	Chousuke	and technically it's just one version behind, as 22 is beta
06:01	_ato	and then M-, to return to where you were
06:01	noidi	Chousuke, okay, thanks... I'll upgrade to 22b then
06:02	noidi	_ato, thank you! I didn't know about the jump functions in slime!
06:02	noidi	I thought I needed tags like with C
06:02	noidi	so I never bothered :)
06:04	_ato	yeah, that's one of the other nice things about having the editor connnected to the compiler like that. It doesn't need tags, it can just eval this:
06:04	_ato	,(meta #'drop-while)
06:05	_ato	huh.. no clojurebot? well then the result is: {:ns #<Namespace clojure.core>, :name drop-while, :file "clojure/core.clj", :line 1857, :arglists ([pred coll]), :doc "..."}
06:08	hiredman	oh, right
06:08	hiredman	stupid registeration requirement
06:09	hiredman	clojurebot: ping?
06:09	clojurebot	PONG!
06:09	hiredman	,(meta #'drop-while)
06:09	clojurebot	{:ns #<Namespace clojure.core>, :name drop-while, :file "clojure/core.clj", :line 1875, :arglists ([pred coll]), :doc "Returns a lazy sequence of the items in coll starting from the first\n item for which (pred item) returns nil."}
06:09	_ato	weirdly it replied to me in privmsg which I thought also required registration
06:09	_ato	or have they changed it now so messaging a channel requires registration but privately doesn't any more?
06:09	hiredman	I don't know
06:10	Chousuke	_ato: isn't it configurable by the user?
06:10	_ato	Chousuke: ah.. yeah that's probably it
06:11	kotarak	_ato: I have trouble with clojars and every scp which is not scp.
06:11	Hali_303	do cross-reference shortcuts work for you? I tried C-c C-w c, but it doesn't work "Evaluation aborted"
06:11	kotarak	_ato: neither jsch nor ganymede work with clojars :(
06:12	Hali_303	C-c < this doesn't work either
06:13	hiredman	I thought clojars used jsch
06:13	_ato	Hali_303: nope they don't work for me either. Probably more SLIME features not implemented for Clojure
06:13	Hali_303	_ato: sad :/ that's a feature that is very useful in practice
06:14	kotarak	jsch + "real" ssh = works, ganymede + "real" ssh = works, "real" scp + clojars = works, jsch + clojars = nope, ganymede + clojars = nope
06:14	_ato	kotarak: hmm I wrote a (pure) jsch client that works
06:15	_ato	I knew it didn't work with maven
06:15	kotarak	_ato: huh? May I use that? I tried with the jsch ant scp task, didn't work.
06:15	_ato	maven also uses jsch I think
06:15	kotarak	_ato: I'd like to have a look what that does differently.
06:16	kotarak	_ato: jsch is a little annoying because there are lots of incompatibilities with jsch, ant, gradle, bleh.
06:16	_ato	it might be because Clojars "talks" on stderr, to give feedback about whether it succeeded or not and to give an error if there's a problem (like if you're trying to post to a groupId you don't own, or you didn't include the POM or whatever)
06:16	kotarak	_ato: But I'd like to know why it doesn't work.
06:17	_ato	kotarak: http://github.com/ato/lein-clojars/blob/master/src/leiningen/push.clj
06:19	kotarak	_ato: thank you for the link. I'd really like to support clojars directly from clojuresque. At the moment I'm shelling out to scp. Not very nice and doesn't for windows
06:20	_ato	I wondered whether it's because I'm doing this to hook up the SSH session's error stream to stdout: (.setErrStream System/err true)
06:20	_ato	maybe the other JSch scp implementations don't do that and a buffer somewhere gets blocked up
06:20	_ato	maven just hangs when you try to make it scp something to clojars
06:21	_ato	that's just a guess though, I might be completely wrong. I never got around to trying to debug it
06:23	kotarak	_ato: I get a failure that some read failed and it says connection broken or so. I also thought about the buffer question. I'll see and have a try.
06:24	_ato	okay. If you find out what it is, let me know so I can try and fix it on the server side
06:24	kotarak	k
06:24	kotarak	_ato: btw. goot to see you back :)
06:24	kotarak	s/goot/good
06:25	_ato	:)
06:27	kotarak	_ato: another btw: the search doesn't work. When you search for clojureql you find two uploads which are not the official one. And the official one called "clojureql" is not found...
06:29	_ato	yep.. search fails to find things without a description (root cause: the result of concatenating a string to a null in sqlite is null :-P)
06:30	_ato	fix will be out soon (I know I've been saying that for a while... just keep getting distracted by other things)
06:30	kotarak	LauJensen: clojureql needs a description on clojars to show up in the search
06:32	kotarak	_ato: oh, and another thing (sorry, collected some topics...): I'd like to be able to delete artifacts again.
06:32	kotarak	eg. my clojares-deploy-testcase for my scp experiments (couldn't do it on my server, because there it worked.... :( )
06:34	_ato	yeah... as far as I know there's no way to tell maven to undeploy something from a repo (as you normally wouldn't want to as it'd break dependencies).
06:34	_ato	for now just email me if you need something deleted and I'll do it by hand
06:35	_ato	at some point I'm going to ditch maven completely from the clojars codebase, and just have it serve up a maven-compatible view of the repo
06:35	LauJensen	kotarak: I'll fix it, I'm AFK
06:35	kotarak	_ato: the thing in question is org.clojars.kotarak/clojuresque-clojars-deploy-testcase, but you can leave it around for now, then I have something to test while investigating the scp issue.
06:38	kotarak	_ato: /again sorry/ and the foo-1.2.3.pom change is not live
06:39	krumholt	in emacs minibuffer i usually get help for functions. this is not working for multimethods or protocol methods. can i get the minibuffer to work for them too?
06:39	_ato	kotarak: what foo-1.2.3.pom?
06:39	kotarak	_ato: clojars didn't recognise the .pom extension. You changed that, but the change didn't go live.
06:40	_ato	aah.. right. Yep, sorry. I'll try to get all that stuff out tomorrow. I've been really slack about it. :(
06:56	hoeck	krumholt: multimethods have no arglist metadata attached, so the minibuffer help has nothing to display
06:57	hoeck	krumholt: but if you define your multimethod with explicit :arglists metadata, minibuffer help works for them too
06:57	krumholt	hoeck, can i do that?
06:57	krumholt	hoeck, super thanks
06:57	kotarak	this can be remedied by defining multimethods as: (defmulti multifn "Some multimethod" {:arglists '([arg1 arg2 arg3])} dispatch-fn)
06:58	krumholt	is that possible for protocol methods too?
06:58	kotarak	Hmm.. Protocol methods don't have arglist info?
07:00	hoeck	no, they have not, but I guess its on the todo-list
07:00	krumholt	no. or at least they don't have minibuffer help
07:01	krumholt	can i add it manually? or will i just have to wait a while?
07:02	patrkris	Does anyone know of any websites written in compojure, that I can have a look at and learn from? Specifically, if I use the HTML functions in compojrure, how do I best organize "templates" and such.
07:02	hoeck	krumholt: yes, defprotocol methods are normal fns bound to a var
07:03	hoeck	krumholt: just use alter-var-root and with-meta
07:04	krumholt	so (alter-var-root myvar (with-meta myvar {:arglist '([arg1 arg2])}) ?
07:06	hoeck	krumholt: sorry, its (alter-meta! #'myvar assoc :argslist '([a]))
07:06	krumholt	hoeck, many thanks
07:07	Raynes	How do I comment out a line of code with paredit active?
07:07	hoeck	krumholt: np :)
07:07	Raynes	If I attempt to comment out a piece of code the old fashioned way, it inserts a newline after the semi-colon.
07:08	AWizzArd	,(doc alter-meta!)
07:08	clojurebot	"([iref f & args]); Atomically sets the metadata for a namespace/var/ref/agent/atom to be: (apply f its-current-meta args) f must be free of side-effects"
07:08	AWizzArd	The docstrings don't mention functions or symbols.
07:10	mikem	tomoj: I saw you had issues building clojure-contrib with pprint.ColumnWriter recently. how did you solve this? the build fails for me building pprint.PrettyWriter
07:11	mikem	ClassNotFoundException: clojure.contrib.pprint.ColumnWriter (PrettyWriter.clj:17)
07:11	hoeck	AWizzArd: the protocol function is bound to a var, just like a normal defn function
07:13	hoeck	Raynes: I don't like this behaviour, so I removed the paredit-semicolon command from my keymap
07:13	Raynes	hoeck: Good idea.
07:15	krumholt	hoeck, it's arglists not argslist :) if anyone else tries but thanks anyway
07:16	hoeck	krumholt: you're right
07:17	_ato	patrkris: it's not particularly clean code, but you could look at my code for the clojars webapp
07:18	_ato	patrkris: I have a clojars.web.whatever module roughly for each page
07:18	_ato	http://github.com/ato/clojars-web/tree/master/src/clojars/web/
07:19	_ato	no idea whether it's the "right" way of doing it
07:19	_ato	I couldn't find any examples either, so just went off and did my own thing :p
07:21	_ato	"controller" like functionality is usually in the same file but a different function
07:24	_ato	this might be another one to look at, the old compojure.org code: http://github.com/weavejester/compojure.org/tree/master/src/org/compojure/
09:30	dabd	..
09:48	tomoj	mikem: hmm
09:48	tomoj	on a different project I just added 1.2.0-master-SNAPSHOT deps and it worked fine
09:49	tomoj	haven't tried again yet to build contrib manually
09:49	tomoj	and I still don't really understand what happened when I had the problem
09:49	tomoj	but, good luck :/
09:52	mikem	tomoj: i see. strange, i'm not doing anything special, just building clojure from source, then clojure-contrib. i wonder how it works on the build servers
09:52	tomoj	apparently it doesn't
09:52	tomoj	well, something's not quite right anyway
09:53	tomoj	clojure-contrib is red at build.clojure.org
09:53	mikem	good point :) looking at that now...
09:53	tomoj	not sure whether that means tests failed or what exactly
09:54	tomoj	the jars I got seemed to work fine, so someone must be building successfully...
09:54	mikem	it's the same thing I'm seeing: http://build.clojure.org/job/clojure-contrib/lastFailedBuild/console
09:54	mikem	maybe the jars are from the last successful run
09:54	tomoj	yep, looks like the error I got as well
09:54	tomoj	ah, good point
09:54	mikem	i don't know enough about maven or clojure-contrib to figure it out :S
09:55	tomoj	if the last successful build was also tagged as 1.2.0-master-SNAPSHOT, I'd get that from lein deps I guess
09:55	tomoj	but..
09:55	tomoj	"last success: N/A"
09:55	tomoj	heh
09:55	mikem	just noticed an email to the mailing list asking about this
09:56	tomoj	oh, thanks
10:07	leafw	what is the preferred way to build clojure-contrib? There isn't a build.xml anymore
10:07	leafw	with mvn ? UGH
10:09	krumholt	i want to map across a sequence with a function. but i need to apply the function to the first argument once, to the second twice, to the third 3 times ... how can i do that?
10:10	leafw	sp Stuart Sierra, what motivated the removal of the ant build.xml ?
10:11	leafw	clojure.git still has it; so now one needs ant for one and mvn for the other? Uh? Or are the plans from removing build.xml from clojure.git as well?
10:11	noidi	krumholt, by applying twice, do you mean (f (f x))?
10:11	noidi	or (f x) (f x) for side effects
10:12	krumholt	noidi, (f (f x))
10:12	krumholt	no siede effects
10:12	krumholt	-e
10:12	noidi	ok
10:12	_ato	leafw: http://groups.google.com/group/clojure-dev/browse_thread/thread/cd807a1e5d7ce77d
10:15	somnium	,(map #(take %1 (iterate inc %1)) (range 5))
10:15	clojurebot	(() (1) (2 3) (3 4 5) (4 5 6 7))
10:15	leafw	thanks. I forgot the music moved to clojure-dev
10:16	AWizzArd	is there a reader macro for entering symbols or keywords that contain spaces?
10:16	somnium	krumholt: something like that?
10:16	AWizzArd	,(keyword "hello guys")
10:16	clojurebot	:hello guys
10:16	krumholt	somnium, i am trying to understand that
10:17	AWizzArd	,:hello guys
10:17	clojurebot	:hello
10:18	_ato	AWizzArd: I doubt a keyword/symbol that contains spaces could really be considered valid
10:18	krumholt	somnium, i want something like that just simpler
10:18	krumholt	,(defn foomap [fn lst] (if (empty? lst) lst (let [temp (map fn lst)] (cons (first temp) (foomap fn (rest temp))))))
10:18	clojurebot	DENIED
10:18	somnium	,(let [f #(* % %)] (map #(take %1 (iterate f %2)) (iterate inc 1) (range 10 14)))
10:18	clojurebot	((10) (11 121) (12 144 20736) (13 169 28561 815730721))
10:19	AWizzArd	_ato: well, see what i did to Clojurebot some seconds ago
10:19	AWizzArd	,(keyword "a b c d e")
10:19	clojurebot	:a b c d e
10:19	AWizzArd	,(symbol "Once upon a time...")
10:19	clojurebot	Once upon a time...
10:19	_ato	,{:a 1 :a 2}
10:19	clojurebot	{:a 1, :a 2}
10:19	_ato	is that a valid map? ;-)
10:20	AWizzArd	_ato: no, that's a bug
10:21	krumholt	,((fn foomap [fn lst] (if (empty? lst) lst (let [temp (map fn lst)] (cons (first temp) (foomap fn (rest temp)))))) inc [1 1 1 1 1])
10:21	clojurebot	(2 3 4 5 6)
10:21	_ato	it just doesn't verify input. same with symbol/keyword
10:21	krumholt	thats what i want
10:24	_ato	,(namespace (symbol "foo///bar////baz"))
10:24	clojurebot	"foo///bar///"
10:25	AWizzArd	I don't know why (symbol "a b c d e") should not be a valid symbol.
10:26	AWizzArd	Common Lisp also allows arbitrary symbol names
10:26	Raynes	,(def !$&*)
10:26	clojurebot	DENIED
10:26	Raynes	lol'd
10:27	AWizzArd	yeah, it needs a reader macro to enter them
10:27	_ato	,(let [!$&* 5] (inc !$&*))
10:27	clojurebot	6
10:28	krumholt	what happens to clojurebot if i evaluate an endless loop? and can i try? :)
10:28	AWizzArd	,(symbol "hello\nworld")
10:28	clojurebot	hello world
10:28	Raynes	krumholt: Wont let you.
10:28	AWizzArd	krumholt: the bot gives you only a few seconds of execution time
10:28	Raynes	,(iterate inc 0)
10:28	clojurebot	Eval-in-box threw an exception:java.lang.OutOfMemoryError: Java heap space
10:28	_ato	AWizzArd: what would you use a symbol with whitespace in it for?
10:29	Raynes	_ato: Kicks. :p
10:29	AWizzArd	_ato: can be a nicer name. As in strings, whitespacecansometimesbeanicethingtohave
10:29	krumholt	,((fn [x] (x x)) (fn [x] (x x)))
10:29	clojurebot	java.lang.StackOverflowError
10:29	tomoj	I never realized that arraymaps could have duplicate keys like that
10:29	_ato	but as a variable/function name?
10:30	AWizzArd	probably not a common thing, but for symbolic computation.. why not?
10:30	AWizzArd	"Some men see things as they are and say 'why'? I dream things that never were and say: 'why not?'" :)
10:32	_ato	heh. In my mind it's similar question to why not have reader macros. Just makes things hard to understand. If you want to do symbolic computation with names that have spaces, why not just use strings as your symbols?
10:32	AWizzArd	Strings don’t support metadata yet
10:32	tomoj	yet?
10:32	AWizzArd	tomoj: even if they never will have my statement is true j)
10:32	tomoj	how could they ever?
10:33	AWizzArd	;)
10:33	tomoj	ah
10:33	tomoj	hehe
10:33	tomoj	just the implicature
10:33	_ato	then define your own box type
10:34	AWizzArd	_ato: I just think that either (symbol "a b c") should throw an error or we could have a reader macro to give us a notation for such symbols
10:36	_ato	Indeed, it probably should throw an error. Just as that map should throw an error. (although their may be efficiency concerns about checking it every single time for the map)
10:36	tomoj	hmm
10:36	AWizzArd	with your hashmap example i would certainly agree that an error message would be the right thing
10:36	tomoj	I would've expected it to just take the latest value
10:36	AWizzArd	for symbols.. i could accept both solutions, error or reader syntax
10:36	tomoj	s/latest/last/
10:36	AWizzArd	tomoj: yes
10:37	AWizzArd	that would also be okay
10:37	tomoj	,(assoc {:a 1 :a 2} :b 3 :c 4 :d 5 :e 6 :f 7 :g 8 :h 9 :i 10)
10:37	clojurebot	{:a 2, :c 4, :b 3, :f 7, :g 8, :d 5, :e 6, :i 10, :h 9}
10:37	AWizzArd	,(hash-map 1 :x, (long 1) :y)
10:37	clojurebot	{1 :x, 1 :y}
10:38	AWizzArd	,(apply identical? (keys (hash-map 1 :x, (long 1) :y)))
10:38	clojurebot	false
10:38	tomoj	I assume there it's just iterating through the arraymap and copying over, so the last value stays through the conversion?
10:38	tomoj	haha
10:38	cgrand	AWizzArd: there is this old ticket http://www.assembla.com/spaces/clojure/tickets/17-GC-Issue-13-%09validate-in-%28keyword-s%29-and-%28symbol-s%29 and in the old "todo" http://clojure.org/todo there's "arbitrary symbols in \| \|"
10:38	Raynes	(let [tehmapz {:a 1 :a 2}] (+ (:a tehmaps) 1))
10:38	AWizzArd	this is okay, because (int 1) and (long 1) are different things.
10:38	Raynes	,(let [tehmapz {:a 1 :a 2}] (+ (:a tehmaps) 1))
10:38	clojurebot	java.lang.Exception: Unable to resolve symbol: tehmaps in this context
10:38	AWizzArd	cgrand: oh good
10:38	Raynes	,(let [tehmapz {:a 1 :a 2}] (+ (:a tehmapz) 1))
10:38	clojurebot	2
10:39	Raynes	That's what I get for using semi-leetspeek in variable names.
10:39	tomoj	as long as it's consistent I guess it doesn't matter too much that it gets printed that way?
10:39	tomoj	just weird
12:39	qed	am I identified now?
13:02	Chousuke	qed: apparently, since you can send to the channel
14:37	alexyk	I use this to get fn names, but it errors on unbound vars:
14:37	alexyk	,(->> (ns-interns ns) (map (fn [[k v]] [k (fn? (var-get v))])) (filter second) (map first))
14:37	clojurebot	()
14:37	alexyk	how do I guard against unbound vars?
14:41	alexyk	it happens when I do a defn and it fails to compile; but apparently the var is left semi-defined.
14:49	cgrand	alexyk: .isBound
14:49	cgrand	,(.isBound #'print)
14:49	clojurebot	true
14:51	alexyk	cgrand: cool. But why if you do (defn xyz some-screwed-up-wrong-thing), xyz is somehow defined, but causes:
14:51	alexyk	java.lang.IllegalStateException: Var mongol.repliers/fetch-graph-xyz is unbound.
14:52	alexyk	also, is there an opposite of fn?, aside from (not (fn? ...)) ?
14:53	alexyk	i.e. an attempt to defn xyz fails with an error, such as some name used in defn is undefined.
14:53	alexyk	but that leaves a trace oof xyz in the namespace, which then leads to that error
14:59	alexyk	is there a grep for a list of strings?
15:00	arohner	alexyk: you're wanting to search through a list of strings for a regex?
15:01	alexyk	yessir
15:01	arohner	map over re-find
15:01	arohner	,(doc re-find)
15:01	clojurebot	"([m] [re s]); Returns the next regex match, if any, of string to pattern, using java.util.regex.Matcher.find(). Uses re-groups to return the groups."
15:03	alexyk	arohner: hm, ok; but wondered why isn't there just a grep. :)
15:03	arohner	it's your lucky day
15:04	arohner	clojure.contrib.str-utils2/grep
15:05	alexyk	ah, super.
15:06	alexyk	the flushing handle, the car, and the heater all broke, and the internet is slow, but the grep is there -- a lucky day indeed!
15:06	alexyk	:)
15:07	alexyk	it'll keep me warm in this New England frost :)
15:09	arohner	nice
15:09	arohner	yeah, it's really cold in Texas. It's almost 40!
15:10	alexyk	wow
15:36	alexyk	somnium: how do I force reload congomongo?
15:38	alexyk	is there a way to force (use ...)?
15:42	chouser	there'a :reload option to use and require
15:42	alexyk	so it looks the state of some vars in the namespace for congomongo got corrupted; I need to reload it... is there a way?
15:44	chouser	there'a :reload option to use and require
15:44	dnolen	alexyk: you can ns-unmap those problem vars I think.
15:44	chouser	there's :reload option to use and require
15:45	chouser	hello
15:45	alexyk	dnolen: ok so some things in ns X got messed up; just found remove-ns and did (remove-ns 'X)
15:45	alexyk	now (use 'X) fails since each var is still aliased to 'X from my current ns!
15:46	alexyk	how can redo (use 'X) now?
15:48	alexyk	so I did ns-unmap, and now (use 'X) returns nil and doesn't refill the namespace
15:48	dnolen	i think you need to do something different here
15:49	dnolen	you need to get all the mappings in your namespace
15:49	dnolen	filter out ones that don't belong
15:49	dnolen	and unmap those one by one.
15:50	dnolen	could probably done quickly with a set operation
15:52	chouser	hello?
15:52	clojurebot	BUENOS DING DONG DIDDLY DIOS, fRaUline chouser
15:52	chouser	yay, it works again.
15:52	krumholt	can i get a better timing than currentTimeMillis on the jvm?
15:53	chouser	dnolen: did you want the :reload option for use and require?
15:54	dnolen	chouser: my ^ was actually a response to alexyk having ns issues.
15:54	chouser	ah, alexyk, right sorry.
15:55	dnolen	chouser: so there's a :reload?
15:55	chouser	I was fighting freenode moderation and lost the log
15:55	alexyk	I do need :reload asap!
15:55	chouser	yes, :reload is there
15:55	chouser	and :reload-all
15:55	chouser	check the docstring for 'require'
15:56	chouser	you and use :verbose as well to confirm it's reloading what you want it to
15:56	chouser	note that won't unmap any vars, so you might still want to do that, depending.
15:57	alexyk	what happened is this: I Ctrl-C'ed a congomongo op, and the state of internal vars of somnium.congomongo (let's call is SC) got effed up good. It was used with (use 'SC). Things like (fetch-one ...) failed with mongo internal error. Now I did (remove-ns 'SC) since any (use 'SC) did nothing. Now I have fetch-onne referring to SC/fetch-one and (use 'SC) fails still. Should I try :reload or do something else?
15:57	dabd	I'm working a bit with JSON and I was thinking of how I could generate
15:57	dabd	a JSON object that conforms to a given JSON Schema. Something like JSON
15:57	dabd	Beans but instead of beans I would have a bunch of functions. The
15:57	dabd	functions could be automatically generated by a macro given a JSON
15:57	dabd	Schema. Is something like this already implemented or is there a better solution?
15:57	dabd
15:58	chouser	alexyk: I guess I'd try (use 'SC :reload-all :verbose) and see whatn happens
15:59	alexyk	no names of vars are printed, only three lines for load, in-ns, refer
15:59	hiredman	dabd: why?
16:00	alexyk	then trying to use a var says it's unbound
16:00	hiredman	clojure json libraries tend to transform json into clojure lists and maps
16:00	hiredman	why do you need some special functions?
16:01	alexyk	back later...
16:15	dabd	hiredman: I'm already using Clojure's JSON libraries but I end up writing code that builds up the Clojure data structures that are mapped to JSON. But I thought I could write a macro that writes the code for me by providing a JSON Schema.
16:21	dabd	hiredman: This http://paste.lisp.org/+20NO is an example of code I'm writing to construct JSON. I think this could be automatically generated from a JSON Schema much like JAXB generates Java classes from a XML Schema.
16:41	alexyk	ok, back to the ns problem. So there's nothiing in SC after it was remove-ns'd, and apparently use 'SC does nothing with or without :reload-all.
16:42	alexyk	I think it's rather general -- you have a ns with internal vars which represent a state, it's messed up; you want to purge the ns and reload it and init it into a virgin state. How?
16:43	dnolen	alexy: you could just unmap everything
16:44	alexyk	dnolen: am not getting it! unmap references from current ns A, to the purged SC?
16:44	dnolen	I think so.
16:45	dnolen	(map #(ns-unmap 'A %) (ns-map 'A)), then reload your A library
16:47	alexyk	aha
16:47	alexyk	dnolen: but I have many valuable vars in A!
16:47	alexyk	I need only to unload those referring to SC
16:48	dnolen	(ns-imports 'A)
16:48	alexyk	hotter
16:48	dnolen	oops
16:48	dnolen	sorry
16:48	dnolen	that's for java imports
16:49	alexyk	yeah, I need only references to SC
16:50	dnolen	you're going to have to write a filter for this
16:51	dnolen	ns-map returns all the symbols in the current name space, each with it's name and it's var
16:54	alexyk	dnolen: with a grep on the qualified name I guess
16:55	dnolen	no, something like this (str (:ns (meta (second (first (ns-map 'A))))))
16:55	dnolen	then you can check if that string is = to "SC"
16:55	dnolen	if so unmap
16:55	dnolen	actually
16:56	dnolen	you can also just compare the actually namespace object SC
16:56	dnolen	sorry
16:56	avarus	hi
16:58	alexyk	dnolen: cool, I was looking for that
17:00	alexyk	dnolen: you mean without str, the result of :ns? how do I make an ns object to compare to from literal 'SC?
17:03	alexyk	how do you supply a literal regex-pattern to re-find?
17:04	dnolen	#"pattern"
17:05	alexyk	ok, almost there: (->> (ns-map ns) (map second) (filter #(= (-> % meta :ns str) "somnium.congomongo")) (map #(ns-unmap ns %)))
17:06	alexyk	the only error is, in the end we get a list of Var's, and unmap needs Symbol
17:06	alexyk	how do I convert a #'var to 'var symbol?
17:07	Chousuke	hmm
17:07	Chousuke	,(.sym #'+)
17:07	clojurebot	+
17:09	alexyk	aha! but printed without ', even though a Symbol
17:09	Chousuke	of course
17:09	Chousuke	'+ would be (quote symbol) :)
17:09	Hun	no need to quote it in the output
17:09	alexyk	lovely
17:10	alexyk	here's the killing buzzsaw: (->> (ns-map ns) (map second) (filter #(= (-> % meta :ns str) "somnium.congomongo")) (map #(ns-unmap ns (.sym %))))
17:12	alexyk	but, still same result -- now (use 'somnium.congomongo), with or without :reload-all, says nil, and an attempt (mongo! :db "name") results in: java.lang.IllegalStateException: Var somnium.congomongo/mongo! is unbound. (NO_SOURCE_FILE:0)
17:13	alexyk	somehow it fails to actually refill the namespace
17:14	alexyk	when I change to it via (in-ns 'SC) and try (ns-map ns), I get java.lang.Exception: Unable to resolve symbol: ns-map in this context. Is this expected?
17:14	alexyk	i.e. shouldn't any namespace refer to clojure.core?
17:17	dnolen	alexyk: you may be getting yourself into a weird REPL state mire, is it really not possible to restart your REPL? is getting your program to the correct state really that hard that you can't write some functions or put some dummy declarations in a comment to eval later?
17:18	alexyk	dnolen: my repl is precious indeed, it contains 30 GB of painfully computed data
17:18	alexyk	it used to get more from mongo but mongo is mangled
17:18	alexyk	since I've seen it before, I want to solve the general problem of reloading a namespace
17:18	chouser	alexyk: you could try load-file directly on the .clj files that define the things you want to fix
17:19	alexyk	chouser: ok
17:20	dnolen	alexky: I see.
17:20	alexyk	is there an uptime for repl, and also a check of how much Xmx remains available?
17:20	alexyk	I'm flushing the data on disk all the time, still loading it is much longer than saving it, so I try to not reload
17:22	alexyk	ok now this is tremensously weird: I can still store into mongo while it's all undefined! I have a mongo-store defn in ns, which uses some SC defns. And it works! How is that possible?
17:23	alexyk	did those aux defns close upon somnium.congomongo/defns when they were defined and kept the references and use them, despite remove-ns/use again cycle?
17:24	KirinDave	Huh...
17:24	KirinDave	… #() is kinda brittle, isn't it?
17:27	avarus	I don't get "let" ...
17:28	KirinDave	avarus: Which part?
17:28	avarus	in general :)
17:28	cgrand	KirinDave: how's #() brittle?
17:28	hiredman	let x be equal to the result of some expression in the following expressions
17:28	avarus	the book says something about bindings
17:28	KirinDave	cgrand: I'm just surprised how it expands.
17:28	hiredman	avarus: it binds a name to a value
17:28	KirinDave	cgrand: I'm used to the way lambda in common lisp works, which is more amenable to certain things.
17:28	avarus	hiredman: oh my god...so simple :)
17:28	hiredman	a limited lexical scope
17:29	avarus	ya, got it, thanks :)
17:29	KirinDave	avarus: For more edification on the subject, expand let into lambda forms.
17:29	avarus	I'm just reading programming clojure and I just arrived "bindings" and the example there confused me
17:29	cgrand	KirinDave: I'm not an expert in other lisps but isn't Clojure's lambda is fn, no?
17:29	hiredman	KirinDave: doesn't really work in clojure
17:30	KirinDave	hiredman: It doesn't?
17:30	hiredman	you can recur across lets, but not across fns
17:30	KirinDave	hiredman: "recur across lets?"
17:31	cgrand	you can't recur across a fn, you always recur to the nearest fn or loop
17:31	hiredman	,((fn [x] (when x (let [y (not x)] (recur y)))) true)
17:31	clojurebot	nil
17:31	hiredman	,((fn [x] (when x ((fn [y] (recur y)) (not x)) )) true)
17:32	KirinDave	hiredman: That sort of defeats the purpose of what I was trying to explain, but that's interesting.
17:32	hiredman	BOOM
17:32	clojurebot	Execution Timed Out
17:32	hiredman	I wrote a macro that turned a let looking binding form into a sequence of fns
17:33	hiredman	not a Sequence sequence, just a bunch of nested fns
17:33	KirinDave	Right.
17:33	KirinDave	And it didn't work, because let is a special form in clojure, not just an expansion.
17:33	KirinDave	But that doesn't mean that it doesn't have value as an exercise, which is why I was recommending it to avarus.
17:34	hiredman	KirinDave: well, to fair, it is a special form in most lisps that I've seen
17:34	KirinDave	hiredman: I didn't say it wasn't.
17:34	hiredman	sure
17:34	KirinDave	So I'm not sure what kind of balance we're striking. :)
17:34	hiredman	~sicp
17:34	clojurebot	sicp is http://www.codepoetics.com/wiki/index.php?title=Topics:SICP_in_other_languages:Clojure:Chapter_1
17:34	hiredman	clojurebot: not that one
17:34	clojurebot	clojurebot: google nothing
17:35	hiredman	~sicp
17:35	clojurebot	sicp is http://web.mit.edu/alexmv/6.001/sicp.pdf
17:35	KirinDave	cgrand: Anyways, I made a mistake that seems to make clojure's compiler shit the bed and return "No message" for the error.
17:35	KirinDave	Or rather, I wrote some code and then it was used in a way I didn't intend. :)
17:51	avarus	I start to like clojure :>
18:06	hiredman	hmm
18:08	clojurebot	test
18:08	hiredman	clojurebot: ping?
18:08	clojurebot	PONG!
18:08	hiredman	a wild ride
18:32	technoma`	replaca: just checked in that change to clojure-http-client
18:33	technoma`	_ato: hey,
18:33	technoma`	somebody squatted on my clojure-http-client group, so I can't upload
18:33	technoma`	also, I squatted on org.clojure
18:33	technoma`	any chance those could be fixed?
18:34	technoma`	_ato: also: welcome back! assuming you're back.
18:36	_ato	technomancy: clojure-http-client is now yours
18:36	technomancy	_ato: thanks!
18:37	technomancy	how's things going?
18:40	_ato	technomancy: not bad, I'm gonna devote today to Clojars hacking. It's badly needing some love. :-)
18:40	technomancy	was starting to get worried about you
18:40	technomancy	glad to see you're back in action
18:41	technomancy	what's on the table for today?
18:43	_ato	fixing data migration to couchdb -- yeah I still haven't done that. Holidays (without phone coverage) + work has kept me away. I've also been experimenting with compiling a really tiny subset of Clojure to LLVM byte code, which has been a bit distracting.
18:43	technomancy	wow, fun
18:43	technomancy	I'm excited about the couchdb migration
18:43	technomancy	especially if it means we can move jars into couch attachments and then mirror via replication... that would be hawt
18:44	_ato	yes, I want to ditch relying on maven itself, it just makes things painful
18:45	_ato	I did a basic browse screen which I might improve with categories/tags of some sort. At the moment it's just browse by alphabet sort of thing
18:53	technomancy	cool
18:57	technomancy	sorry I dropped the ball on implementing browse myself
18:59	_ato	don't worry about it. :-)
19:49	replaca	technomancy: thanks!
19:59	AWizzArd	,(/ 12800 1600)
19:59	clojurebot	8
20:30	kiba	hello
20:53	kiba	hmm
20:58	abrooks	technomancy: Would you be interested in patches to add a run task to leiningen? (i.e. "lein run <ARGS>" would call the main entry point with the provided args.)
20:58	abrooks	technomancy: BTW, I love leiningen -- it's great stuff!
20:59	kiba	but I determine that I should build my first web application in clojure
20:59	kiba	which I have no idea about building web application in the first place
21:29	hiredman	clojurebot: ping?
21:29	clojurebot	PONG!
21:33	technomancy	abrooks: there's a thread on the mailing list about someone who already added that as a plugin
21:33	technomancy	I may merge it to lein proper down the road
21:34	abrooks	Hm. I must have missed it -- thanks for letting me know. Consider this my +1. :)
21:36	abrooks	Oh, there's a leiningen group. I missed that.
21:58	abrooks	technomancy: Having now used leiningen-run I'll give a verified +1. Quite nice. Particularly if eval-in-project gets nailgun support.
22:00	abrooks	Hm. I guess it's not eval-in-project getting nailgun, it's really lein/lein.bat.
22:01	abrooks	By eval-in-project we're already in a running JVM aren't we...
22:02	hiredman	I think eval-in-project spins up a new jvm
22:03	hiredman	I think someone did some kind of nailgun plugin for lein
22:04	hiredman	~google lein nailgun
22:04	clojurebot	First, out of 521 results is:
22:04	clojurebot	Index of /repo/lein-nailgun/lein-nailgun/
22:04	clojurebot	http://clojars.org/repo/lein-nailgun/lein-nailgun/
22:26	kiba	hmm
22:37	hiredman	compojure seems to be the framework most people use
22:37	hiredman	it has it's own irc channel and google group, dunno how active they are
22:38	kiba	not very active
22:39	kiba	documentation is so poor that it's hard to learn
22:40	hiredman	kiba: you might checkout http://data-sorcery.org/category/compojure/
22:58	technomancy	kiba: the mailing list for compojure is probably more active than the channel
23:33	kiba	how do I load other files
23:33	kiba	?
23:41	mtd	kiba: (load "filename-without-clj-extension") ?