#clojure – 2014-06-01

2014-06-01

00:04	duck1123	:( minimal aleph test-case works fine. It's my birthday, shouldn't this work for me today?
00:06	rhg135	duck1123, it's never that easy
01:54	tolstoy	duck1123: If it comes to it, http-kit might be a good alternative.
02:23	hadronzoo	Will leiningen use credentials.clj.gpg to authenticate not only on deploy, but also when searching for artifacts?
02:36	tolstoy	I'm guessing cljsbuild is not quite a good citizen in this case.
02:36	tolstoy	Oops. (cljsbuild hook is, uh, problematic for me)
04:13	no7hing	is this still the canonical way to get a record’s type in clojurescript? http://stackoverflow.com/questions/17190575/how-do-i-get-the-type-of-a-record-in-clojurescript
04:26	martinklepsch	are there other clojure devops things than pallet?
06:45	AeroNotix	https://gist.github.com/AeroNotix/334e54e5b11ce6a608ac
06:46	AeroNotix	Caused by: java.lang.RuntimeException: Unable to resolve symbol: in this context
06:46	AeroNotix	what's going on there?
06:46	blur3d	, (def track1-data ["B4000000000000002^CardUser/John^131210100000019301000000877000000?" "B4000000000000002^CardUser/John ^131210100000019301000000877000000?"])
06:46	clojurebot	#<CompilerException java.lang.RuntimeException: Unable to resolve symbol: in this context, compiling:(NO_SOURCE_PATH:0:0)>
06:48	TEttinger	"B4000000000000002^CardUser/John ^131210100000019301000000877000000?"
06:48	TEttinger	uh sorry
06:48	amalloy	you probably have a non-breaking space or other weird whitespace character in there
06:48	AeroNotix	hm
06:49	AeroNotix	ffs
06:49	blur3d	the double quotes are being evaluated
06:50	AeroNotix	when I copied the strings to a new def, it works...
06:50	AeroNotix	lolunicode
06:50	AeroNotix	Or something
06:50	AeroNotix	in other news, I need to generate track1 cc information, does anyone have a list or some code to generate this?
06:50	TEttinger	yeah there's probably an unprintable garbage char in the string
06:50	blur3d	yeah, something weird.. I got it working with single quotes and alphanum chars only
06:51	AeroNotix	#FuckIt
06:51	AeroNotix	next problem
06:51	TEttinger	copying pasting might remove the unprintables haha
06:51	wagjo	AeroNotix: you have an EF BB BOM between last string and ])
06:51	AeroNotix	wagjo: lol
06:52	AeroNotix	I really don't know how that happened
06:53	blur3d	try typing it out by hand from scratch
06:53	AeroNotix	blur3d: I did, already said I did. It works
06:54	blur3d	kk
06:54	AeroNotix	I need some Track1 CC info now, the one I pasted above, works. But I need MOAR
06:54	AeroNotix	just for unit testing
08:12	enen	hey guys iv been meaning to get into clojure for some time now. while i can google on how to get started and wont bother you with that, i cant decide if i should try and learn also vim or emacs in the process. would it be worth my time?
08:14	Rhymor	enen: emacs has a great lisp integration. Vim is good for text editing (as opposed to text authoring)
08:15	Rhymor	If all you want to do is program, start with LightTable
08:15	enen	yes, light table looks great, i was thinking to go with it
08:15	Rhymor	It has an emacs and vim keybinding plugin. So you can experiment with these.
08:15	enen	yeah exactly
08:16	Rhymor	enen: in short, emacs is a programmer’s editor, vim is a system administrator’s.
08:17	enen	Rhymor: thats an interesting thought
08:17	locks	+1 for Light Table
08:18	locks	I was trying clojurescript the other day
08:18	locks	opened a browser tab on LT, connected, bam live reload
08:18	enen	im a fan of bret victors stuff, thats why id like to try light table
08:19	enen	locks: coming from an ios dev background where i have to wait 30 secs to a minute for my simulator to run my code this looks like a dream haha
08:22	locks	yeah can't wait for some of the LT updates
08:23	ambrosebs	enen: also check out Cursive Clojure if you're shopping.
08:24	wagjo	,(+ 2 2)
08:24	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
08:25	ambrosebs	,+
08:25	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
08:27	wagjo	seems like I've broken clojurebot (again), sorry
08:30	locks	,)
08:30	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
08:30	locks	,)
08:30	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
08:30	locks	NooOoo
08:48	scape	opinions/feedback welcome :) https://github.com/viperscape/kuroshio/blob/master/examples/chan.clj
09:19	AWizzArd	rhg135: please check your priv query windows
11:55	irctc	hi everyone, I have a question about using nREPL (cider) in Emacs. In my clojurescript project, I start the repl server using cider-jack-in command. However, whenever I try to execute a function from the clojure file, a compilerexception is thrown with the message unable to resolve symbol. What could be the problem? Does cider load all dependencies automatically?
11:56	irctc	anyone is there?
12:02	yotsov	irctc: you could try asking this in #clojurescript I guess. I don't use emacs so can't help
12:07	hyPiRion	irctc: It's silent on sundays
12:22	visof	hello is the best way to write concurrent fast code is using future?
12:45	agumonkey	hello aliens
13:54	mindbender1
13:54	Jaood
14:11	andyf	ambrosebs: Is it time to change the recommendation at this link, given that tools.analyzer(.jvm) might be ready to build tools now? https://github.com/clojure/jvm.tools.analyzer#future-deprecation
14:21	bbloom	amalloy_: burned by zero arity assoc AGAIN
14:23	andyf	0 arity, or 1 arity?
14:27	fifosine1	Quick question: I'm looking at the function definition "(defn thumbnail-link [{:keys [userid name]}]", what's going on here with the map?
14:31	jeremyheiler	fifosine1: map destructuring. you can call it like: (thumbnail-link {:userid "foo" :name "bar" :somethingelse "baz"}) and it will pull the specified values out for you.
14:31	mindbender1	fifosine1: That's destructuring. You can read more here and research further: http://blog.jayfields.com/2010/07/clojure-destructuring.html
14:31	fifosine1	ok, will read, thank you
14:36	fifosine1	mindbender1: Oh wow, the :keys directive is pretty cool
14:37	mindbender1	Yes it is. There are many cool stuff in clojure as you'll soon find out.
14:58	pwiki_	What is recommended for working with dates and time in ClojureScript? Is cljs-time mature enough, or should I use a JS library like MomentJS? Does someone has experience with this?
15:00	benmoss	pwiki_: i recommend using the Google Closure classes
15:00	benmoss	i’m not familiar with cljs-time really
15:00	jeremyheiler	benmoss: it appears that is what cljs-time does.
15:01	benmoss	so to the extent that it’s just a wrapper it’s probably “mature"
15:01	clojurebot	It's greek to me.
15:01	jeremyheiler	yeah probably.
15:02	pwiki_	ok, then I will give cljs-time a try, thanks
15:17	benmoss	do people use chunked-seq stuff directly in their code much?
15:17	benmoss	chunk-buffer chunk-append chunk-cons etc
15:18	benmoss	i would guess its just an implementation detail that maybe you see in the wild in some deftypes but otherwise pretty rare?
15:45	fifosine	Has anyone here worked through Web Development in Clojure?
16:08	platz	Anyone tried Caribou?
16:08	platz	looks like kind of a batteries-included web framework as opposed to a minimal one
16:08	platz	for better or worse
16:11	amalloy	benmoss: i have never used chunk-*
16:14	benmoss	yeah all i could find was some stuff about using custom chunk sizes for optimization
16:16	serendipity1	why did Rich Hickey ultimately chose jvm over clr?
16:17	turbofail	because he didn't want to be trapped on windows?
16:17	turbofail	don't even bring up mono
16:18	bbloom	turbofail: seems like he did a fair bit of professional c# development, so i'm not so sure that he's super worried about windows
16:18	AWizzArd	Well, the C in CLJ stands for C# actually.
16:18	bbloom	there's an old interview w/ fogus where he said he wanted to do twice as much, not the same stuff twice
16:18	AWizzArd	C#, Lisp, Java
16:18	bbloom	and the jvm version was dramatically more popular, hence jvm wins
16:18	turbofail	that too
16:19	turbofail	part of the reason the JVM was more popular though was because it was more cross platform
16:20	serendipity1	thanks bbloom, that clears up things. I was wondering because he earlier was implementing dotLisp or smthg like that
16:21	bbloom	serendipity1: iirc he said in interviews that dotlisp was his "everybody's first lisp" project... presumably during the time he was writing C# professionally
16:21	benmoss	can someone explain to me why `chunk` isn’t idempotent? https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/ChunkBuffer.java#L30
16:22	serendipity1	If interop with c# is my prime concern should I choose IronScheme or ClojureClr?
16:22	bbloom	serendipity1: they are very different languages
16:24	bbloom	benmoss: once you call chunk, you've shared the private buffer... so you can't reuse it
16:24	serendipity1	I want to learn a homoiconic language but have ability to interact with c#
16:25	benmoss	bbloom: so just to enforce the pattern it makes the buffer unusable?
16:25	arrdem	serendipity1: I picked up several Scheme dialects, then Common Lisp, then Clojure. Clojure is the only one I use. homoiconicity is overrated, langage communities and semantics make all the difference.
16:25	bbloom	benmoss: yeah, when you call "chunk" on the buffer, you're releasing that resource, effectively
16:25	bbloom	serendipity1: we're bias in here, but learn ClojureCLR
16:25	benmoss	cool, thanks
16:26	bbloom	serendipity1: and if you're just learning, i suspect interop is relatively unimportant... unless you've got something in particular in mind you want to build with it
16:26	bbloom	serendipity1: IronScheme attempts to confirm to the scheme spec, which makes interop harder. clojure is designed to be hosted, so it's interop support will be much smoother
16:28	serendipity1	in my regular c# code i find myself in place where language starts getting in the way. i have higher order templates, lambdas, code generation all over the place and it just feels a little bit too verbose
16:28	bbloom	serendipity1: you won't find anybody disagreeing here :-P
16:29	serendipity1	:)
16:31	serendipity1	but as far as I understand I could also do all these fun lisp-macro magic in IronScheme, right?
16:32	arrdem	As an SRSN \| n >= 5 compliant Scheme IronScheme definitely has macros, so yes but macros are overrated.
16:33	bbloom	serendipity1: yes... with the caveat that i don't know anything about iron scheme... although traditionally scheme has a more complex macro system than clojure or common lisp, and so you may wish to prefer to learn about macros in a simpler context
16:33	arrdem	s/SRSN/RNRS/g
16:40	serendipity1	arrdem, why do you find homoiconicity and macro system overrated? I've some experience with homoiconicity (coding in mathematica, which is great but slow as hell) and it's the exact thing that makes me consider lisp for some auxiliary coding. In C# there's a mismatch between regular code and code generation. And then there is the generator for the code generator :)
16:43	arrdem	serendipity1: for some problems runtime code generation is awesome. I'm building a compiler for clojure this summer. Runtime code generation is what I do. That Clojure and other lisps have support for this baked in is awesome. However, after reading a ton of books on macros and runtime code generation I have yet to sit down on a problem and go "ah yes I can just generate code and eval it". There was one instance where I actually did do clojure code generati
16:44	turbofail	run-time code generation is great for transmitting functions over the wire
16:44	platz	The ugliness of code gen in non-homoiconic languages serves as a deterrent to over-use, and in C# or Scala it's not that bad, only requires an up-front learning investment
16:45	arrdem	yeah. Honestly I think that partial is probably the only runtime code generation that gets used "regularly".
16:46	arrdem	but that's just my 2c.
16:46	Bronsa	arrdem: your previous message cut at "I actually did do clojure code generati"
16:46	arrdem	Bronsa: :/ I gotta get ERC to resume splitting long messages.
16:47	arrdem	on I have yet to sit down on a problem and go "ah yes I can just generate code and eval it". There was one instance where I actually did do clojure code generation, but that was a parser generator and arguably not a nice one either.
16:47	Bronsa	arrdem: tell me how to do it once you figure it out :)
16:48	turbofail	partial?
16:48	arrdem	Bronsa: I think it's something to do with channel spam protection settings.
16:48	turbofail	i don't think partial does any code gen
16:50	andyf	inc Bronsa
16:51	andyf	How does one do that? It's been so long since I've seen an example I've forgotten.
16:51	arrdem	turbofail: functions in jvm clojure exist only as IFn classes. In order to express a partial application you have to generate a new one-off class instance which serves to wrap the partial'd arguments together with the apply machinery to insert the application arguments.
16:51	andyf	(inc Bronsa)
16:51	lazybot	⇒ 22
16:52	andyf	Not for something just now, but for enhancements and fixes to tools.analyzer(.jvm) throughout Dec and Jan :)
16:52	Bronsa	hah, thanks
16:53	gastove	Okay room, question: can you pass a namespace as an argument to a function?
16:53	andyf	Sounds like those libs are getting use by core.async and core.typed now?
16:53	turbofail	arrdem: are you sure? i don't think that's what partial actually does
16:53	turbofail	er ok it does make a new instance
16:53	turbofail	but not a new class
16:54	Bronsa	andyf: yep
16:54	andyf	gastove: Yes, you can, but that doesn't necessarily mean you can do anything you might want to be able to do with it.
16:54	turbofail	thus not really code-gen
16:54	turbofail	,(map class [(partial + 3) (partial - 5)])
16:54	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
16:54	andyf	,(the-ns 'clojure.core)
16:54	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
16:54	turbofail	interesting
16:54	Bronsa	wat
16:54	arrdem	wat
16:55	arrdem	how did map just do that..
16:55	gastove	andyf: hrm, okay. Thanks! I'm probably better off defining some sort of attribute map and loading _that_ instead.
16:56	arrdem	turbofail: you're right, because in generally you could write a single partial that just does (fn [& args] (apply f (concat partial-args args))) which makes the generated f just an instance with a seq and an ifn.
16:56	amalloy	someone probably broke all of clojurebot's eval again
16:56	amalloy	,1
16:56	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
16:56	arrdem	turbofail: so not really code gen, just instance creation.
16:56	andyf	gastove: The expression I attempted to evaluate above (the-ns 'clojure.core), successfully returns a Java object of type clojure.lang.Namespace in my REPL. What were you hoping to do with that namespace arg in the function?
16:56	amalloy	he'll recover. in the mean time, lazybot's eval is more fun anyway, right???
16:56	lazybot	amalloy: How could that be wrong?
16:56	clojurebot	Pardon?
16:59	gastove	andyf: uh, honestly, probably something that isn't idomatic Clojure (I'm newish to this). I was defining a namespace that describes several pieces of an MVC-esque model -- then in a controller, I wanted to load that namespace, then pass it to a function that checks to see if migrations have already been done for that model.
17:00	gastove	andyf: so something like (defn migrate [model] (if-not (migrated? model/table-name) (do-db-things-with model/spec))
17:00	andyf	gastove: ok. I don't think I have any experience to say whether that would be a good approach or not.
17:01	serendipity1	arrdem, For me code generation mainly comes here: I find that when I heavily adhere to SRP I often need to abstract away the whole pattern of wiring between objects. And IoC containers come somewhat short for that. What often changes when I find a better solution to the problem is also the whole pattern of communication between objects.
17:01	gastove	andfy: Honestly, I don't either :-P
17:02	serendipity1	So when the wiring/communication pattern is generated then only localized change in the generating code is required. So when I work towards more robust abstractions it also pushes me towards code generation more and more, and code generation in non-homoiconic languages, I feel, doesnt hack it
17:08	serendipity1	anyways, g'night everyone, thanks for some lisp/scheme clarifications
17:11	arrdem	Bronsa: ever consider core.typed-ing tanal? I've been kicking the idea around and I'm not sure it's possible since you want to support host local tanal AST extensions and ctyped would probably require a total AST definition.
17:17	Bronsa	arrdem: ambrose started working on annotating taj a while ago but I believe he decided core.typed is not suited for validating its ast (I may remember wrongly though)
17:18	Bronsa	arrdem: I don't know if that changed but core.typed is'nt really suited for validation of runtime emitted data, that's more of a job for something like Schema
17:18	Bronsa	isn't*
17:20	Bronsa	arrdem: also since core.typed depends on taj that would probably cause circular dependency issues
17:22	arrdem	Bronsa: the reason I say this is a typing problem is that a type system lets you say no other datastructure could ever exist where schema just says "well it looks good for now" which is much weaker.
17:22	arrdem	Bronsa: definitely WRT the circular dependencies, probably even using the new lightweight build.
18:06	noncom\|2	is java interop of form Class/staticMember resolved during runtime or compile time?
18:07	bbloom	compile time
18:08	noncom\|2	how can i take it to realtime?
18:08	noncom\|2	would (constantly Class/m) work?
18:08	bbloom	noncom\|2: the simplest thing to do is:
18:08	bbloom	#(Class/m %)
18:09	noncom\|2	but the static member is a variable, it is not a method...
18:09	noncom\|2	it can change in runtime though, so compile time does not cut it
18:09	bbloom	oh, you want a field?
18:09	noncom\|2	aha
18:09	bbloom	there are some methods on clojure.lang.Reflector
18:10	noncom\|2	no simpler way? :)
18:10	bbloom	not w/o more details about what you're trying to accomplish
18:11	noncom\|2	i have a java library which relies on a static variable. this variable is assigned after ini (it is OpenGL context). i want to have a simple function to get it like (my-ns/gl-context).
18:12	noncom\|2	the task is to make the function return the current value not the compile-time one..
18:12	bbloom	i'm not sure i follow
18:12	bbloom	is there only one such static variable? or many of them?
18:12	noncom\|2	hmm.. maybe i misunderstand something..
18:12	noncom\|2	one variable
18:12	noncom\|2	the global open gl context
18:13	bbloom	it's a public static field?
18:13	noncom\|2	yes
18:13	bbloom	calling it like a function with no arguments should just get the field
18:13	bbloom	(TheClass/theField)
18:13	noncom\|2	ok, will try now
18:15	noncom\|2	hmmm... nope. it returns a opengl context instance, just like with calling plain Class/method (w/o parens) but it does not give a desired result. maybe there is another mistake somewhere, i have to look for
18:16	noncom\|2	but, hey, (Class/field) and Class/field give seemingly same results..
18:16	noncom\|2	so i think what you adviced is right, but i have some other error in the code to look for. this was not the cause, as it looks like
18:16	bbloom	yes
18:18	noncom\|2	found it! yep was another place :) thanks for helping though!
18:30	whilo	hi, in clojure i cannot override the pr serialisation of a record, right? in cljs it is a protocol: IPrintWithWriter
18:32	bbloom	(doc print-method)
18:32	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
18:32	whilo	in clojure it is a multimethod: print-method
18:32	bbloom	,(doc print-method)
18:32	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
18:32	bbloom	&(doc print-method)
18:32	lazybot	⇒ "; "
18:32	bbloom	argh.
18:35	iwillig1	okay
18:35	arrdem	is someone ddosing the bot with REPL breaking code?
18:36	locks	bot is still broken?
18:37	arrdem	,(+ 1 2)
18:37	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: clojure.core/with-open requires an even number of forms in binding vector in clojure.core:>
18:38	arrdem	locks: I'll take that as a yes...
18:38	locks	:(
18:43	whilo	bbloom: hmm, print-method in core.clj has no docstring, do you know a way to overwrite the behaviour of a multimethod?
18:44	whilo	just curious, i am doing it wrong probably anyway
18:44	whilo	so i hopefully don't need it
18:44	whilo	(i have to rethink my current attempt)
18:44	bbloom	whilo: google it, it's kinda a weird ugly spot in clojure core, but works fine once you get it down
18:45	whilo	ok, thx. nice dendrology talk btw.
18:48	whilo	bbloom: got it, thx
18:48	bbloom	whilo: glad you liked it, thanks!
19:04	noncom\|2	does nrepl have any dependencies besides clojure and java?
19:05	noncom\|2	there is no project.clj, so i am not sure
19:05	noncom\|2	but it look sso
19:05	noncom\|2	*so
19:07	noncom\|2	looking at the sources seems like proving that
19:09	kschrader	anyone know how to use the #uuid reader with a Var?
19:10	kschrader	it works fine if I pass it a UUID String
19:10	kschrader	but as soon as I put that same sting in a Var it blows up
19:11	kschrader	Example: https://gist.github.com/kschrader/c41891bd6bc78dd2b653
19:11	noncom\|2	maybe @a?
19:12	kschrader	nope, can't deref a String
19:13	whodidthis	(def a #uuid "538ba7b2-5034-4146-a1d0-884eebfce27f")
19:13	hiredman	kschrader: reading and evaluating are different things
19:14	hiredman	when you (def a "foo") a is read in as the symbol a, but evaluated to the string "foo"
19:14	hiredman	#uuid "..." is transformed by the read (before evaluation has happened) in to a uuid object
19:14	kschrader	ah, that makes sense
19:14	kschrader	thanks hiredman
19:19	noncom\|2	kschrader: (read-string (str "#uuid \"" a"\""))
19:19	noncom\|2	works
19:20	kschrader	noncom\|2: that's helpful, thanks
19:20	kschrader	makes sense now
19:27	noncom\|2	kschrader: although i'd use something like (java.util.UUID/fromString a)
19:31	gfredericks	noncom\|2: the pom.xml isn't that hard to read
19:33	noncom\|2	gfredericks: yeah, but i never read one before :D
19:33	noncom\|2	but still it looks like there is nothing external, from whata i can say
19:33	noncom\|2	looked at the pom too
19:33	gfredericks	I think contrib projects tend to be pretty low on dependencies
19:34	gfredericks	at worst they depend on another contrib project? not 100% sure about that
19:34	noncom\|2	looks like nrepl is clear on that - nothing external
19:34	noncom\|2	this is very good for me
19:40	gfredericks	how long can GC pauses get?
19:41	gfredericks	<8GB of ram
19:52	munderwo	Hi all. Im trying to :require a defrecord from another namespace. code here https://www.refheap.com/86178
19:54	munderwo	It seems like im not doing the correct :require somewhere or :import. Any ideas?
19:59	tolstoy	munderwo: Just a guess, but might you need to (:require [raft.log :refer [->LogEntry]])
19:59	tolstoy	Or maybe just use (LogEntry. 0 0 "Foo")?
20:04	yedi	are their any clojure libs / interfaces that make weka not terrible to use
20:06	yedi	ahem, there*
20:09	munderwo	tolstoy: hmm. didnt seem to do the trick. I've got a work around for now. I tried the (LogEntry. 0 0 "Foo") as well. but same error. Not sure what the dealio is. would like to work it out for future tho.
20:18	tolstoy	munderwo: The way I've done similar things is to have a "make-log-entry" function in the original namespace.
20:18	tolstoy	munderwo: Then just use that to construct new entries.
20:20	tolstoy	The record definition stays in its own namespace, etc.
20:20	munderwo	tolstoy: ahh ok fair enough. I've got an "append" in that namespace. so I guess I could just change that around to take the vaues rather than the record. thanks!
20:20	tolstoy	munderwo: For what it's worth, I also tend to define a "protocol" that the record implements for additional functions that take that record. Then just import the protocol.
20:21	tolstoy	But that's probably over-engineering for my uses. I like to practice, though. ;)
20:21	munderwo	Yeah I looked at protocols. And it looked like It could do what I wanted. But then it didnt seem like you could specify the record that you wanted it to work on? I might have missed something though
20:22	munderwo	You could when you did the extend-protocol but not the base?
20:22	tolstoy	(defrecord Blah [p1 p2] IProtocol (fn-1 [_]) (fn-2 [_])))
20:23	tolstoy	I use it for service lifecycle stuff.
20:23	tolstoy	A protocol with "start" and "stop".
20:23	tolstoy	Then I have different records, each implement (at last) that protocol.
20:23	tolstoy	defrecord WebService or defrecord LogService and so on.
20:26	tolstoy	Each of those records held the "state". DB conns, or web-server instance, whatever. And each with a (make-web-server <port>) fn.
20:27	tolstoy	Not much better than, say, using atoms in a map. shrug. protocols seems more useful to library writers.
20:27	munderwo	fair call.
20:28	munderwo	Hey any idea why '=' doesnt work on a defrecord after persisting to disk and reading it back? (in edn format?) is it because = is memory based not value based?
20:28	munderwo	and if so how do I make it memory based? apart from a function?
20:29	tolstoy	Don't know. Maybe try == instead?
20:29	hiredman	tolstoy: == is only form numbers and the docstring for it says so
20:29	tolstoy	Oh, maybe that's number only.
20:30	hiredman	munderwo: = on defrecords considers types, which on the jvm depends on which classloader loaded the class
20:30	hiredman	in clojure that can all get very tricky
20:31	hiredman	so likely the reader is instaniating a different jvm class for your defrecord when you read it in than it has for what ever you have in memory
20:31	hiredman	which could happen a few different ways
20:32	hiredman	if you created an instance of a defrecord, (def foo (->Foo)) then reloaded the definition of Foo, new instances of your Foo record won't be equal to the one defed in foo
23:59	pcn	Has anyone else had problems with tutorial-05 in modern-cljs? I'm getting emitted code that's somewhat different from the example, but I don't know enough js to know if it's kosher