#clojure – 2015-08-16

2015-08-16

04:02	ttt_fff	has no one written a datomic clone out of respect for rich hickey ?
04:57	thearthur	ttt_fff It would also be a non trivial amount of work
05:33	wasamasa	lol, "out of respect"
05:35	oddcully	out of respect for the amount of work
05:37	wasamasa	I don't get it
05:37	wasamasa	sounds like you either want to go for duplicate effort for the heck of it or not pay for the existing solution
05:39	oddcully	guess paying is the issue. the same question was asked last week. suprisingly no contender showed up
05:41	oddcully	or maybe some demand, that such a base component must be OSS since the vendor could go out of business.
05:41	oddcully	well there will be a next time we can ask :*)
05:53	wasamasa	right
07:12	xtrntr	what does thsi error mean
07:12	xtrntr	Exception in thread "main" java.lang.RuntimeException: Unmatched delimiter: )
07:49	kwladyka_	xtrntr, you have too much or too less "(" or ")"
07:49	kwladyka	use editor with structural editing
07:50	kwladyka	like intellij idea
07:51	wasamasa	or emacs with paredit
08:14	hyPiRion	uh, so in EDN, the only way to encode the comma character (','), is by writing "\u002c"?
08:15	hyPiRion	Well I guess according to the spec, but I found it a bit strange that it's the only whitespace character without a more convenient name
09:08	pepijndevos	What is the neatest way to write a seq where each element might not be there? (remove nil? [(when x y) (when a b)])
09:48	andrewchambers	Yo, is there an equivalent of slurp for binary data?
09:48	andrewchambers	I just need a byte array
09:53	justin_smith	andrewchambers: you can make a ByteBuffer, use .put repeatedly to fill it from the source, then call .array on it to get a byte array
09:54	justin_smith	andrewchambers: or use something like apache commons-io IOUtils
09:59	hyPiRion	andrewchambers: https://github.com/ztellman/byte-streams is good
10:02	vise890	hi all. i'm messing around with core.logic and i'm trying to make a recursive goal. I'm using conde to differentiate between the recursive and base cases, but that blows the stack. is there an exclusive version of conde or am i missing something bigger?
10:02	vise890	here's the code btw: http://pastebin.com/FcPm7MtC
10:17	crocket	Exceptions feel like a mistake.
10:18	crocket	My program crashed because an http request failed due to a network problem.
10:18	crocket	I had to catch IOException in my clojure program.
10:18	crocket	I didn't see it coming.
11:54	lodin_	Is there any benchmark library that outputs relative performance, like http://perldoc.perl.org/Benchmark.html#EXAMPLES ?
11:59	sg2002	lodin_: There's https://clojuredocs.org/clojure.core/time for simple use cases.
11:59	sg2002	lodin_: For more complex you may want to look into https://github.com/ptaoussanis/timbre that has a macro called "profile".
12:00	lodin_	sg2002: I'm currently using criterium for benchmarking. I'll check out timbre.
12:02	lodin_	sg2002: timbre seems really useful for finding bottlenecks. Not really what I'm after though (but it will surely be useful later).
12:03	lodin_	sg2002: I'm guessing that it wouldn't be too hard to write a comparison function from criterium's output, but I was guessing/hoping that someone had done that already.
13:10	luxbock	is there a prettier way to do this: ##(map (fn [[k v] i] {k (assoc v :bar i)}) {:a {:foo 1}, :b {:foo 1}, :c {:foo 1}} (range)) ?
13:10	lazybot	⇒ ({:c {:bar 0, :foo 1}} {:b {:bar 1, :foo 1}} {:a {:bar 2, :foo 1}})
13:11	luxbock	my actual use case is adding :id-keys for inner maps, where the outer map is a sorted-map, so I'm doing (into (empty my-sorted-map) XXX)
13:34	Hazel_eyes	Hi
13:34	Hazel_eyes	anyone care to chat?
13:35	luxbock	I feel like I'm far too OCD about the way my code gets indented
13:35	luxbock	https://gist.github.com/luxbock/e192b33d2d578ef8b696
13:35	luxbock	clearly the second form is easier to read?
13:35	seangrov`	,(let [gs #{{:a 10}}] (get-in gs [{:a 10}]))
13:35	clojurebot	{:a 10}
13:36	luxbock	I like Clojure's let and other binding forms but I wish `cond` would group its tests and results together with an additional set of parens/braces
13:45	seangrove	,(let [ms #{{:a 10}}] (get-in ms [{:a 10}]))
13:45	clojurebot	{:a 10}
13:45	seangrove	,(let [ms #{{:a 10}}] (update-in ms [{:a 10}] assoc :b 20))
13:45	clojurebot	#error {\n :cause "clojure.lang.PersistentHashSet cannot be cast to clojure.lang.Associative"\n :via\n [{:type java.lang.ClassCastException\n :message "clojure.lang.PersistentHashSet cannot be cast to clojure.lang.Associative"\n :at [clojure.lang.RT assoc "RT.java" 785]}]\n :trace\n [[clojure.lang.RT assoc "RT.java" 785]\n [clojure.core$assoc__4130 invokeStatic "core.clj" 191]\n [clojure.cor...
13:45	seangrove	Is that obviously intuitive to other people, or does it seem like update-in should work?
13:46	seangrove	Asking because it seems convenient/nice, but there's probably a very solid reason why it doesn't work
13:52	justin_smith	seangrove: update-in works with things that have keys
13:53	seangrove	,(update-in [0 0 0] [1] inc)
13:53	clojurebot	[0 1 0]
13:54	seangrove	justin_smith: Just not sure that a vector has a stronger notion of keys than a set does ;)
13:58	justin_smith	seangrove: according to clojure, it's an associative structure and the keys are indexes
14:21	justin_smith	,(associative? [])
14:21	clojurebot	true
14:21	justin_smith	,(associative? #{})
14:21	clojurebot	false
15:09	seangrove	justin_smith: Sure, but would it make sense to implement Associative for sets?
15:10	seangrove	Seems like it'd be a win overall, but was trying to figure out if there was some guarantee that Associative demands (like Counted), that sets couldn't satisfy
15:11	justin_smith	sets are counted, maybe it's the lookup thing?
15:11	justin_smith	I mean they do membership test as if its lookup, but maybe that's not good enough?
15:13	seangrove	justin_smith: Sounds like a question for the ML! Or maybe the dev-ml
15:15	justin_smith	seangrove: what happens when you use update-in on a key and it turns into an existing key?
15:16	justin_smith	(update-in #{{:a 0} {:a 0 :b 1}} [{:a 0}] assoc :b 1)
15:16	justin_smith	I guess that's equivalent to removing a key
15:16	justin_smith	s/key/member
15:17	seangrove	justin_smith: Yeah, interesting question.
15:17	seangrove	Seems like removing a member would be intuitive enough (a bit strange perhaps)
15:19	seangrove	justin_smith: But overall, seems like a convenient, intuitive addition, no?
15:20	seangrove	Actually came across it because it would be very nice for some datomic operations
15:21	seangrove	Since all cardinality-many refs are sets
16:04	justin_smith	interesting
16:05	justin_smith	so it saves you having to use into [] for example
16:16	mcktrtl	In the clj-http docs it shows an example for using :retry-handler and says "Apache's http client automatically retries on IOExceptions" -- this means retry handler only works for some exceptions right, not all?
16:35	justin_smith	mcktrtl: it depends on what exceptions the client can throw, right?
16:35	mcktrtl	im specifically trying to retry when receiving status=504
16:36	mcktrtl	its not working so I'm assuming IOExceptions means something else but I can't find what it means
16:36	justin_smith	what kind of exception does it throw when status is 504?
16:38	mcktrtl	ExceptionInfo clj-http: status 504 clj-http.client/wrap-exceptions/fn--10592 (client.clj:196)
16:39	justin_smith	since that isn't an IOException it seems like you would have to catch ExceptionInfo and check for status 504
16:40	mcktrtl	what exactly is an ioexception out of curiosity?
16:40	justin_smith	it's a subclass of Exception
16:40	justin_smith	as is ExceptionInfo
16:40	mcktrtl	oh sorry, right, I meant like, what's a real life example of an event that would raise that
16:40	justin_smith	the idea is that if people use predictable subclasses, you can decide how to handle different errors smartly...
16:40	justin_smith	end of stream
16:41	justin_smith	lost connection
16:41	justin_smith	network not available would be another one
16:41	mcktrtl	ahh ok
16:42	justin_smith	but really, the event doesn't raise the error - somebody's code does, and we hope they are consistent enough to use IOException for events that are related to IO...
16:44	mcktrtl	hehe
19:04	slester	Is there an idiomatic way of going through a list of functions (that return either a tuple or false/nil) and exiting once a tuple is returned?
19:07	simon1234	Hi there! I want to override the ILookUp interface when defining a record, is this possible? Been browsing and googling for some times now, but I don't how if this is possible, note: this is for Clojure (not clojurescript where ILookup is a protocol and not a java interface). Thanks, cheers!
19:18	amalloy	simon1234: no. records are maps; if you could muck with ILookup they wouldn't act like maps
19:18	amalloy	if you want a custom type that's totally from-scratch you can use deftype, which doesn't act like a map unless you do it yourself
19:19	simon1234	amalloy: ok, I will give a look at deftype then, thanks! Though, it could make sense to derefine some interface to provide some specific implementation, maybe calling the (super) implementation if that makes tsense
19:50	justin_smith	slester: (first (keep #(%) [f g h ...])) - though this is not guaranteed not to run any functions after the first non-nil result
19:52	slester	justin_smith, interesting, thanks. I may end up having to refactor it to use something else, not sure how to tackle my issue haha
19:52	justin_smith	slester: (reduce (fn [_ f] (let [x (f)] (if (nil? x) nil (reduced x)))) nil [f g h ...]) this one would only call each function once, and not call any after the first non-nil result, but it's a bit uglier
19:53	justin_smith	slester: the difference is that you can't ever guarantee non-execution with laziness, so this version does not use laziness
20:19	crocket	How should I handle exceptions in clojure? Can I avoid it?
20:33	justin_smith	crocket: try/catch and I guess you can avoid code that throws exceptions?
20:33	crocket	justin_smith, I don't like exceptions
20:33	justin_smith	but some stuff is going to hit build ins that throw exceptions (like calling nil, or using a numeric method on nil)
20:34	justin_smith	crocket: the people who designed the vm don't agree
20:34	crocket	I used clj-http, and it threw a subclass of IOException which I didn't see coming.
20:34	crocket	Now, I'm looking for a replacement of clj-http that doesn't throw exceptions for network problems.
20:35	justin_smith	yes, I think clj-http throws stupid exceptions
20:35	crocket	Because clj-http uses an apache component which throws stupid exceptions.
20:35	crocket	http kit looks promising although it seems to throw Error and RuntimeException.
20:35	justin_smith	crocket: well, network problems are a different issue - the underlying network lib throws an exception if there is a network level error
20:35	crocket	like no route to host?
20:36	crocket	My program crashes because there was no route to host at the moment.
20:36	crocket	Very stupid
20:36	crocket	crashed*
20:36	justin_smith	crocket: the easy answer is never use clj-http outside of a try/catch block
20:36	crocket	I'm not sure what kind of exceptions clj-http will throw.
20:36	justin_smith	then catch all of them
20:37	crocket	Even, OutOfMemoryException?
20:37	justin_smith	there's no such thing, that's an error
20:37	crocket	There are many exceptions that you shouldn't catch.
20:37	justin_smith	no, there are errors you should not catch
20:37	justin_smith	there are no exceptions you should not catch
20:37	justin_smith	taht is the difference between exceptions and errors
20:37	crocket	OutOfMemoryException is a subclass of RuntimeException which is a subclass of Exception.
20:37	justin_smith	there is no such thing as OutOfMemoryException
20:37	justin_smith	it does not exist
20:38	justin_smith	you are thinking of OutOfMemoryError
20:38	crocket	Ok
20:38	justin_smith	there is no exception that is unsafe for catching
20:38	crocket	So, is it ok to catch just Exception?
20:38	crocket	Not sure, if it's ok to catch RuntimeException.
20:38	justin_smith	right, and that catches all subclasses, including the ones clj-http throws
20:39	justin_smith	crocket: this is the difference between Exception and Error. Catching RuntimeExceptions is fine http://docs.oracle.com/javase/7/docs/api/java/lang/RuntimeException.html
20:40	justin_smith	crocket: if you look at the known subclasses on that page, none of those are unsafe to catch and handle
20:40	crocket	ok
20:41	crocket	Do you know better error handling methods?
20:41	crocket	I mean theoretically
20:41	justin_smith	sure, Haskell has a nice setup, even common lisp has recoverable conditions
20:42	crocket	https://docs.oracle.com/javase/8/docs/api/java/lang/NullPointerException.html is a RuntimeException.
20:42	crocket	In theory, I can just catch and log NPE.
20:42	justin_smith	right
20:42	crocket	It doesn't seem to be a good idea to keep running after NPE.
20:42	justin_smith	?
20:43	justin_smith	there's all sorts of reasonable cases for catching and handling an NPE
20:43	crocket	NPE indicates a programmer error that you should fix.
20:43	justin_smith	for example if I try to make a number out of some part of a response from a server
20:43	justin_smith	or if I load corrupted data from disk
20:45	crocket	not a valid reason to throw NPE
20:45	justin_smith	crocket: but it's a valid reason to get one given the tools we have
20:51	crocket	means it is a mistake in java.
20:52	justin_smith	sure, the whole thing is set up under the premise that exceptions are a useful way to cut short failed code paths, which is obviously something you disagree with
20:53	justin_smith	crocket: I'm going to try to remember the source, but basically there is a basic method in the jvm, for like counting array elements or finding an item, that works by iterating until it finds what it is looking for, then throwing an exception to break out of the loop
21:19	quizme	hi, i have some clojure code that I would like to execute from clojure. i guess the simplest naive thing to do would be execute the python from the command line and block until the script finished ?
21:21	weebz	quizme do you mean execute some python from clojure?
21:21	quizme	weebz yah
21:21	quizme	weebz what's the best way to orchestrate that? it's a job that needs to run everyday
21:22	quizme	weebz or maybe i should make a daemon in both clojure an python to read from a common redis work queue
21:23	weebz	I can't speak for best practice as I'm still somewhat new to clojure, but if you're you're just firing off a python script you can use clojure.java.shell
21:23	weebz	https://clojuredocs.org/clojure.java.shell/sh
21:24	quizme	weebz yeah ...
21:24	quizme	weebz ideally i would like to be able to "start from where it left off" cuz there might be a lot of items to process
21:25	quizme	weebz and have some web page to see what the status of the jobs is.
21:26	weebz	quizme I'm not really sure what you're specifically getting at. How to communicate back and forth between python and clojure?
21:26	weebz	Or how to orchestrate everything?
21:28	quizme	i would like to be able to spawn multiple python workers to make it more scalable.
21:30	quizme	so the multiple workers should not be working on th same job, and it should be able to communicate to the clojure process to do what needs to be done next.
21:30	quizme	basically i have a pipeline of steps. some of the steps are done in python and some of the steps are done in clojure.
21:32	quizme	for example, say i have a lot of users and each user has a lot of widgets, and my pipeline has to process all widgets of all users, and the pipeline consists of N steps. C steps are done in Clojure and P steps are done in Python, C + P = N.
21:33	TEttinger	quizme: might be a good use case for Jython, that JVM version of Python (specifically Python 2 I think)
21:34	TEttinger	since then you could call back and forth in the same program
21:34	quizme	Tettinger yeah... true....
21:35	quizme	Tettinger that's not a bad idea
21:35	TEttinger	if you need to use something like SciPy or NumPy it probably doesn't have those
21:35	TEttinger	it might, I dunno
21:35	quizme	Tettinger oh yeah i need NumPy
21:36	TEttinger	I'll check if there's anyone who's already done something like this...
21:36	quizme	i don't know of any worker queue libraries that work both in python and clojure, but i don't want to get too complicated and bump up to RabbitMQ or anything like that.
21:37	TEttinger	oh boy. this is definitely in the "little bit complicated" realm, but http://codespeak.net/execnet/
21:38	TEttinger	you could use this to communicate from your NumPy normal Python to a teensy Jython setup that's really part of your clojure app
21:38	TEttinger	(like lazybot has a JRuby plugin, even though lazybot is 100% clojure)
21:39	TEttinger	$jruby 3 ** 4
21:39	TEttinger	$jr 3 ** 4
21:39	TEttinger	hm, might not be enabled
21:42	TEttinger	quizme: there's also ZeroMQ, which has bindings everywhere, typically pretty idiomatic ones for modern languages like Clojure and Python. it isn't a messaging queue like RabbitMQ, and is probably better described as an improvement over plain sockets.
21:42	TEttinger	I think it's common these days for RPC-type stuff like you're talking about
21:43	quizme	tettinger hmm ...
21:47	TEttinger	communicating between VMs and other VMs or from VMs to non-VMs is traditionally a hard problem.
21:49	TEttinger	some languages are meant to support a specific kind of communication, like Lua being embedded in C (but C++ is less easy), or Clojure being able to call any JVM language. there's some weird experiments that the LLVM people did to run both Java and C# on one VM, and IKVM lets you compile JVM stuff into DLLs that .NET langs like C# can use
21:49	TEttinger	the first two examples work great! the second two, uh not so much
21:50	crocket	Interprocess communication tends to be independent of language.
21:53	TEttinger	crocket: yeah, I mean there's plain old TCP as an option for sending stuff between processes on different machines. the question is really I think what format the data should be in when the communication happens?
21:53	crocket	There are libraries for that
21:53	crocket	Apache avro, protocolbuffer, and so on
21:54	TEttinger	msgpack, thrift yeah
22:35	slester	justin_smith, is there any reason `(reduced (list 1 2 3))` would just return `1`? Pretty confused right now
22:37	slester	https://github.com/slester/amiss/blob/master/src/amiss/core.clj#L408-L415 is somehow returning just :soldier
22:39	amalloy	why on earth would you use a reduce and ignore the accumulator parameter to return either (reduced x) or false, instead of just using (first (filter f xs))?
22:41	slester	'why on earth' is a bit harsh
22:43	slester	because I still want it to return false not '()
22:43	slester	and I need it to short-circuit when it has found something
22:44	amalloy	first and filter short circuits, of course. that's what laziness is for
22:45	slester	and if none match?
22:46	amalloy	then you get nil. you can convert nil to false, if that's important to you
22:47	amalloy	,(first (filter even? (range)))
22:47	clojurebot	0
22:47	amalloy	,(first (filter even? [1 3 5]))
22:47	clojurebot	nil
22:47	amalloy	,(or (first (filter even? [1 3 5])) false)
22:47	clojurebot	false
22:50	amalloy	but since nil is treated the same as false in boolean contexts you can usually just leave it alone
22:50	slester	ok, it's converted, but that doesn't change that I'm getting out just a keyword instead of a list somehow
22:56	AWizzArd	Idea? [(class x) (class z)] => [[B [B] and (= (seq x) (seq z)) => true but (= x z) => false
22:57	AWizzArd	Two byte arrays with the same contents and the same order, but they are not =
22:58	AWizzArd	,(= (byte-array 0) (byte-array 0))
22:58	Bronsa	,(= (object-array [1 1]) (object-array [1 1]))
22:58	clojurebot	false
22:58	clojurebot	false
22:58	Bronsa	clojurebot: hello?
22:58	clojurebot	BUENOS DING DONG DIDDLY DIOS, fRaUline Bronsa
22:58	Bronsa	lol
22:59	Bronsa	AWizzArd: arrays compare via identity
22:59	AWizzArd	Seems so. Didn’t expect that.
23:00	Bronsa	AWizzArd: that's a java-ism
23:01	Bronsa	,(equals (object-array [1 2]) (object-array [1 2]))
23:01	clojurebot	#error {\n :cause "Unable to resolve symbol: equals in this context"\n :via\n [{:type clojure.lang.Compiler$CompilerException\n :message "java.lang.RuntimeException: Unable to resolve symbol: equals in this context, compiling:(NO_SOURCE_PATH:0:0)"\n :at [clojure.lang.Compiler analyze "Compiler.java" 6704]}\n {:type java.lang.RuntimeException\n :message "Unable to resolve symbol: equals in t...
23:01	Bronsa	,(.equals (object-array [1 2]) (object-array [1 2]))
23:01	clojurebot	false
23:04	slaterr	,'(1,2,3)
23:04	clojurebot	(1 2 3)
23:04	nakiya	hello
23:05	crocket	How does clojure cope with event callbacks without continuation?
23:16	slaterr	are there functions like this?
23:17	slaterr	(1 2 3) => (() (1) (1 2) (1 2 3))
23:17	slaterr	(1 2 3) =>((1 2 3) (2 3) (3) ())
23:17	crocket	A power set?
23:18	crocket	That looks similar to a power set.
23:18	slaterr	I think it is
23:19	slaterr	well actually it isn't
23:20	luxbock	#(take (inc (count %)) (iterate rest %)) would work for the second
23:20	crocket	That seems to involve recursion.
23:22	slaterr	,(iterate rest '(1 2 3))
23:22	clojurebot	((1 2 3) (2 3) (3) () () ...)
23:23	slaterr	,(iterate butlast '(1 2 3))
23:23	clojurebot	((1 2 3) (1 2) (1) nil nil ...)
23:25	crocket	一、二、三、四.....
23:27	slaterr	not sure about an elegant way of solving the first. without busting out manual recursion
23:32	TEttinger	,(reductions (fn [coll _] (rest coll)) '(1 2 3 4) '(1 2 3 4))
23:32	clojurebot	((1 2 3 4) (2 3 4) (3 4) (4) ())
23:34	slaterr	neat
23:36	TEttinger	,(reductions (fn [_ ct] (take ct '(1 2 3 4))) (range 5))
23:36	clojurebot	(0 (1) (1 2) (1 2 3) (1 2 3 4))
23:36	TEttinger	there's probably a much better way to do the second
23:36	TEttinger	,(reductions (fn [_ ct] (take ct '(1 2 3 4))) () (range 1 5))
23:36	clojurebot	(() (1) (1 2) (1 2 3) (1 2 3 4))
23:40	nakiya	,(+ 1 2)
23:40	clojurebot	3
23:40	slaterr	,(reduction + '(1 2 3 4 5)
23:40	clojurebot	#<RuntimeException java.lang.RuntimeException: EOF while reading>
23:40	slaterr	,(reductions + '(1 2 3 4 5)
23:40	clojurebot	#<RuntimeException java.lang.RuntimeException: EOF while reading>
23:40	slaterr	,(reductions + '(1 2 3 4 5))
23:40	clojurebot	(1 3 6 10 15)
23:41	TEttinger	,(reductions conj () '(1 2 3 4))
23:41	clojurebot	(() (1) (2 1) (3 2 1) (4 3 2 1))
23:41	TEttinger	hm
23:41	TEttinger	,(reductions conj [] '(1 2 3 4))
23:41	clojurebot	([] [1] [1 2] [1 2 3] [1 2 3 4])
23:41	TEttinger	there we go
23:42	TEttinger	that would be the better way
23:42	slaterr	what is the difference between () and [] there?
23:42	slaterr	looks good
23:42	palinkas	hi guys, I have an interesting problem
23:42	TEttinger	() is a seq, can be created by calling seq on any other collection or by calling list with some elements, or even with '(1 2 3) if you want everything in it quoted
23:43	palinkas	I want to run multiple clojure runtimes in a single JVM
23:43	TEttinger	[] is a vector, can be created with [] literals like [1 2 3] , with (vector 1 2 3), or by converting using vec .
23:43	justin_smith	amalloy: the reason for trying reduce after giving a lazy answer was in case there was a hard requirement that further functions not be called
23:43	palinkas	one basic app, that can change behaviour dynamicly and I want to switch between the different version of the logic
23:44	palinkas	the logic won't be just a single namespace -> that's why I'd need multiple runtimes
23:44	TEttinger	seqs append to the beginning with conj, and do not support indexing by element efficiently (you can use nth to get an element, but it will be slower than a vector), but can be lazy and potentially infinite if so desired.
23:45	palinkas	I have also want to share some of the basic infrastructure between those clojure runtime (common classes, common logic implemented in java)
23:46	justin_smith	palinkas: there are experimental things for doing multiple clojures in one vm, I don't know how reliable any of them are. If you can find a way to do it via protocols instead (two different implementations of one protocol, switching out the implementation to get your optional behaviors), that will likely be much more robust.
23:46	TEttinger	vectors append to the end with conj, can be used as functions with an integer arg to get that element, like ([10 20 30] 0) returns 10, and tend to be more common in idiomatic clojure when describing data than quoted lists
23:46	TEttinger	make sense, slaterr?
23:47	TEttinger	there's a bit of trickiness regarding a seq and a lazyseq, which you will no doubt discover when you try to turn a lazyseq into a string
23:47	slaterr	yes. it adds an element at the point where it makes sense for the collection type
23:47	justin_smith	palinkas: protocols were made for precisely the scenario of being able to swap out different implementations without changing your code (except the part where the object implementing that protocol is introduced)
23:47	TEttinger	yep
23:47	TEttinger	and since seqs are linked lists, appending to the beginning makes sense
23:48	palinkas	my basic idea was: the users write some clojure files (clj namespaces), that can depend on each other (in some cases)
23:48	TEttinger	vectors are array-like I think. but I think that's more useful as a way to understand them than as their actual internals
23:48	justin_smith	palinkas: so you want a sandbox?
23:49	justin_smith	there are definitely ways to have sandboxed clojure envs
23:50	palinkas	so, it can happen that: v1 (ns.a - v1 , ns.b - v1 (requires ns.a should use v1)) -> user changes stuff -> v2 ( ns.a - v2 , ns.b - v3 (requires ns.a, should use v2 ))
23:50	palinkas	justin_smith: yes, pretty much
23:50	justin_smith	palinkas: because without a sandbox you are effectively giving root to every user
23:51	TEttinger	(important note, slaterr: vectors, seqs, the {} maps that should not be confused with the fn map, the #{} sets, and a few others like the sorted variants of maps and sets, are all immutable, so a function that is given one as an argument won't change the original after the function resolves, though that function could return a modified copy)
23:51	justin_smith	palinkas: you could probably do this with multiple clojail instances https://github.com/Raynes/clojail
23:51	crocket	(doc reductions)
23:51	clojurebot	"([f coll] [f init coll]); Returns a lazy seq of the intermediate values of the reduction (as per reduce) of coll by f, starting with init."
23:51	palinkas	justin_smith: thx, I'll check it out
23:52	crocket	TEttinger, () is a list.
23:52	justin_smith	,(list? ())
23:52	clojurebot	true
23:52	crocket	seq is an abstraction that a list and a vector implements.
23:52	crocket	,(seq? [])
23:52	clojurebot	false
23:52	crocket	ouch
23:53	TEttinger	crocket: heh
23:53	justin_smith	yeah, () is also a seq
23:53	justin_smith	,(seq? ())
23:53	clojurebot	true
23:53	TEttinger	I know what I'm talking about here, crocket.
23:53	TEttinger	,(let [coll (range 4)] (do (pr coll) (print " ") (map inc coll) (pr coll)))
23:53	clojurebot	(0 1 2 3) (0 1 2 3)
23:54	TEttinger	slaterr: coll doesn't change despite map returning a modified copy
23:55	slaterr	yes I get that. but I am still trying to decypher your message. maps, a collection, is different than map a function, is that what you meant? no clue what #{} is, or sorted variants of maps and sets
23:56	justin_smith	that's why I try to use hash-map to talk about the collection type
23:57	justin_smith	(type #{})
23:57	justin_smith	,(type #{})
23:57	clojurebot	clojure.lang.PersistentHashSet
23:57	justin_smith	it's a set
23:59	TEttinger	cool stuff time!
23:59	TEttinger	,(map inc '(1 2 3))
23:59	clojurebot	(2 3 4)