2013-09-23
| 00:00 | tufflax | allenj12: yes, but you said in the comment "sum up all the distances in board, as well as the board itself" |
| 00:00 | tufflax | what does that mean |
| 00:00 | allenj12 | tufflax: o sorry my bad i just meant that the board should be passed into the manhatten distance function aswell |
| 00:00 | allenj12 | tufflax: that was poor wording |
| 00:00 | allenj12 | tufflax: sorry |
| 00:00 | tufflax | allenj12: np :P |
| 00:10 | allenj12 | tufflax: the addition of map got me answer of 14N :P still trying to get around it |
| 00:11 | tufflax | allenj12: Is 14N not good? :P |
| 00:11 | tufflax | I don't understand what you mean by "get around it" |
| 00:12 | allenj12 | tufflax: well i worked it out the answer should 21 (or 19 my partners work :P) and idk where the N came from lol :P :P |
| 00:12 | coventry | Is there a right way to say (ns foo (:require [x.core :only [foo] :rename {foo bar}]))? |
| 00:13 | tufflax | allenj12: N means that it's a BigNumber, you can mostly ignore that. I don't know where it came from either |
| 00:13 | tufflax | But obviosly you have miscalculated or the code is wrong in some other way then |
| 00:13 | tufflax | if it should be 21 or 19 :p |
| 00:14 | allenj12 | tufflax: hmm kk yea ill double check my other code |
| 00:14 | tufflax | Or maybe BigInteger, you can check with (class the-number) |
| 00:15 | tufflax | ,(class 1N) |
| 00:15 | clojurebot | clojure.lang.BigInt |
| 00:15 | tufflax | allenj12: there ^ |
| 00:16 | allenj12 | tufflax: ty im checking some numbers using manhatten distance manually it seems to be mostly right |
| 00:17 | allenj12 | tufflax: some are off tho |
| 00:17 | TEttinger | did somebody say manhattan distance? |
| 00:18 | marcopolo | so http://sprunge.us/VaBV doesn't appear to work with the master branch of cljs, let me try running nolen's core.async tests on it |
| 00:18 | TEttinger | ah, you mentioned A* earlier |
| 00:18 | TEttinger | allenj12, is this open source? |
| 00:18 | allenj12 | Tettinger: why yes we did apparently there is something wrong with our method of calculating it :P |
| 00:18 | TEttinger | I already have an open source implementation of dijkstra pathfinding, if you're interested |
| 00:19 | TEttinger | I could honestly use A* since it's faster when you have a ton of options |
| 00:19 | allenj12 | Tettinger: just an intro to ai hw im picking up clojure while i do this which i smaking it a little harder :X |
| 00:19 | TEttinger | oh ok |
| 00:19 | allenj12 | Tettinger: :D |
| 00:20 | TEttinger | well I'd love to take a look when you're done, I'm not aware of an A* impl in Clojure yet |
| 00:20 | TEttinger | hm |
| 00:20 | TEttinger | apparently the only difference between A* and dijkstra is the heuristic |
| 00:20 | hfaafb | is there any semantic or mechanical difference between calling (:key foo) and (foo :key) where foo is a map? |
| 00:21 | brehaut | hfaafb: 'object/record' maps vs 'associative' maps |
| 00:21 | allenj12 | Tettinger |
| 00:21 | TEttinger | yes? |
| 00:21 | clojurebot | yes isn't commutative |
| 00:22 | allenj12 | Tettinger: apparenlty our method of man dis is wrong.. ill show you what i got so far tho for A* the search is optimal if you have an admissable heuristic |
| 00:22 | allenj12 | Tettinger: thats all i know tho |
| 00:22 | allenj12 | Tettinger: as far as comparing between the 2 |
| 00:23 | brehaut | hfaafb: keyword first slightly indicates an objectish/record map , whereas keyword second slighly indicates an associative thing that happens to have keywords as keys. |
| 00:23 | TEttinger | ok, sure, I'll pastebin up mine if you want it. that's just what someone mentioned in a gamedev channel |
| 00:23 | brehaut | hfaafb: if you were using core.typed, the former would be HMaps, the latter (Map keyword any) (or whatever value) |
| 00:24 | allenj12 | Tettinger: https://www.refheap.com/18908 |
| 00:24 | allenj12 | also if anyone has a new to to calculate manhatten distance on a eight puzzle using a list then that would be helpful XD |
| 00:25 | TEttinger | eight puzzle? |
| 00:25 | hfaafb | thank you brehaut |
| 00:26 | allenj12 | Tettinger: yea a common puzzle of tile you can slide around if you googlde it you should get an idea |
| 00:27 | TEttinger | oh yeah that |
| 00:29 | TEttinger | allenj12, Manhattan distance is for 2D planes, right? I'm not sure what the list represents, being 1D. is it the standard 1d collection representing multiple dimensions thing, where indices 0-2 are the first row, 3-5 are the second row...? |
| 00:29 | allenj12 | Tettinger: usually in other functional languages i know it is standard so i thought i would try it it |
| 00:30 | TEttinger | yeah, I just mean... there's a few ways to do this |
| 00:30 | TEttinger | oh I get it |
| 00:31 | TEttinger | yeah I see the / and mod 3 |
| 00:33 | TEttinger | found it |
| 00:33 | TEttinger | (apply + (map #(manhatten-distance board %) board)) |
| 00:34 | TEttinger | it's finding the distance of the board to itself, allenj12 |
| 00:34 | TEttinger | agh nvm |
| 00:34 | TEttinger | didn't see it was checking goal in the function |
| 00:45 | TEttinger | allenj12, it seems like it acts correctly, I'm not sure what's up |
| 00:45 | TEttinger | it might be the usage of floor? |
| 00:46 | TEttinger | nvm, found the bug |
| 00:46 | allenj12 | Tettinger: for my test case at the bottom (manhatten-distance testt 3) it should not be 0 |
| 00:47 | allenj12 | Tettinger: yea |
| 00:52 | TEttinger | got it |
| 00:55 | TEttinger | http://ideone.com/zix3tO allenj12, the issue was only one abs was used, you needed to abs both terms |
| 00:58 | TEttinger | uh crud, a bad version went to ideon |
| 00:58 | allenj12 | Tettinger: omg lol ty |
| 00:59 | allenj12 | Tettinger: hmm now im getting 22 |
| 00:59 | allenj12 | juki |
| 00:59 | TEttinger | haha |
| 01:07 | ddellacosta | (in cljs, if it makes a difference:) why do I get an error with the message "no matching clause…" in a condp if I have ":else nil" at the end? The whole point is that I don't care what happens if I haven't matched my previous conditions. |
| 01:09 | ddellacosta | never mind, I guess I'm confusing how condp and cond work. Making it simply "nil" did the trick. |
| 01:28 | callen | ddellacosta: case didn't fit your problem? |
| 01:34 | ddellacosta | callen: well, I had a predicate expression I was applying values to, so condp seemed to fit here. But not sure it was the best choice. |
| 01:38 | callen | sight unseen, hard to say. Seems reasonable at first blush. |
| 01:52 | hfaafb | i'm a fairly new clojure programmer and would like to put my toes in the water, if anyone has any comments on some code i wrote i would appreciate it :) https://github.com/samuelcferrell/chjess/blob/master/src/chjess/core.clj |
| 01:55 | mtp | i'm fairly new to reading clojure code but it looks like something i'd write if i were any good at chess |
| 01:56 | TEttinger | ditto, mtp |
| 01:56 | mtp | (if i wrote it, i'd probably think it needs a bit more abstraction, too) |
| 01:56 | TEttinger | I have no idea what it does, but it seems clean |
| 01:57 | mtp | and i'd start by naming the (mapcat (partial x board) (filter blah)) pattern i keep seeing |
| 01:58 | mtp | and also by clarifying the part where it says ;;promotion check, perhaps by naming the edges of the board |
| 02:01 | metellus | you could replace (enemy ...) with a map {:black :white :white :black} |
| 02:01 | metellus | although I don't know that you gain anything from that... |
| 02:03 | Jarda | hmm |
| 02:03 | hfaafb | metellus: that makes a lot of sense! thanks |
| 02:04 | Jarda | so if I have a function name 'author-name' and it takes an id as argument, uses a db to fetch the author name and returns it |
| 02:04 | Jarda | then I guess it isn't a 'pure function' |
| 02:05 | Jarda | as the return value isn't necessarily the same even with the same argument value |
| 02:05 | Jarda | because of introduction of state, I guess |
| 02:06 | Jarda | is this a real problem and if so, how do we handle this? |
| 02:06 | SeanCorfield | yup, databases are a big ol' pot of state |
| 02:06 | SeanCorfield | but you can isolate it to the edges of your program and keep the rest of it pure |
| 02:06 | Apage43 | alternatively you could model it as a function of 2 args, an id, and a snapshot of a database |
| 02:07 | SeanCorfield | that's the Datomic approach - database as value |
| 02:08 | Jarda | btw which database would you recommend? I'm used to couchdb and postgresql and this current project I'm learning clojure for can be described woth with relations or without |
| 02:09 | SeanCorfield | regular sql db stuff is a bit messy tho'... you can reason about it as a set of transformations on data that are snapshotted to disk in between times, if that helps isolate state changes from pure functions |
| 02:09 | SeanCorfield | Jarda: fwiw, i use both mysql and mongodb with clojure |
| 02:10 | SeanCorfield | all the sql DBs are "equal" in some respect, since you ultimately use clojure.java.jdbc to interact with any of them |
| 02:10 | SeanCorfield | for the noSQL DBs, you'll have a separate library for each... i'm only familiar with mongodb (and congomongo) |
| 02:10 | Jarda | ok |
| 02:11 | SeanCorfield | there's not yet any standardization in the noSQL world... |
| 02:11 | Jarda | Apage43: with cluctch? |
| 02:11 | SeanCorfield | but then there are four distinct types of noSQL DB :) |
| 02:11 | Apage43 | no, I mean, I've actually worked *on* CouchDB =P |
| 02:11 | Jarda | Apage43: oh :D |
| 02:11 | SeanCorfield | clutch is supposed to be really good for couchbase |
| 02:11 | Apage43 | When I use it from clojure I just use clj-http though |
| 02:12 | Apage43 | Couchbase doesn't support the CouchDB api |
| 02:12 | Apage43 | if we have anything that points to clutch, I need to get it removed |
| 02:12 | Jarda | clutch doesn't support couchbase AFAIK |
| 02:12 | Apage43 | correct |
| 02:13 | Apage43 | rather anything built for CouchDB doesn't. |
| 02:13 | SeanCorfield | oh... i didn't realize they were so different? |
| 02:13 | SeanCorfield | that seems... misguided... |
| 02:13 | Apage43 | Couchbase is Membase (clustered memcached thingy) with CouchDB's view engine |
| 02:13 | SeanCorfield | how confusing... |
| 02:14 | Apage43 | so K/V is over memcached protocol |
| 02:14 | Jarda | Couchbase is missing the HTTP API completely |
| 02:14 | Jarda | the HTTP API of Apache CouchDB that is |
| 02:15 | SeanCorfield | CouchDB sounded interesting if you needed auto-versioning (which we didn't, which is why we went with MongoDB) |
| 02:15 | Apage43 | at any rate, as the NoSQL things go, they're all pretty different in terms of what they provide |
| 02:15 | Apage43 | CouchDB's revisions can't be used for version tracking anyway |
| 02:15 | Apage43 | they're for conflict detection only |
| 02:15 | Jarda | Apage43: you can pretty much use them for versioning though |
| 02:16 | Apage43 | It doesn't make any guarantee that they'll stick around though |
| 02:16 | SeanCorfield | I've heard a lot of people push CouchDB for versioned content... but never dug deep myself |
| 02:16 | Jarda | (if you realize that the revisions might be lost at certain circumstances) |
| 02:17 | Apage43 | Apache CouchDB has an attachments feature which kind of works well for archiving things hanging off a document |
| 02:17 | Apage43 | but the actual "revisions" are for detecting whether updates are in conflict during replication |
| 02:17 | Jarda | I'd like to try Neo4j, not that I'd need a graph database. Just to make myself familiar with it :) |
| 02:18 | SeanCorfield | I had a play with it during a Neo4J preso at a Java UG... using the Clojure REPL... It's definitely an interesting approach |
| 02:19 | Apage43 | they perform a similar job to vector clocks in Riak |
| 02:21 | callen | you couldn't pay me to use CouchDB again. |
| 02:22 | callen | bad experience using it with analytics data. CouchBase I might consider. |
| 02:22 | callen | that was years ago, so it might've gotten better. I have scars though. |
| 02:22 | Apage43 | I haven't followed actual CouchDB for a couple years, so I couldn't speak to what's happened in the meantime |
| 02:24 | Apage43 | I think the most interesting thing in that space to me are the CouchDB-compatible mini implementations, like PouchDB and TouchDB (now called Couchbase Lite, but actually *is* a CouchDB API compatible thing.. which is confusing.. |
| 02:26 | rhg135 | Is there a simple way to get a resource in cljs+nodejs like clojure.java.io? |
| 02:27 | callen | rhg135: JS doesn't have resources. |
| 02:27 | callen | rhg135: use the node.js file APIs. |
| 02:28 | rhg135 | Figured |
| 02:28 | rhg135 | K |
| 02:28 | TEttinger | Jarda, graph DBs are interesting. they seem to have a pretty limited group of people who actually use them though |
| 02:30 | callen | They're not super-duper practical for common-case stuff. |
| 02:30 | s4muel | i am interested in their application to complex software systems rather than people. service dependencies, infrastructure, etc. looking at bitsy, an embedded graph db atm |
| 02:30 | callen | and in most cases where you're actually storing a graph, it's often faster to just use adjacency lists and caching. |
| 02:31 | Apage43 | mhm. Graphs are hard. |
| 02:32 | Apage43 | Especially big ones. |
| 02:45 | allenj12 | can someone help me with this now? https://www.refheap.com/18909 i posted the question commented in my code it has to do with priority-maps |
| 02:47 | Ember- | and your problem in it is? |
| 02:48 | Ember- | ah, comment |
| 02:48 | allenj12 | Ember-: yea i thought it would be better to ask in there |
| 02:48 | Ember- | path finding stuff it seems |
| 02:49 | Ember- | so that "list of possible moves" should be the path found, right? |
| 02:50 | Apage43 | allenj12:I think you might be looking for the frequencies function |
| 02:50 | Apage43 | ,(frequencies [:up :up :right :right :right]) |
| 02:50 | clojurebot | {:up 2, :right 3} |
| 02:51 | allenj12 | no i will have at most only one oeration as it lists only posibble moves no repeats |
| 02:51 | Ember- | so just the first step of the found path? |
| 02:51 | Apage43 | ah, I've misinterpreted |
| 02:52 | allenj12 | so options lists a set of possible moves.. i wanna score how well those moves do and map the score with that move |
| 02:52 | allenj12 | so |
| 02:52 | Ember- | ah |
| 02:52 | allenj12 | if :up makes a manhatten distance of 3 then i want (:up 3) |
| 02:53 | Apage43 | ahh |
| 02:53 | allenj12 | so the priority map can sort it by the value and i can apply the operation in the next function |
| 02:53 | allenj12 | by getting the key somehow :P |
| 02:54 | allenj12 | sry if it was not clear enough |
| 02:56 | Apage43 | (into (priority-map) (for [move options] [move (manhattan-distance (move-tile move board) goal)])) |
| 02:56 | Apage43 | pm/priority-map i guess in your case |
| 02:56 | allenj12 | hmm ill give it a shot |
| 02:56 | Apage43 | in general you can conj [k v] pairs into maps |
| 02:56 | Apage43 | and into conj's a list of things into an existing structure |
| 02:57 | Apage43 | ,(into {} [[:a 1] [:b 2]]) |
| 02:57 | clojurebot | {:a 1, :b 2} |
| 03:13 | allenj12 | Apage43: getting close it seems but not quite there https://www.refheap.com/18910 havin some trouble |
| 03:15 | allenj12 | Apage43: not sure what that error means at the bottom imma keep playing with it |
| 03:15 | Apage43 | it means (find-index) got called with a number |
| 03:16 | Apage43 | somehow, and it wasn't expecting to |
| 03:16 | Apage43 | allenj12: It looks like you got your arg order mixed up on line 38 |
| 03:17 | Apage43 | or something |
| 03:17 | Apage43 | but move-tile is calling (find-index) with a number for its "board" arg, which I think is wrong |
| 03:21 | allenj12 | Apage43: kk ill snoop around |
| 03:29 | allenj12 | Apage43: very weird im using light table and it seems everything is going where it needs to be as i call move-tile with board first and move-tile has board first |
| 03:30 | Apage43 | I'm just reading "space (find-index 0 board)" and "find-index [board tile]" |
| 03:30 | Apage43 | which looks like that always gets called with 0 as board and board as tile |
| 03:31 | Apage43 | which is also backwards from how its called in possible-move? |
| 03:31 | Apage43 | compare lines 38 and 50 in your last refh |
| 03:31 | Apage43 | https://www.refheap.com/18910#L-38 |
| 03:34 | allenj12 | Apage34: ah yes silly but now im getting a new error again ill keep looking but ill repost it too |
| 03:35 | allenj12 | Apage34: https://www.refheap.com/18911 |
| 03:38 | Apage43 | allenj12: if you use assoc with indices on board, or use board as a function, it is important that board is a vector not a list |
| 03:38 | Apage43 | ,([1 2 3 4] 0) |
| 03:38 | clojurebot | 1 |
| 03:38 | Apage43 | ,('(1 2 3 4) 0) |
| 03:38 | clojurebot | #<ClassCastException java.lang.ClassCastException: clojure.lang.PersistentList cannot be cast to clojure.lang.IFn> |
| 03:39 | allenj12 | Apage34: isnt there a nth function? would that work? |
| 03:40 | Apage43 | That works in place of using it as a function, but you still won't be able to assoc onto it |
| 03:40 | Apage43 | &(doc nth) |
| 03:40 | lazybot | ⇒ "([coll index] [coll index not-found]); Returns the value at the index. get returns nil if index out of bounds, nth throws an exception unless not-found is supplied. nth also works for strings, Java arrays, regex Matchers and Lists, and, in O(n) time, for sequences." |
| 03:41 | allenj12 | Apage43: interesting |
| 03:41 | Apage43 | but in general if you're doing random-access style stuff you should use vectors over lists |
| 03:46 | Viesti | hmm, what's the idiomatic way to have a named constant? (don't like to use "magic numbers") |
| 03:47 | allenj12 | Apage43: IT WORKS! |
| 03:47 | Apage43 | aw yeah! |
| 03:54 | s4muel | Viesti: (def ^:const planck 6.626...) |
| 03:55 | callen | (inc s4muel) |
| 03:55 | lazybot | ⇒ 1 |
| 03:55 | callen | just for using planck. |
| 03:56 | s4muel | seemed natural |
| 04:02 | allenj12 | Im off for tonight |
| 04:02 | allenj12 | Apage43: thank you |
| 04:09 | Viesti | s4muel: thanks |
| 05:54 | allenj12 | hey does anyone know how to get a key from a map value? say if i took the first of a map |
| 05:55 | hyPiRion | use `key` |
| 05:55 | hyPiRion | ,(key (first {:a :b})) |
| 05:55 | clojurebot | :a |
| 05:55 | allenj12 | ah k lol sry if that was silly |
| 05:55 | hyPiRion | ah, no worries |
| 07:19 | reprob | Hi! Is it possible to edn/read something containing a clojure regex? (Specifying a :default option didn't work.) |
| 07:21 | carlosgaldino | is it possible to destructure the args of a function that was passed as an argument to another function? |
| 07:22 | carlosgaldino | smth like: (defn my-function [function [a & args]]) |
| 07:25 | augustl | carlosgaldino: the only place those args would be available is inside the function, so I can't think of any use case for doing that |
| 07:26 | carlosgaldino | augustl: it's just an exercise. I was trying to implement `curry` but it's not working and I wanted to make sure that I was getting the arguments the right way |
| 07:26 | Bronsa | reprob: no, the edn reader doesn't include the regex literal |
| 07:26 | reprob | Bronsa: Ah ok, thanks. I'll use the Clojure reader with *read-eval* off then... |
| 07:26 | carlosgaldino | augustl: since I'm very new to clj I'm just like, really don't know what I'm doing. :) |
| 07:28 | augustl | carlosgaldino: yeah, the args to a function passed to another function is not available to that other function |
| 07:44 | carlosgaldino | augustl: ok, thanks |
| 08:09 | jave | I'm trying to make a spring mvc controller in Clojure, more or less like this: |
| 08:09 | jave | http://stevendick.github.io/blog/2013/08/13/write-a-spring-mvc-controller-in-clojure/ |
| 08:09 | jave | but I cant get it to work |
| 08:10 | jave | It doesnt seem like the mothod annotations get compiled into the class |
| 08:11 | jave | if I look at another java based controller I can find the requestmapping annotation argument inside the class. I cant find it in my clojure based controller class |
| 08:14 | schmir | is there a leiningen plugin helps with building a distribution for end users? i.e. helps with building an uberjar, and a .bat/shell wrapper? |
| 08:37 | doomlord_ | what are the chances of a native clojure implementation apearing |
| 08:41 | AimHere | https://github.com/takeoutweight/clojure-scheme <- seems to exist already |
| 08:41 | AimHere | At least in some form |
| 08:42 | AimHere | Clojure's parasiting on host systems is probably part of it's success though! |
| 08:47 | mdrogalis | callen: Heh, yeah. More just meant for repl use. |
| 09:26 | G0SUB | doomlord_, AimHere: There's also https://github.com/halgari/clojure-metal/ |
| 09:31 | doomlord_ | thats different , but still pretty interesting. |
| 09:32 | doomlord_ | ah, it can go to llvm |
| 09:37 | augustl | G0SUB, doomlord_: I wonder if that's legal.. Perhaps Rich will sue everyone, just like Oracle sued Android over Java :P |
| 09:37 | G0SUB | augustl: what's illegal here? |
| 09:38 | TimMc | The name would need changing. |
| 09:38 | augustl | G0SUB: the Clojure trademark, and the interface of Clojure itself |
| 09:38 | doomlord_ | it could only be good for the clojure ecosystem, right? |
| 09:39 | augustl | didn't Oracle sue Google for copying the "looks" (not sure what the correct term is..) of Java in Dalvik? |
| 09:39 | G0SUB | augustl: there's nothing illegal here, FWIW it's created by Tim Baldridge, Cognitect employee & co-author of core.async |
| 09:39 | G0SUB | augustl: he also wrote clojure-py |
| 09:40 | G0SUB | and that Oracle reference is a joke in poor taste, IMHO. |
| 09:40 | G0SUB | all of these use-cases fall under "Fair Use" |
| 09:40 | doomlord_ | edn looks neat, i'd always wanted json to support keywords/symbols |
| 09:40 | augustl | G0SUB: I see |
| 09:45 | G0SUB | augustl: Oracle was just plain stupid (and arrogant) to sue Google over API. That's why they lost. |
| 10:10 | silasdavis | If I am in the namespace some-long-and-unwieldy-namspace, and I use the fully qualified keyword ::foo |
| 10:10 | silasdavis | from another namespec can I coveniently reference :some-long-and-unwieldy-namspace/foo? |
| 10:11 | silasdavis | I was hoping I could use a require on some-long-and-unwieldy-namspace |
| 10:11 | justin_smith | silasdavis: (:require some-long-and-unweildy-namespace :as slaun) ... ::slaun/foo |
| 10:12 | silasdavis | ah, thanks |
| 10:12 | justin_smith | np |
| 10:13 | justin_smith | I think :: brings in the same resolution rules as for vars, no prefix means current namespace by default, and then a prefix gets looked up as an alias |
| 10:34 | mdrogalis | Night Vision debugger lives! https://github.com/MichaelDrogalis/night-vision |
| 10:43 | ToBeReplaced | i'm still not grokking core.async -- am I able to open/close readers inside of a go block? |
| 10:44 | mdrogalis | ToBeReplaced: You don't close readers, you close channels. |
| 10:45 | mdrogalis | ToBeReplaced: Try not to overthink it. It's just a queue that can be consumed by lightweight threads. |
| 10:45 | cmajor7 | simple macro question.. still have difficulties fully grasping them: https://www.refheap.com/18919 |
| 10:47 | ToBeReplaced | mdrogalis: if i have a file, and i want to write its lines onto a channel, how would i do that? i would have done (go (with-open [... ] ...)) |
| 10:48 | mdrogalis | ToBeReplaced: Oh, you meant readers in that sense. Sure, why not? |
| 10:49 | ToBeReplaced | mdrogalis: with-open reflects inside of go blocks, and i'm having issues trying to do it directly with try/finally -- it seems to close from the finally block early |
| 10:50 | vijaykiran | cmajor7: I'm not sure how you can return two values from a macro |
| 10:51 | justin_smith | the only way to return two things is to put them inside a structure |
| 10:51 | justin_smith | period, macro or function or special form |
| 10:52 | vijaykiran | cmajor7: in your case, the m is always returns the let which will return the last form |
| 10:52 | ToBeReplaced | mdrogalis: in fact (go (let [rdr ...] ...)) succeeds where (go (with-open [rdr ...] ...)) fails -- so i'm thinking that i just don't understand something |
| 10:52 | justin_smith | you can use ~@ to expand a structure inline inside a syntax quote though |
| 10:53 | cmajor7 | justin_smith: use "~@" on what structure.. I feel "the truth" is somewhere, just can pinpoint it.. can you give me a simple example? |
| 10:53 | justin_smith | ,`(1 2 ~@[3 4] 5) |
| 10:53 | clojurebot | (1 2 3 4 5) |
| 10:54 | justin_smith | of course [3 4] did not have to be literal, it could have been a binding |
| 10:55 | mdrogalis | ToBeReplaced: What do you mean it 'reflects'? |
| 10:55 | mdrogalis | AFAIK, with-open is just a macro that expands to close the reader at the bottom of try/finally |
| 10:55 | ToBeReplaced | reflection on .close |
| 10:55 | ToBeReplaced | the type hint seems to go awry inside of the go block |
| 10:55 | cmajor7 | justin_smith: trying… (defmacro m [n] `(~@[{:a 42} {:a 42 :b (+ 1 n)}])) but it seems to evaluate.. e.g. it treats {: a 42} as a funtion |
| 10:57 | mdrogalis | ToBeReplaced: Hm, weird. |
| 10:57 | mdrogalis | Maybe ask tbalridge whenever he turns up. |
| 10:57 | justin_smith | cmajor7: yes, that is what putting something by an open paren does |
| 10:58 | justin_smith | the ~@ has to be where the args are used, not generated |
| 10:58 | justin_smith | the macro problem: they make you need to use more macros |
| 10:58 | ToBeReplaced | mdrogalis: yeah i've been having weird reflection problems -- https://www.refheap.com/18905 |
| 10:59 | mdrogalis | Perhaps file it on Jira. |
| 10:59 | cmajor7 | justin_smith: not sure I understand :) how does it need to be changed to return two maps? |
| 10:59 | justin_smith | cmajor7: the ~@ flattens out a structure into a list, but it cannot flatten the return value into multiple values because there is literally no way to return multiple values that are not in a structure |
| 11:00 | ToBeReplaced | mdrogalis: no CA :( |
| 11:01 | cmajor7 | justin_smith: ok. the way I was thinking about it is to substitute two maps at the compile time, hence thought of macros, is there a different way to do it? and/or this cannot be done with macro? |
| 11:01 | justin_smith | cmajor7: in order to return multiple values you would need to port scheme's multiple return value feature or something, it would be a pain, and would not be that useful in my experience |
| 11:01 | justin_smith | cmajor7: the way to do it is a macro where you use the values, not where you return them |
| 11:02 | cmajor7 | justin_smith: I see, so macros can do anything _as long_ as that anything is a single form? |
| 11:02 | justin_smith | or just a function that destructures them and calls the other function |
| 11:02 | justin_smith | cmajor7: yeah |
| 11:03 | mdrogalis | ToBeReplaced: You don't need a CA to file a bug on Jira. |
| 11:03 | mdrogalis | You do need one to fix/contribute code. |
| 11:03 | cmajor7 | justin_smith: thx. "is a macro where you use the values, not where you return them", can you elaborate? |
| 11:03 | ToBeReplaced | ah, didn't realize |
| 11:03 | mdrogalis | :) |
| 11:05 | justin_smith | cmajor7: (defmacro m [double a b c] `(do-something ~@double a b c)) |
| 11:05 | justin_smith | where double ends up bein the two maps you generated elsewhere |
| 11:05 | justin_smith | forgot to unquote a b c but hopefully you get the gist |
| 11:06 | justin_smith | so the thing making the two maps can just be a function returning a vector or list of length 2 |
| 11:06 | justin_smith | the macro happens where you use them |
| 11:07 | justin_smith | but you really don't need a macro here, you could as well do (let [[a1 a2] double] (do-something a1 a2 b c)) |
| 11:07 | justin_smith | no macro needed, just destructuring |
| 11:20 | cmajor7 | justin_smith: right. the things is there are other maps coming from other sources, so these two maps is just an example.. I can always just return [m1 m2] and flatten them later, since they still come in as an ArraySeq.. I'd just need to get over a fact that a macro can be evaluated to a single form only |
| 11:26 | sm0ke | Hello, in core.async how do i define a callback function, when channel has value? |
| 11:28 | yedi | has there been a write up on Aurora yet? |
| 11:30 | mdrogalis | sm0ke: There are no callbacks. |
| 11:30 | fredyr | sm0ke: i think one of the ideas w/ core.async is to get away from callbacks |
| 11:30 | mdrogalis | Just write the code sequentially. |
| 11:30 | fredyr | sm0ke: and instead have blocking semantics |
| 11:30 | fredyr | also what mdrogalis said |
| 11:30 | sm0ke | fredyr: i dont get it why async in first place than? |
| 11:31 | scriptor | because async isn't just about callbacks |
| 11:31 | scriptor | it's so you can write asynchronous code that *looks* blocking, but is actually non-blocking |
| 11:31 | mdrogalis | sm0ke: The compiler inserts the callbacks for you. |
| 11:31 | sm0ke | what!!! |
| 11:32 | justin_smith | sm0ke: core.async turns a series of blocking reads into async callbacks on the value returned by the thing writing |
| 11:32 | sm0ke | really? wow but i get writes on channel..i still dont get the reads. |
| 11:33 | sm0ke | i mean i dont understand |
| 11:33 | scriptor | pastebin your code and we can take a look |
| 11:33 | scriptor | ah |
| 11:33 | sm0ke | :P |
| 11:33 | sm0ke | really compiler inserts the callbacks...that the most awesome thing i head all day |
| 11:34 | fredyr | i guess the best way to think of it is ordinary threads writing/reading on blocking channels |
| 11:34 | justin_smith | sm0ke: the "reads" are translated into a callback that does everything you use the value for in a callback. But you don't have to write a callback any more than you write gotos to implement a function. |
| 11:34 | mercwithamouth | slurping/barfing is freaking awesome =P |
| 11:36 | scriptor | sm0ke: Nolen's article helped me out http://swannodette.github.io/2013/07/12/communicating-sequential-processes/ |
| 11:36 | scriptor | the first example is pretty complex, but then he moves on to a simpler one |
| 11:37 | sm0ke | scriptor: ok so what you are saying is..whenever clojure compiler sees a (go (<!..the outside form is turned into a callback? |
| 11:38 | justin_smith | sm0ke: it goes deeper than that, every <! is a callback |
| 11:38 | justin_smith | or is translated into one, within the go block |
| 11:39 | justin_smith | so your typical go block is a deep set of nested callbacks, that read like normal step by step blocking code |
| 11:40 | sm0ke | wow i need some time and googling to digest all this |
| 11:40 | bbloom | justin_smith: should point out: their execution is nested, but the callbacks themselves are not: they are a state machine built out of a switch statement |
| 11:40 | justin_smith | bbloom: thanks |
| 11:40 | scriptor | has anyone written up anything on the internals of core.async? |
| 11:40 | bbloom | scriptor: yes, 1 sec |
| 11:40 | justin_smith | sm0ke: this is actually simpler than callbacks to use, that is the whole point |
| 11:40 | scriptor | the source for (go …) is a bit intimidating |
| 11:40 | bbloom | scriptor: http://hueypetersen.com/posts/2013/08/02/the-state-machines-of-core-async/ |
| 11:41 | justin_smith | it is callbacks and all that mess that require the heavy mental lifting from a user of the api |
| 11:41 | scriptor | bbloom: nice, thanks! |
| 11:41 | justin_smith | ccore.async makes it easier, much easier |
| 11:42 | tim__ | scriptor: also, my Clojure/Conj talk in 2 months |
| 11:43 | justin_smith | bbloom: isn't a state machine semantically equivalent to nested callbacks? |
| 11:43 | sm0ke | wow after learning about (compile.. this is the second biggest mindblowing thing i stumbled upon in clojure.. |
| 11:43 | sm0ke | although i guess this is borrowed from golang? |
| 11:43 | bbloom | justin_smith: the point i was clarifying is that the callbacks aren't nested lexical closures |
| 11:43 | tbaldridge | justin_smith: perhaps semantically, but not from a performance perspective. The state machine version is pretty much alloc free. |
| 11:44 | justin_smith | right, nice |
| 11:44 | tbaldridge | justin_smith: or can be made so, it's kindof half-alloc free in core.async right now. |
| 11:44 | justin_smith | I am actually very interested in ways to make alloc-free versions of clojure functions |
| 11:45 | scriptor | tbaldridge: looking forward to the slides/video then |
| 11:45 | justin_smith | in common lisp there was this concept of "cons free programming", how to write code for tight loops that does not churn garbage |
| 11:45 | justin_smith | are there instructional resources for doing the equivalent in clojure? |
| 11:46 | justin_smith | I mean I guess we could each suss out how to do it, but a set of tricks and best practices woudl be cool |
| 11:46 | bbloom | justin_smith: if you need instructions on how to do that, you shouldn't be doing that |
| 11:46 | tbaldridge | justin_smith: pretty much any doc on Clojure performance tuning will talk about type-hinting (that removes boxing), trying to lift closures to outer loops, etc. |
| 11:46 | justin_smith | bbloom: hey, I want to do DSP in clojure, fuck me right? |
| 11:47 | bbloom | justin_smith: you don't need to optimize that much unless you've measured & know what you're doing. at which point, it will be obvious how to experiment w/ transients, arrays, mutation, etc |
| 11:47 | pyr | anyone using clojure.core.typed in anger ? |
| 11:47 | justin_smith | bbloom: when I have the same loop called N times a second, with the same code path every time, hints on how to do idiomatic clojure things without churning the heap would be nice |
| 11:48 | bbloom | justin_smith: take a look at HipHipArray |
| 11:48 | justin_smith | cool, thanks |
| 11:48 | tbaldridge | justin_smith: and a profiler, and a hammock |
| 11:48 | bbloom | https://github.com/prismatic/hiphip |
| 11:48 | Sorella | sm0ke: kinda. core.async is based on Hoare's (et al?) communicating sequential process work. Go is also based on that. |
| 11:49 | tbaldridge | sm0ke: that being said, Rich has very few issues with Go's implementation with CSP, mostly around syntax. In his core.async podcast Rich basically said "Go did it right, we're copying them, we're not ashamed of it" |
| 11:49 | justin_smith | I have done dsp in other languages, and the things I tended to find useful was looking at other people's dsp code, and the rules of thumb for successful dsp (never print or allocate memory in your processing thread, etc.) |
| 11:50 | Apage43 | It seems like you'd figure those out pretty quick |
| 11:52 | justin_smith | so we've determined that the resource I am asking about does not exist because I am simply too stupid to figure it out on my own, and only people who can invent these things from scratch should be doing this sort of thing. OK. |
| 11:52 | tbaldridge | justin_smith: or perhaps bbloom just lacks tact. Every performance optimizing programmer has to start somewhere. |
| 11:53 | tbaldridge | justin_smith: take a look at hiphop though, it was kindof designed for this sort of thing |
| 11:56 | bbloom | justin_smith: tbaldridge: my apologies, didn't mean to be insulting. just was trying to make the point that it's the sort of thing that you should avoid doing in clojure unless you have to and then the solutions in clojure are roughly the same as they are across the JVM ecosystem and beyond: use arrays, mutation, etc |
| 11:56 | tbaldridge | (inc bbloom) |
| 11:56 | lazybot | ⇒ 18 |
| 11:57 | tbaldridge | justin_smith: don't forget, you're asking how to do dsp in a dynamic language. That's a tad like asking how to do NASCAR racing with a F150. |
| 11:57 | mtp | that's how it usually goes |
| 11:57 | justin_smith | tbaldridge: sure, yeah |
| 11:57 | tbaldridge | you can do it, it's going to take some work, creativity that's all. |
| 11:58 | justin_smith | right now I just do the DSP in a dsl (csound) and do the tricky logic in clojure |
| 11:58 | bbloom | justin_smith: have you looked at java-based solutions that you can access via interop? |
| 11:58 | tbaldridge | justin_smith: for instance, even writing your DSP code in JAVA and then wiring it up with Clojure |
| 11:59 | justin_smith | yeah, that would make sense |
| 11:59 | Apage43 | this is the whole overtone thing |
| 11:59 | Apage43 | it gives you DSLs over supercollider |
| 11:59 | seangrov` | (dec bbloom) ; gotta keep the ego in check |
| 11:59 | lazybot | ⇒ 17 |
| 11:59 | bbloom | dammit. |
| 11:59 | justin_smith | I don't like overtone because definstrument looks like clojure code but isn't, it ends up being a bigger cognative overhead than just using csound code is |
| 11:59 | arrdem | (dec seangrove) ; use of dec is discouraged |
| 11:59 | lazybot | ⇒ -2 |
| 12:00 | Bronsa | lol |
| 12:00 | arrdem | (dec arrdem) ; for fairness |
| 12:00 | lazybot | You can't adjust your own karma. |
| 12:00 | arrdem | #winning |
| 12:00 | bbloom | (doseq [user (:users clojure)] (dec user)) |
| 12:01 | arrdem | it'd be awesome if the bots did expose a user/nicks var |
| 12:02 | seangrov` | arrdem: Fair enough :) |
| 12:02 | tbaldridge | (set! tbaldridge (inc 9000)) |
| 12:02 | tbaldridge | that's what i'd do if it did. |
| 12:03 | seangrov` | &(println "alive?") |
| 12:03 | lazybot | ⇒ alive? nil |
| 12:03 | seangrov` | Ah, probably can't get carma from lazy bot then |
| 12:03 | seangrov` | &(do (println "(inc seangrov`)") 1) |
| 12:03 | lazybot | ⇒ (inc seangrov`) 1 |
| 12:05 | jonasen | bbloom: I vaguely remember you being interested in facebooks React.js. Do you know of any good resources (except the github repo) to learn how it works and how it's designed? It'd be an awesome match for clojurescript/hiccup I think and I'd be interested in building (a proof of concept) port. If it's not too complicated that is :) |
| 12:06 | bbloom | jonasen: they docs were updated a month or so ago to be much clearer |
| 12:06 | bbloom | i still believe very strongly that their approach is the best out there in the js world & would port very nicely to clojure |
| 12:07 | seangrov` | bbloom: Wanted to say thanks for the talk that unexpected surfaced here in #clojure on concatenative clojure, I really enjoyed it, and it opened my mind a bit towards stack-based languages |
| 12:07 | bbloom | seangrov`: cool! I'm glad you enjoyed it :-) |
| 12:07 | jonasen | bbloom: I've tried to find the reconciliation algorithm in their code base on github but haven't been able to find it |
| 12:07 | bbloom | jonasen: i think the word to look for is "transaction" |
| 12:07 | jonasen | bbloom: ok |
| 12:08 | bbloom | jonasen: they are pretty helpful over in #reactjs too |
| 12:10 | bbloom | jonasen: i was wondering if it would be possible to modify react's core to be reusable with different data structures. i think they already do dependency injection everywhere for testing, so i bet one could swap in hiccup-style data, but there are some other complications such as the component "classes" |
| 12:12 | jonasen | bbloom: Yes, it's probably possible to build a wrapper of sorts. |
| 12:12 | jonasen | Don't know how well it would work with advanced compilation |
| 12:13 | bbloom | jonasen: i was hoping not to have to wrap at all, but rather parameterize their engine |
| 12:22 | jonasen | bbloom: I think I found the key classes: ReactCompositeComponent and ReactNativeComponent. I will try to dig deeper and see if I'm able to make sense of the implementation |
| 12:23 | bbloom | jonasen: it took me a while to sort it all out. it's dense |
| 12:26 | wei_ | what's a good way to write a logical xor? e.g. (xor true true) => false |
| 12:27 | bbloom | wei_: (defn xor [p q] (and (or p q) (not (and p q)))) |
| 12:27 | wei_ | i guess that's just not= ... |
| 12:27 | bbloom | wei_: heh, yeah but i was just gonna say that :-) |
| 12:27 | wei_ | cool :) thanks bbloom |
| 12:38 | silasdavis | what's the best way to call a sequence of methods just for their side-effects? |
| 12:39 | tbaldridge | silasdavis: dorun |
| 12:39 | tbaldridge | ,(doc dorun) |
| 12:39 | clojurebot | "([coll] [n coll]); When lazy sequences are produced via functions that have side effects, any effects other than those needed to produce the first element in the seq do not occur until the seq is consumed. dorun can be used to force any effects. Walks through the successive nexts of the seq, does not retain the head and returns nil." |
| 12:39 | TimMc | wei_: not= works only if you're dealing with "physical" booleans. |
| 12:39 | tbaldridge | silasdavis: compare to: |
| 12:39 | tbaldridge | ,(doc doall) |
| 12:39 | clojurebot | "([coll] [n coll]); When lazy sequences are produced via functions that have side effects, any effects other than those needed to produce the first element in the seq do not occur until the seq is consumed. doall can be used to force any effects. Walks through the successive nexts of the seq, retains the head and returns it, thus causing the entire seq to reside in memory at one time." |
| 12:40 | wei_ | what are non-physical booleans? |
| 12:40 | TimMc | If you have logical booleans, you'd need to call (boolean ...) on the args first. |
| 12:40 | rasmusto | tbaldridge: oh, I think I found the memory issues in my code :O |
| 12:40 | wei_ | ah, right |
| 12:40 | TimMc | wei_: Well, ##(if 1 "logical true" "logical false") |
| 12:40 | lazybot | ⇒ "logical true" |
| 12:40 | TimMc | &(if nil "logical true" "logical false") |
| 12:40 | lazybot | ⇒ "logical false" |
| 12:41 | tbaldridge | rasmusto: ? |
| 12:44 | TimMc | So this should work: (defn xor2 [a b] (not= (boolean a) (boolean b))) |
| 12:44 | dnolen | pyr: it seems like people are starting to do that, check #typed-clojure |
| 13:00 | rasmusto | tbaldridge: dorun vs doall |
| 13:01 | rasmusto | I have a (doall (pmap ...)) over a very large sequence, and I'm just using it for side effects |
| 13:03 | silasdavis | tbaldridge, I would I use it to execute [fn1, fn2, ...] |
| 13:03 | silasdavis | like (do (fn1) (fn2) (fn3) ...) |
| 13:05 | rasmusto | ,(doall (map #(%) [#(print 1) #(print 2) #(print 3)])) |
| 13:05 | clojurebot | 123(nil nil nil) |
| 13:05 | justin_smith | is there a known workaround for the loss of println output into the ether when printing inside asynch processes in nrepl / emacs? |
| 13:06 | justin_smith | I guess I could start using a logging lib in my repl, but that somehow feels wrong |
| 13:07 | clgv | justin_smith: you probably have an stdout buffer in emacs. the missing messages should show up there. |
| 13:08 | justin_smith | I am connecting to an nrepl in a terminal |
| 13:08 | glosoli | What about Emacs |
| 13:08 | glosoli | ? |
| 13:08 | justin_smith | it is not showing up in the terminal, the nrepl buffer, or the nrepl connection buffer |
| 13:08 | glosoli | justin_smith: C-x-b nrepl-server |
| 13:08 | glosoli | It might be hidden, depending on the conf |
| 13:09 | justin_smith | there is no such buffer, because I am connecting to a process in a terminal |
| 13:09 | glosoli | But it shows up in buffer search :) |
| 13:09 | glosoli | hmm |
| 13:09 | glosoli | never done that |
| 13:10 | justin_smith | I should see "pending: (this is a test)" instead I get "pending: (this is", or sometimes "pending: " |
| 13:10 | justin_smith | or sometimes nothing |
| 13:10 | justin_smith | seemingly random |
| 13:10 | glosoli | justin_smith: ah yes you are right hmm interesting |
| 13:11 | justin_smith | so you can reproduce it by printing in future calls maybe? |
| 13:12 | mattyw | I've just suddenly started seeing SAXParseException when I try to evaluate using vim-fireplace. Has anyone else seen this? |
| 13:12 | glosoli | justin_smith: dunno man :) I am kinda of a newbie here, it's first time I heard of future calls :) |
| 13:14 | silasdavis | can I write (dorun (for [f [#(println "foo") #(println "bar")]] (f))) more directly |
| 13:15 | justin_smith | glosoli: (dotimes [i 30] (future (println "hello" "there" "world:" i))) |
| 13:15 | justin_smith | notice how few numbers are in that output |
| 13:16 | glosoli | http://imgur.com/AGRtR8H |
| 13:16 | glosoli | yeah |
| 13:16 | justin_smith | even better: (dotimes [i 30] (future (locking *out* (println "hello" "there" "world:" i)))) |
| 13:17 | justin_smith | in a repl that loses almost all the printouts, in a naked repl you see them all |
| 13:18 | justin_smith | hmm, looks like you are getting most of the output, if not all, I am losing most of it |
| 13:18 | justin_smith | try the locking version I just pasted |
| 13:19 | glosoli | http://imgur.com/YBBbUeF |
| 13:19 | glosoli | Seems like I got all of them |
| 13:19 | justin_smith | something must be wrong with my setup |
| 13:19 | glosoli | justin_smith: maybe older version of nREPL ? |
| 13:20 | justin_smith | mine is pretty new, let me see if elpa has an update |
| 13:20 | Raynes | Reddit has ruined me. |
| 13:20 | Raynes | I saw all those imgur links and was fully expecting to see cats. |
| 13:20 | justin_smith | heh |
| 13:20 | Raynes | Was utterly disappointed to find out otherwise. |
| 13:20 | glosoli | cat pics lol, da best |
| 13:21 | Raynes | glosoli: http://i.imgur.com/b24K57j.gif |
| 13:21 | glosoli | aaah so these are the cat pictures everyone are crazy about lol |
| 13:23 | silasdavis | is (fn [f] (f)) a built-in function? |
| 13:23 | xeqi | &(apply (fn [] 3)) |
| 13:23 | lazybot | clojure.lang.ArityException: Wrong number of args (1) passed to: core$apply |
| 13:24 | technomancy | maybe force? |
| 13:24 | xeqi | blah, the ones I can think of don't have 0 arity |
| 13:24 | technomancy | nope |
| 13:24 | technomancy | deliver is the 1-arity one |
| 13:25 | justin_smith | ,(deliver 3) |
| 13:25 | clojurebot | #<ArityException clojure.lang.ArityException: Wrong number of args (1) passed to: core$deliver> |
| 13:25 | technomancy | ,(deliver println :ohai) |
| 13:25 | clojurebot | :ohai\n |
| 13:25 | silasdavis | context is (dorun (map #(%) fns)) |
| 13:25 | xeqi | &(.invoke (fn [] 3)) |
| 13:25 | lazybot | ⇒ 3 |
| 13:25 | technomancy | use doseq |
| 13:26 | silasdavis | (doseq [f fns] (f))? |
| 13:26 | xeqi | yep, that is better in that context |
| 13:26 | silasdavis | ok cheers |
| 13:30 | TimMc | silasdavis: Try trampoline. :-P |
| 13:31 | bordatoue | could anyone tell me which version is domain defined for core.logic |
| 13:35 | justin_smith | glosoli: updating my nrepl.el did the trick, it seems |
| 13:36 | glosoli | justin_smith: must have been some bug |
| 13:36 | TimMc | &(trampoline (-> f #(f))) |
| 13:36 | lazybot | java.lang.StackOverflowError |
| 13:37 | silasdavis | TimMc, nice that works... provided the function doesn't return a function you don't want to run |
| 13:37 | TimMc | Oh hmm, not quite. |
| 13:37 | TimMc | silasdavis: :-) |
| 13:37 | TimMc | &(trampoline (fn f [] f)) |
| 13:37 | lazybot | Execution Timed Out! |
| 13:38 | silasdavis | boing |
| 13:40 | technomancy | nice |
| 13:42 | TimMc | &(recur) is the shortest infinite loop I can make, I think. |
| 13:42 | lazybot | Execution Timed Out! |
| 13:46 | TimMc | And ##((-> f #(f))) is my new personal best for stack overflow, beating out the previous record of (#(% %) #(% %)) |
| 13:46 | lazybot | java.lang.StackOverflowError |
| 13:46 | Raynes | &(range) |
| 13:46 | lazybot | java.lang.OutOfMemoryError: Java heap space |
| 13:46 | Raynes | That's technically an infinite loop. |
| 13:47 | Raynes | Unfortunately it also uses heap space :P |
| 13:47 | TimMc | Raynes: (dorun (range)) hasn't been an infinite loop since 1.3 came out. |
| 13:48 | TimMc | RIP auto-promoting +. |
| 13:48 | rasmusto | &(dorun (range)) |
| 13:48 | TimMc | (Well, it's +' now, but whatever.) |
| 13:48 | lazybot | Execution Timed Out! |
| 13:48 | TimMc | rasmusto: That will take too long, you'll want to start not too far before Long/MAX_VALUE. |
| 13:48 | silasdavis | TimMc, what does ## do? |
| 13:48 | TimMc | silasdavis: Inline bot trigger. |
| 13:49 | silasdavis | ha |
| 13:49 | silasdavis | ah |
| 13:49 | `cbp | Hmm would a random error inside a compojure defroutes make it silently fail and cause the routes to not actually be added? |
| 13:50 | rasmusto | &(dorun (range (dec Long/MAX_VALUE) Long/MAX_VALUE)) |
| 13:50 | lazybot | ⇒ nil |
| 13:52 | TimMc | &(dorun (range (dec Long/MAX_VALUE) Double/POSITIVE_INFINITY)) |
| 13:52 | lazybot | java.lang.ArithmeticException: integer overflow |
| 13:52 | shoerain | So uh... low hanging fruit question: I want to reload a project from inside `lein repl`. Can I do that? |
| 13:52 | justin_smith | shoerain: (require ... :reload-all) |
| 13:52 | TimMc | shoerain: clojure.tools.namespace apparently has a refresh fn as well. |
| 13:52 | rasmusto | shoerain: check out clojure.tools.namespace.repl/refresh[-all] also, it's pretty cool |
| 13:52 | justin_smith | that doesn't help with for example async stuff still running |
| 13:53 | shoerain | whoa, fast. |
| 13:58 | shoerain | TimMc, rasmusto thank you 2 for clojure.tools.namespace.repl/refresh |
| 13:59 | callen | mdrogalis: right. for exploring other peoples' code :) |
| 14:00 | rasmusto | shoerain: no problem. I just got it set up last week and it works well for the most part. I'm still having issues with some AOT stuff though... |
| 14:01 | shoerain | so is there tab-completion for functions in a namespace? I have (ns overtone-tutorial.core (:gen-class)), can I do (overtone-tutorial.core/<TAB-character>)? |
| 14:01 | justin_smith | would "aot" include macros and protocols? because those can be tricky too |
| 14:01 | mdrogalis | callen: :) |
| 14:01 | justin_smith | shoerain: in emacs by default it is alt-tab or esc-tab |
| 14:01 | rasmusto | justin_smith: not in my case, I don't think i'm using non-standard macro stuff |
| 14:01 | callen | shoerain: why are you gen-class'ing? |
| 14:02 | shoerain | since (overtone-tutorial.core/<TAB>) |
| 14:02 | shoerain | callen: I couldn't tell you, since I don't know myself. |
| 14:02 | justin_smith | shoerain: yeah, tab is just indent by default, alt-tab or esc-tab (both map to M-tab internally) |
| 14:02 | rasmusto | justin_smith: I never understood that, doesn't alt-tab usually do something different? :P |
| 14:03 | justin_smith | rasmusto: only if your window managers steal it |
| 14:03 | justin_smith | emacs used it before "window managers" were a thing |
| 14:03 | justin_smith | iirc |
| 14:03 | rasmusto | ah gotcha |
| 14:04 | justin_smith | gen-class is only really needed if java code needs to call your code |
| 14:04 | justin_smith | afaik |
| 14:04 | justin_smith | and it is one of those aot things that will make reloading messy |
| 14:04 | technomancy | it can make uberjars easier too, but yeah, not required |
| 14:04 | shoerain | shucks, but my .inputrc gets in the way of esc-tab (vim bindings here), and m-tab is taken by window manager. hm |
| 14:05 | rasmusto | shoerain: evil-mode steals my 'esc' :p |
| 14:05 | justin_smith | technomancy: is that because the user of the uberjar is a java program, or is there something else I am missing? |
| 14:06 | technomancy | justin_smith: `java -jar myjar-standalone.jar` vs `java -cp myjar-standalone.jar clojure.main -m mymain.ns` |
| 14:06 | justin_smith | ahh |
| 14:06 | justin_smith | yeah |
| 14:06 | technomancy | there is also lein-otf |
| 14:06 | shoerain | rasmusto, justin_smith: so I should look into customizing `lein repl` (presumably nrepl) for tab completion? |
| 14:07 | justin_smith | oh, is this not in emacs, just the repl? |
| 14:07 | justin_smith | I was confused |
| 14:07 | technomancy | and profile-scoped targets with limiting :aot to the uberjar profile |
| 14:07 | rasmusto | hrm, c.t.n.r/refresh-all fails for me if I :aot [my.ns]... Says it can't find a namespace |
| 14:07 | shoerain | justin_smith: yup just the repl |
| 14:08 | rasmusto | shoerain: are you an emacs user? or would you like to learn? |
| 14:08 | justin_smith | yeah, then tab should complete things if the namespaces are loaded |
| 14:08 | shoerain | i haven't hooked up editors and repls (so far in python and javascript) |
| 14:09 | shoerain | justin_smith: it's a generic completion, I guess I want (clojure.tools.namespace.repl/<tab>) to return all possible inputs |
| 14:09 | shoerain | er, s/inputs/functions in that namespace/ |
| 14:09 | justin_smith | I think you want editor integration for that kind of thing |
| 14:10 | technomancy | no, you should have completion in lein repl |
| 14:11 | shoerain | there's some form of static completion... if I call c.t.n.r/foo, then I get 'foo' completed when I call it again. Even if 'foo' doesn't exist |
| 14:11 | technomancy | .inputrc might be confusing it |
| 14:15 | shoerain | this is a record of what i'm doing: http://asciinema.org/a/5572 |
| 14:16 | rasmusto | shoerain: i read that as "asciienema" |
| 14:16 | mtp | asciilemma |
| 14:16 | shoerain | well |
| 14:16 | shoerain | it has some noise too (from supercollider and /usr/bin/java) |
| 14:17 | justin_smith | if you :use you don't have to type out the namespace, that is kind of the point of :use |
| 14:17 | rasmusto | shoerain: it looks like it's completing the whole "clojure.tools.namespace..." thing, it should complete the stuff from "clojure...." as well |
| 14:17 | shoerain | so :use is similar to (use 'namespace.core), but in the repl? |
| 14:17 | shoerain | if so, I was wondering why the syntax is differet |
| 14:18 | rasmusto | shoerain: :use is just inside of the (ns) macro |
| 14:18 | rasmusto | but he's saying that if you "use" something, you don't have to fully qualify it |
| 14:18 | rasmusto | or "require c.t.n.r :refer [refresh]" |
| 14:19 | shoerain | oh yeah sure. I just wanted to look at functions in that namespace though |
| 14:19 | rasmusto | shoerain: yeah, I'm not sure what the completion is doing |
| 14:19 | shoerain | if I just <tab> with nothing typed in, I get every possible function/keyword |
| 14:20 | shoerain | rasmusto: so completion should be working, but it's not working for me? |
| 14:20 | rasmusto | shoerain: if I do "clojure<tab>" I get all the completions from core and others |
| 14:20 | technomancy | shoerain: try moving .inputrc and re-running lein repl |
| 14:20 | justin_smith | technomancy: his term session he posted showed he had no .inputrc |
| 14:20 | shoerain | should i output lein/clojure versions too? technomancy I did; there's a `cat ~/.inputrc` command that doesn't return anything |
| 14:21 | technomancy | oh oops |
| 14:21 | shoerain | anyway i'll check when i'm back later, thanks for the help |
| 14:21 | shoerain | mebbe i'll just install arch and be done with this ubuntu-ness... overtone was more of a pain to get started here |
| 14:22 | supersym | fun, I knew there had to be someone who tried that before :P (code by voice) https://www.youtube.com/watch?v=8SkdfdXWYaI |
| 14:22 | rasmusto | shoerain: are you running leiningen from apt? |
| 14:27 | wei_ | is there a way to get :body-params from compojure's destructuring syntx? |
| 14:29 | rasmusto | is there a way to auto-require ~/.lein/profiles.clj dependencies in the user namespace for projects? I've looked at making a :dev profile per-project with a user.clj, but I'd like something that works for multiple projects and doesn't clutter project dependencies |
| 14:30 | xeqi | wei_: (GET "/route" {:keys [body-params] :as request} ...) |
| 14:30 | technomancy | rasmusto: maybe :init in :repl-options? |
| 14:31 | technomancy | rasmusto: IMO that approach isn't great since it encourages you to stay in the user ns |
| 14:31 | technomancy | much better to expose the functionality you want in your editor |
| 14:31 | wei_ | thanks xeqi. so no way to get it with the compojure-specific destructuring e.g. (GET "/route/:id" [id] ...) ? |
| 14:31 | technomancy | https://github.com/technomancy/nrepl-discover |
| 14:31 | silasdavis | is there a built-in for (comp (partial into {}) concat) ? |
| 14:32 | rasmusto | technomancy: okay. ~/.lein/profiles.clj + editor seems like a good approach |
| 14:32 | silasdavis | hash-maps as args |
| 14:32 | technomancy | silasdavis: just into? |
| 14:32 | rasmusto | I guess the only issue is sharing tools w/ other ppl |
| 14:32 | silasdavis | technomancy, madness |
| 14:32 | technomancy | or merge if you need many args |
| 14:32 | silasdavis | thanks |
| 14:32 | technomancy | =) |
| 14:33 | technomancy | rasmusto: I've been thinking about exposing ops from reply and nrepl.el as things like "~toggle-trace my.ns/some-var" |
| 14:33 | technomancy | things that you'd normally invoke via M-x nrepl-toggle-trace |
| 14:33 | technomancy | might sketch out an implementation in grench first |
| 14:34 | rasmusto | hm, sounds interesting |
| 14:34 | callen | technomancy: the Grenchmen are on the forefront of (social) engineering? |
| 14:34 | technomancy | slime is similar; you can enter ,commands which aren't interpreted as code input |
| 14:34 | rasmusto | do you have a recommended way to run clojure code in the nrepl.el repl with an emacs hotkey? |
| 14:35 | technomancy | callen: I wish someone would record a cover of some grenchmen songs like http://wondermark.com/more-littlest-butt/ |
| 14:35 | technomancy | rasmusto: no, but that should be added to the docs |
| 14:36 | callen | technomancy: o lawd. |
| 14:36 | technomancy | you're welcome |
| 14:37 | justin_smith | rasmusto: C-M-x ? |
| 14:37 | justin_smith | that will run the surrounding sexp in the default nrepl instance |
| 14:37 | technomancy | I think he means code that isn't under the point |
| 14:37 | callen | I, too, want the littlest butt ;_; |
| 14:37 | rasmusto | justin_smith: yeah, this would be code that I define in my init.el |
| 14:38 | justin_smith | oh |
| 14:38 | rasmusto | stuff like (require '[clojure.pprint :refer [pprint]]) (assuming I'm in the nREPL mode) |
| 14:39 | callen | I use a macro for that. |
| 14:39 | callen | my refresh macro in the REPL dumps (require '[clojure.tools.namespace.repl :refer [refresh]]) (refresh) (require '[clojure.pprint :refer [pprint]]) |
| 14:40 | callen | I jam that and go back to what I was doing |
| 14:40 | technomancy | I don't think you should do it inside a session |
| 14:40 | justin_smith | callen: so then you then run the macro manually, right? |
| 14:40 | callen | I do so because I don't really live in 'user |
| 14:40 | technomancy | I think "make this operation available everywhere" should be supported by tooling |
| 14:40 | callen | justin_smith: I make a conscious decision to refresh my REPL, yes. |
| 14:40 | technomancy | reply does this to a degree |
| 14:40 | callen | justin_smith: it's partly a hack around not having a universal injection faculty. |
| 14:40 | callen | partly because I co-opted some of how Stuart Sierra works. |
| 14:41 | callen | my test REPL does: (require '[clojure.tools.namespace.repl :refer [refresh]]) (refresh) (run-all-tests) |
| 14:41 | callen | I jam "up-arrow return-key" to run my tests, ^^ |
| 14:41 | rasmusto | callen: I was looking at his dev/user.clj, and it works, but requires that you move deps from ~/.lein/profiles to the project |
| 14:42 | callen | rasmusto: nothx. I work on too many projects for that. |
| 14:42 | technomancy | user.cjl doesn't compose |
| 14:42 | callen | that too. |
| 14:42 | rasmusto | compose in what way? |
| 14:43 | technomancy | you can only have one of them |
| 14:43 | technomancy | it's basically the anti-namespace |
| 14:43 | technomancy | the namespace which isn't namespaced |
| 14:43 | rasmusto | may as well be called 'global? |
| 14:43 | technomancy | basically |
| 14:43 | rasmusto | okay, hmm |
| 14:44 | rasmusto | good thing user.clj is still just sitting in the index |
| 14:44 | rasmusto | callen: do you have a dotfiles repo I can look at for your emacs macro? |
| 14:45 | callen | I thought this channel was sick of me talking about my dotfiles repo. |
| 14:45 | callen | rasmusto: https://github.com/bitemyapp/dotfiles/blob/master/.emacs.d/clojure/clojure-config.el |
| 14:45 | callen | you'll see my tracing machinery there too. |
| 14:46 | rasmusto | I'll talk dotfiles all day |
| 14:47 | callen | rasmusto: my dotfiles repo has ~420,000 LOC of elisp. |
| 14:47 | technomancy | you monster |
| 14:47 | callen | rasmusto: you really don't want me to talk about my antediluvian Emacs. |
| 14:47 | rasmusto | i made this thing: https://github.com/rasmusto/vim-emacs-clojure-starter-kit and its bad |
| 14:47 | technomancy | anyone remember M-x clojure-install? |
| 14:47 | callen | entire monasteries of monks have gotten lost in my dotfiles. |
| 14:48 | callen | You can find stories of their past generations scribbled on the stone of the labyrinth |
| 14:48 | callen | the bones of friar john... |
| 14:48 | technomancy | callen: have you read http://www.amazon.com/Crazy-Hair-Neil-Gaiman/dp/0060579080 ? |
| 14:48 | technomancy | "Crazy Dotfiles" https://www.youtube.com/watch?v=7mx_l1bBrPk |
| 14:49 | callen | technomancy: I have not read it. |
| 14:50 | technomancy | "Hunters send in expeditions; radio back there positions. Still, we've lost a dozen there. Lost inside my crazy hair." |
| 14:50 | callen | rasmusto: anyway yeah, alembic, c.t.t, c.t.n.r, and c.pp are the relevant parts of that file. |
| 14:50 | technomancy | *their |
| 14:50 | technomancy | this is how I imagine your dotfiles |
| 14:50 | callen | Seems accurate. |
| 14:51 | callen | aka, "Where we're going? We won't need package management where we're going." |
| 14:51 | callen | cue demonic shrieks |
| 14:51 | callen | this is what happens when your development environment started life as timestamped tarballs a decade and change prior. |
| 14:52 | xeqi | huey lewis? |
| 14:52 | rasmusto | rasmusto: thanks :) |
| 14:57 | shaungilchrist | is anyone tackling the lack of pprint for cljs? |
| 14:58 | callen | shaungilchrist: no, but I'm pretty sure there aren't attack dogs that'll kill the first person that tries, so you should give it a whack. |
| 14:59 | callen | if there are attack dogs, plz report as much from your hospital bed. |
| 15:00 | callen | technomancy: thanks for reminding me of my least favorite part of CL. I'd almost suppressed that memory. |
| 15:00 | rasmusto | callen: where do you load up clojure-config.el? (pardon my emacs ignorance) |
| 15:00 | callen | now i have to spend 2 hours in the bathtub scraping my skin |
| 15:00 | callen | rasmusto: oh so you want to follow this rabbit-hole? |
| 15:00 | callen | rasmusto: https://github.com/bitemyapp/dotfiles/ that's the top-level, look at .emacs |
| 15:01 | callen | rasmusto: I add everything in ~/.emacs.d/ to the elisp load path |
| 15:01 | callen | then I require/load-library libraries and config files. |
| 15:01 | callen | at the bottom of my .emacs you will notice my extreme hatred of auto-fill-mode. |
| 15:03 | gleag | callen: Shouldn't you be more afraid of the velociraptors? ;) |
| 15:03 | rasmusto | callen: ah, ok. seems familiar |
| 15:03 | callen | gleag: if a 11 year old kid that "knows Unix" can gymnast-kick a velociraptor in the face, then I trust Katya to protect me. |
| 15:04 | callen | incidentally, I was once an 11 year old kid that "knows Unix" |
| 15:04 | callen | so I'm in good company. |
| 15:04 | callen | rasmusto: the familiarity is a ruse. |
| 15:04 | justin_smith | what was that 3d sgi file system manager called again? |
| 15:04 | callen | rasmusto: I even use a custom color theme :) |
| 15:04 | justin_smith | the one on the screen when she said that |
| 15:05 | callen | justin_smith: http://en.wikipedia.org/wiki/Fsn |
| 15:05 | callen | justin_smith: I googled what you said and clicked the first result. |
| 15:05 | rasmusto | callen: it would feel more familiar if it had escape key timeout hacks in it |
| 15:05 | callen | this is how it works 99% of the time when somebody asks a question in here. |
| 15:08 | TimMc | In this case, who cares? |
| 15:08 | TimMc | It's not a Clojure question, it's off-topic conversation. |
| 15:09 | TimMc | If you're getting burn-out from the number of times someone asks a Jurassic Park-related question, you should seriously consider the amount of time you spend on the internet. :-P |
| 15:09 | technomancy | or teach it to clojurebot |
| 15:09 | TimMc | Or that. |
| 15:09 | callen | ~raptors |
| 15:09 | clojurebot | No entiendo |
| 15:09 | technomancy | what do we have bots around here for anyway if not as labour-saving devices |
| 15:09 | technomancy | clojurebot: earn those botsnacks, slacker |
| 15:09 | clojurebot | Cool story bro. |
| 15:10 | technomancy | dang it |
| 15:10 | gfredericks | "labor" saving |
| 15:10 | technomancy | gfredericks: haven't you seen the jetsons |
| 15:10 | gfredericks | not really |
| 15:11 | technomancy | oh |
| 15:13 | gleag | callen: I actually had http://xkcd.com/292 in mind. |
| 15:14 | callen | I just wrote a "throws?" macro. I think I'm a bad person. |
| 15:14 | callen | gleag: so did I. |
| 15:15 | callen | gleag: he has a whole on-going theme with the velociraptors :) |
| 15:16 | sveri | hi, i am currently trying luminus with http-kit, now, in the docs it is said that i should start the server with lein run -dev, this works basically, but when i change my templates they dont get reloaded, is there a way to achieve that? |
| 15:16 | gleag | callen: Just like Tolkien, Jackson, and Rowling with spiders. :) |
| 15:17 | gleag | I wonder if there are people with lispalienophobia out there. |
| 15:21 | `cbp | sveri: selmer has a function selmer.parser/cache-off! |
| 15:22 | callen | sveri: it's not detecting the file timestamp changes for some reason, which is strange but easily rectified by what `cbp said. |
| 15:24 | sveri | callen: `cbp: ok, thats nice, but where and how do i call that function? |
| 15:25 | _scape | given (["v" "0.032977" "0.716507" "0.166762"] ["v" "0.033133" "0.716507" "0.165748"] ["n" "0.033133" "0.781325" "0.165748"] ["n" "0.017057" "0.716507" "0.145737"]) how would I filter this to be (["v"] [[ "0.032977" "0.716507" "0.166762"] ["0.033133" "0.716507" "0.165748"]]) etc. Basically I want to pull that first element and list the rest, grouping them together |
| 15:25 | sveri | in the handler.clj in the init function? |
| 15:25 | callen | sveri: sure. |
| 15:25 | callen | sveri: give it a whirl. |
| 15:29 | sveri | callen: well, it works, somehow, i changed the navigation entry text and it works when i open the "/" url, but opening other urls it is not actualized |
| 15:32 | rlb | It looks like java still can't do full-resolution filesystem timestamps. If so, then I suspect that leaves me with jni/jna, and if I decided to go the jni route, should the jni bits just be a separate project, or integrate with leiningen somehow? |
| 15:34 | llasram | rlb: I haven't used it yet myself, but might be worth checking out https://github.com/jnr/jnr-posix |
| 15:34 | nDuff | Hrm. |
| 15:34 | `cbp | _scape: group-by does kind of what you want |
| 15:34 | rlb | llasram: OK, thanks. |
| 15:35 | nDuff | I have some quite old cljs code which includes: (def error-queue (atom #queue [])) -- building this against a modern compiler, I'm getting "No reader function for tag queue". |
| 15:35 | shaungilchrist | callen: I think instead of trying to make pprint work in its entirety I will scratch my immediate itch which is pretty printing edn |
| 15:35 | `cbp | _scape: you would have to write a group-by that removes the first item when grouping |
| 15:35 | `cbp | ,(group-by first '(["v" "0.032977" "0.716507" "0.166762"] ["v" "0.033133" "0.716507" "0.165748"] ["n" "0.033133" "0.781325" "0.165748"] ["n" "0.017057" "0.716507" "0.145737"])) |
| 15:35 | clojurebot | {"v" [["v" "0.032977" "0.716507" "0.166762"] ["v" "0.033133" "0.716507" "0.165748"]], "n" [["n" "0.033133" "0.781325" "0.165748"] ["n" "0.017057" "0.716507" "0.145737"]]} |
| 15:36 | _scape | thank you |
| 15:36 | callen | shaungilchrist: you can already pretty-print edn... |
| 15:36 | callen | shaungilchrist: ...because it's Clojure data... |
| 15:37 | callen | I am now going to stare at shaungilchrist in consternation |
| 15:39 | dnolen | nDuff: worked for me at the REPL, did you make sure to clean your project after switching CLJS versions? |
| 15:42 | shaungilchrist | callen: in clojurescript? |
| 15:42 | nDuff | dnolen: "lein clean", definitely, and lein-clrjsbuild says that it's cleaning up after itself. I notice that clojurescript.net doesn't recognize #queue [] as valid either. |
| 15:42 | mdrogalis | dnolen: Coworker excitedly ran over to me. "Immutable JS objects". One word reply: "Mori" D: |
| 15:42 | nDuff | err, lein-cljsbuild |
| 15:42 | nDuff | dnolen: ...which build are you testing against? |
| 15:42 | shaungilchrist | callen: I am looking for run-time pprint of edn in cljs e.g. nice whitespace idiomatic indentation etc. |
| 15:42 | dnolen | nDuff: master |
| 15:43 | dnolen | nDuff: clojurescript.net doesn't track ClojureScript closely as far as I know, it's the CLJS-in-CLJS project |
| 15:43 | dnolen | mdrogalis: heh |
| 15:45 | nDuff | dnolen: hmm. I'm currently trying 0.0-1889, but if master works, maybe switching to a snapshot is the thing to do. |
| 15:46 | dnolen | nDuff: should work in 1889 too there's reader support for #queue and there has been for a long time |
| 15:46 | dnolen | nDuff: minimal project to recreate please |
| 15:53 | dnolen | bbloom: http://www.cs.indiana.edu/~sabry/papers/exteff.pdf |
| 15:54 | dnolen | tbaldridge: you might find this interesting too ^ |
| 15:54 | bbloom | @dnolen: yup. tbaldridge & i were discussing that a week or two ago in here |
| 15:54 | bbloom | good stuff |
| 15:54 | tbaldridge | dnolen: ditto what bbloom said |
| 15:55 | nDuff | dnolen: https://gist.github.com/charles-dyfis-net/087d8be56804c9e9fae2 |
| 15:55 | dnolen | bbloom: heh cool, I saw that Oleg & co were presenting it at Haskell Symposium |
| 15:55 | bbloom | awesome |
| 15:55 | bbloom | it's good stuff |
| 15:55 | nDuff | dnolen: ...there's a makefile in there to create the directory structure and run lein (since gist doesn't allow subdirectories). |
| 15:55 | bbloom | i think this extensible effects stuff is a BFD |
| 15:56 | dnolen | nDuff: does the problem still occur if you use a ns form? |
| 15:57 | nDuff | Shoot. I hadn't paid enough attention that it was a NPE here. |
| 15:57 | nDuff | Not the same error as the original non-minimal case. |
| 15:57 | nDuff | Back to try to figure out how to reproduce... |
| 16:11 | hfaafb | is it feasible to profile a short lived clojure jar in visualVM? |
| 16:11 | hiredman | hfaafb: best to put the short lived code in a loop |
| 16:11 | hiredman | if your short lived code exits before you have time to attach the profiler you cannot profile it |
| 16:12 | hfaafb | thanks |
| 16:13 | justin_smith | you could also load up a repl, attach jvisualvm, and then run the main function manually |
| 16:13 | justin_smith | but a loop will gather better data, likely |
| 16:13 | nDuff | ...if it's short-lived enough that the JVM has never had time to optimize it, it's going to be pretty slow in production. |
| 16:14 | nDuff | I wonder if optimizing the spun-up case is actually fair in this kind of situation. :) |
| 16:16 | noprompt | dnolen: is there any gotcha with dynamic variables in clojurescript? |
| 16:16 | noprompt | *are there |
| 16:16 | dnolen | noprompt: like what? |
| 16:17 | noprompt | dnolen: i'm not sure. someone has been working on making my CSS library compatible with clojurescript and the code that uses dynamic variables isn't working. :-/ |
| 16:18 | noprompt | dnolen: there's probably something subtle we're missing, but i thought i'd ask. |
| 16:18 | dnolen | noprompt: we don't really support bound-fn or similar things |
| 16:20 | noprompt | dnolen: ah ok, so binding is the problem. |
| 16:20 | noprompt | dnolen: wait, nm, my mistake. |
| 16:20 | noprompt | sorry i read that wrong. |
| 16:21 | dnolen | noprompt: binding if you're assuming that you can *capture* a set of bindings like you can with bound-fn in Clojure |
| 16:22 | jmonetta | hi guys, any ideas why this doesn't work? : |
| 16:22 | jmonetta | (defmacro testmacro |
| 16:22 | jmonetta | ([a] (testmacro 5 a)) |
| 16:22 | jmonetta | ([b a] `(+ ~a ~b))) |
| 16:24 | noprompt | dnolen: do you have a moment to glance at some code? |
| 16:25 | callen | jmonetta: I liked that you tried that. |
| 16:25 | dnolen | noprompt: only if it's a minimal case |
| 16:25 | noprompt | dnolen: yeah it's just the setup/application. |
| 16:26 | dnolen | noprompt: got a gist or link? |
| 16:26 | noprompt | dnolen: https://github.com/jeluard/garden/blob/cljs/src/cljx/garden/compiler.cljx#L48-L52 , #L62-L66, #L198 |
| 16:27 | noprompt | dnolen: just the var decl, macro, and how it's used. |
| 16:27 | jmonetta | callen: sorry, didn't got that |
| 16:27 | jmonetta | what am I doing wrong there? |
| 16:27 | noprompt | dnolen: the code works perfectly fine in clj, but in cljs it looks like *selector-context* is not getting bound. |
| 16:28 | dnolen | noprompt: yeah that's not what I'd call a minimal case :) |
| 16:28 | dnolen | noprompt: a stand alone project with a some code that doesn't work |
| 16:28 | noprompt | dnolen: oh :P |
| 16:30 | noprompt | dnolen: just to be clear though, binding [*some-var* some-val] should work fine in cljs though? |
| 16:30 | noprompt | i haven't messed with dyn vars at all in cljs yet. |
| 16:35 | dnolen | noprompt: yes binding is supported |
| 16:36 | noprompt | dnolen: ok that's what i thought. guess i have some investigation to do. :) |
| 16:37 | rlb | llasram: amusingly, it looks like jnr-posix does save the fractional seconds for the file timestamps, but provides no way to acces them (i.e. only private/final) -- mtime() only returns the at_mtime value, etc. |
| 16:38 | rlb | so perhaps back to thinking about jni... |
| 16:39 | rlb | (or seeing if jnr-posix might be interested in an extra method...) |
| 16:40 | technomancy | huh; ocaml can have let-scoped "open" declarations (essentially :refers) |
| 16:42 | technomancy | (with-ns [(:require [clojure.java.io :as io])] (io/copy thingy target)) |
| 16:42 | technomancy | that would be a fun exercise to implement |
| 16:44 | rasmusto | technomancy: wouldn't that be easier with a more composable ns macro? |
| 16:46 | Bronsa | technomancy: (defmacro with-ns [opts & body] `(let [ns# *ns*] (ns ~(gensym) ~@opts) ~@body (in-ns (ns-name ns#)))) |
| 16:46 | technomancy | Bronsa: mwahaha |
| 16:46 | rasmusto | oshi |
| 16:46 | technomancy | rasmusto: I don't think so unless you mean moving refers and :as aliases to not be implemented in java |
| 16:47 | technomancy | the way it happens in alias/refer is inherently stateful |
| 16:47 | rasmusto | maybe, I'm not quite sure what I mean. But I think people have talked about wanting inheritable namespace templates and the like |
| 16:48 | technomancy | huh; .removeAlias exists as a method on Namespace |
| 16:48 | technomancy | so you could just make the changes and roll them back |
| 16:50 | callen | My new nickname for http://clojurescript.net/ is the "Un-muggle-ma-tron" |
| 16:53 | mpenet | How bad is it having (def ? ...) in ones library, it syntatically nice on a sql like dsl but I am sure this violates a few laws |
| 16:53 | justin_smith | what law would it violate? |
| 16:53 | mpenet | the readers untold laws |
| 16:54 | technomancy | there are some completely-ignored guidelines about what constitutes a "legal" identifier |
| 16:54 | technomancy | |
| 16:54 | mpenet | I think vars should start with an alpha char, but then again nothing is specified |
| 16:55 | mpenet | well it is: http://clojure.org/reader |
| 16:55 | mpenet | Maybe I will comply then, or just go the heretic way, not sure yet |
| 16:55 | rasmusto | (defn myfn® []) |
| 16:56 | mpenet | actually "?" should be fine |
| 16:57 | technomancy | I wouldn't take those rules too seriously |
| 16:57 | technomancy | clojure breaks them |
| 16:58 | mpenet | Yup, and we have a macro called . another called .. . I guess "?" is fine, as long as you dont start with a num this follows the "rules" I think. |
| 17:00 | Bronsa | mpenet: there's no way a macro called . is going to work |
| 17:00 | mpenet | I know, this was an example of macro naming used in core, not a name I planned to use |
| 17:00 | Bronsa | oh, ok |
| 17:00 | Bronsa | (. is not a macro btw) |
| 17:00 | technomancy | isn't . a special form though? |
| 17:00 | mpenet | ah |
| 17:00 | technomancy | right |
| 17:01 | Bronsa | that's why I said it's never going to work |
| 17:10 | lunk | `source |
| 17:10 | lunk | is the code for clojurebot available somewhere? |
| 17:10 | technomancy | clojurebot: source |
| 17:10 | clojurebot | source is http://github.com/hiredman/clojurebot/tree/master |
| 17:10 | lunk | ty |
| 17:11 | technomancy | sure |
| 17:15 | lunk | `(let [x 5] x) |
| 17:15 | lunk | clojurebot: `(let [x 5] x) |
| 17:15 | clojurebot | Gabh mo leithscéal? |
| 17:16 | lunk | clojurebot: (let [x 5] x) |
| 17:16 | clojurebot | It's greek to me. |
| 17:16 | lunk | technomancy: little help here |
| 17:16 | lunk | :) |
| 17:16 | Bronsa | ,(let [x 5] x) |
| 17:16 | clojurebot | 5 |
| 17:16 | lunk | oh comma |
| 17:16 | lunk | ty |
| 17:16 | justin_smith | without the presence of namespaces, or if people were expected to 'use' your namespace, ? would be a bad name. With namespaces it is just fine, the namespace name makes it make sense |
| 17:17 | lunk | ,(for [x (range 1)] (* x x x x)) |
| 17:17 | clojurebot | (0) |
| 17:18 | noprompt | justin_smith: what would ? mean as far as a namespace name goes? |
| 17:18 | justin_smith | ? as a namespace? dunno |
| 17:18 | scriptor | lunk: if you need a quick repl online, there's always http://tryclj.com/ |
| 17:18 | justin_smith | I meant within one it at least has a context |
| 17:19 | lunk | did someone filter our System calls from clojure bot? (java.lang.System/exit 0) for instance |
| 17:19 | justin_smith | you can also /msg clojurebot |
| 17:19 | noprompt | justin_smith: can you share an example? |
| 17:19 | lunk | scriptor: just thinking about the bot and sanitizing user inputs ;) |
| 17:19 | jkj_ | http://pastie.org/8349900 |
| 17:19 | jkj_ | wtf |
| 17:19 | scriptor | ah |
| 17:19 | jkj_ | i have a very strange performace problem |
| 17:19 | jkj_ | very very strange |
| 17:20 | justin_smith | noprompt (speech-act/? "dog" "here") |
| 17:20 | justin_smith | generates a question |
| 17:20 | lunk | oh goody, security exception, worked better than expected |
| 17:20 | jkj_ | my sort-by took too long.. then i presorted the stuff, but now the (first) to the sequence takes time :D |
| 17:20 | noprompt | justin_smith: oh you were say a var named ? not a namespace named ?. |
| 17:20 | noprompt | *saying |
| 17:20 | justin_smith | yeah |
| 17:20 | justin_smith | hah |
| 17:21 | justin_smith | though ? would make a good name for a namespace that defined a large number of boolean predicates |
| 17:21 | gfredericks | jkj_: I think this is lazy seqs |
| 17:21 | jkj_ | gfredericks: true... filter probably does that |
| 17:21 | gfredericks | jkj_: without the first the lazy seq escapes the (time) call without doing any work |
| 17:21 | noprompt | justin_smith: personally, i don't like vars that are just called ? because it's really unclear what that function does until you look at the source/docs. |
| 17:21 | justin_smith | I think it would depend on the ns, but sure |
| 17:22 | jkj_ | gfredericks: funny that the repl shows so tiny times though the full sequence is shown |
| 17:22 | jkj_ | gfredericks: maybe the time is just not calculated and the sequence is realized on repl's own time? |
| 17:22 | jkj_ | ...or something |
| 17:23 | gfredericks | jkj_: yes that's it |
| 17:23 | noprompt | justin_smith: typically i err on the side of clarity with regard to var names. the fact you're asking if it's a good name suggests to me that it's probably not since there's some doubt there. |
| 17:23 | gfredericks | if you wrapped the filter version with (time (doall ...)) you'd see it slow down |
| 17:24 | noprompt | justin_smith: that being said, it may make perfect sense with regard to your lib. |
| 17:24 | jkj_ | gfredericks: yes. i tried |
| 17:25 | jkj_ | gfredericks: any ideas on finding the longest .startsWith string match fast? |
| 17:26 | justin_smith | ,(apply str (map (comp char read-string (partial apply (partial str "2r"))) (partition 8 "011011000110111101101100"))) |
| 17:26 | clojurebot | "lol" |
| 17:26 | jkj_ | gfredericks: http://pastie.org/8349932 |
| 17:27 | rasmusto | ,(str "2r" "0100101") |
| 17:27 | clojurebot | "2r0100101" |
| 17:27 | rasmusto | oh |
| 17:27 | justin_smith | read syntax for base 2 number |
| 17:28 | justin_smith | ,"2r0100101 |
| 17:28 | clojurebot | #<RuntimeException java.lang.RuntimeException: EOF while reading string> |
| 17:28 | gfredericks | jkj_: if your question is just about performance, then you can probably add a type hint there |
| 17:28 | justin_smith | oops |
| 17:28 | justin_smith | ,"2r0100101 |
| 17:28 | clojurebot | #<RuntimeException java.lang.RuntimeException: EOF while reading string> |
| 17:28 | justin_smith | ,"2r0100101" |
| 17:28 | clojurebot | "2r0100101" |
| 17:28 | rasmusto | I always get a little sad when bits are represented as unicode characters |
| 17:28 | justin_smith | ,2r0100101 |
| 17:28 | clojurebot | 37 |
| 17:28 | gfredericks | jkj_: (.startsWith ^String call (:call %)) |
| 17:28 | jkj_ | gfredericks: thx |
| 17:29 | gfredericks | jkj_: that avoids reflection, which is super slow compared to a quick string operation |
| 17:29 | jkj_ | gfredericks: wow. a huge difference |
| 17:30 | justin_smith | rasmusto: I had thrown that little snippet together to decode a string of bits I saw on the net |
| 17:30 | rasmusto | justin_smith: yeah, there are some 4clojure problems that involve looking at digits, seems like read-string is the cleanest way to do it |
| 17:31 | jkj_ | gfredericks: problem solved. |
| 17:31 | gfredericks | ,strangely I can't reproduce the timing difference: |
| 17:31 | clojurebot | #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: strangely in this context, compiling:(NO_SOURCE_PATH:0:0)> |
| 17:31 | gfredericks | ,(time ((fn [s] (.startsWith s "foo")) "foo bar")) |
| 17:31 | clojurebot | "Elapsed time: 0.156546 msecs"\ntrue |
| 17:31 | gfredericks | ,(time ((fn [s] (.startsWith ^String s "foo")) "foo bar")) |
| 17:31 | clojurebot | "Elapsed time: 0.132136 msecs"\ntrue |
| 17:32 | gfredericks | that's a barely noticeable improvement |
| 17:32 | gfredericks | but *warn-on-reflection* confirms that the first reflects while the second doesn't |
| 17:32 | justin_smith | rasmusto: with 4clojure I actually ended up doing a loop with quot and div |
| 17:32 | justin_smith | but when you are starting with a string rather than number, that is a little different I think |
| 17:33 | rasmusto | oh, hm. I think I've done both |
| 17:33 | justin_smith | I am noisesmith on 4clojure, if you are curious |
| 17:33 | rasmusto | number -> string doesn't make much sense though, except for that palindrome one |
| 17:34 | justin_smith | I was amazed when I used jvisualvm and saw how much of my production app's ram was being used by objects of type string |
| 17:34 | justin_smith | *type char |
| 17:34 | rasmusto | all of my ram is in lazy-seqs usage, go figure |
| 17:35 | lunk | rasmusto: (get-a-job lazy-seqs) |
| 17:36 | rasmusto | lunk: both my code and myself share that trait |
| 17:36 | lunk | :P |
| 17:36 | rasmusto | though I'm not religious |
| 17:36 | mpenet | justin_smith: in my case it's used like that: (select :foo (where {:bar 1 :baz ?})) |
| 17:37 | mpenet | justin_smith: even tho I agree in most case it's probably a bad idea, here it kind of make sense |
| 17:37 | justin_smith | mpenet: so why not :? if it is a placeholder? |
| 17:37 | mpenet | justin_smith: keywords are encoded to strings, backward compatibility |
| 17:38 | mpenet | justin_smith: and I know for a fact that some user use this, had a PR recently for namespaced kw support |
| 17:38 | justin_smith | interesting |
| 17:39 | mpenet | justin_smith: the lib in question FYI https://github.com/mpenet/hayt |
| 17:44 | justin_smith | mpenet: I assume '? and \? have the same issues? |
| 17:44 | rasmusto | ,:\ |
| 17:44 | clojurebot | #<RuntimeException java.lang.RuntimeException: Invalid token: :> |
| 17:45 | justin_smith | ,\? |
| 17:45 | clojurebot | \? |
| 17:45 | jkj_ | gfredericks: (time (lookup "OH8GLV" (time/now))) ... "Elapsed time: 7.306 msecs" ... this does the operation thrise so time dropped from 42ms -> 7ms |
| 17:45 | mpenet | justin_smith: these 2 both work as well actually and generate the same as ? |
| 17:47 | mpenet | justin_smith: but it's not really intended to use like this (and has some potentialy suprising outcomes when used with one of the compilation modes) |
| 17:48 | jkj_ | gfredericks: well.. still takes 3 days. maybe i have to rethink the problem. bruteforce filtering lists might not be best way |
| 17:50 | gfredericks | jkj_: if you wanna get really serious reducers might help |
| 17:51 | gfredericks | specially if you have a big pile'o'cores |
| 17:54 | rasmusto | am I wasting my time building a command line interface with clojure.tools.cli? Should I just make a pretty repl instead? |
| 17:54 | jkj_ | gfredericks: thanks for the hint. |
| 17:55 | jkj_ | hmm.. cloud vendor wants to give me 16GB of mem with 8 cores |
| 17:57 | TEttinger | how much/month, jkj_? |
| 17:58 | TEttinger | http://www.ebay.com/itm/DELL-POWEREDGE-C1100-1U-2x-XEON-L5520-2-26GHZ-QUAD-CORE-CPUS-48GB-MEM-4x-160GB-/380633173091?pt=COMP_EN_Servers&hash=item589f814c63 |
| 17:58 | rasmusto | oh wow |
| 17:58 | jkj_ | TEttinger: ~450 with 30 GB of mem |
| 17:59 | jkj_ | eur |
| 17:59 | TEttinger | yep, ebay once again to the rescue |
| 17:59 | jkj_ | ok 285,34 eur 8 cores and 16 GB |
| 17:59 | TEttinger | a friend bought one with 72 GB RAM |
| 18:00 | TEttinger | I think it's working fine as an VM farm |
| 18:00 | TEttinger | he paid more than that though, that one is an especially good deal |
| 18:00 | jkj_ | i wonder how sun niagara processors go with clojure :P |
| 18:00 | jkj_ | huge amounts of cores |
| 18:00 | TEttinger | oooh |
| 18:01 | technomancy | jkj_: http://www.tbray.org/ongoing/When/200x/2008/05/01/Wide-Finder-2 |
| 18:01 | technomancy | though that task is IO-bound |
| 18:02 | TEttinger | holy wow http://www.ebay.com/itm/Sun-SPARC-T4-1-Base-1x8-Core-2-85GHz-32GB-and-6X300-GB-Disks-1x-Dual-8GB-FC-/390493525109?pt=COMP_EN_Servers&hash=item5aeb3a5475 |
| 18:02 | TEttinger | look at that price |
| 18:02 | gfredericks | hardware is one of my top 1 least favorite things to own |
| 18:02 | TEttinger | there is that |
| 18:03 | TEttinger | if something goes wrong you foot the bill |
| 18:04 | gfredericks | if I've got a bunch of CPU work to do, should I expect reducers to utilize the cores significantly more efficiently than pmap? |
| 18:05 | technomancy | pmap won't get too far ahead of the slowest entry |
| 18:05 | gfredericks | htop is telling me I'm only getting ~65% out of my 8 cores |
| 18:05 | TEttinger | gfredericks, I've kinda assumed reducers were the better version of pmap |
| 18:05 | technomancy | terrible if you have wide variance in costs |
| 18:05 | gfredericks | technomancy: that's an interesting point, hadn't thought about variance |
| 18:06 | gfredericks | I'm not sure how varied my work is here. I made a `pmap-chunkily` that chunks into ~20 jobs first |
| 18:06 | gfredericks | which should lower the variance |
| 18:06 | technomancy | as long as there's an "extra-chunky" version I'm for it |
| 18:06 | gfredericks | I don't have java 7 on this machine though |
| 18:06 | gfredericks | time to spelunk through apt I guess |
| 18:09 | rasmusto | gfredericks: I have a similar pmap issue, I'm about to reorganize stuff so I can get bigger chunks |
| 18:09 | gfredericks | might be hard to fit this problem to reducers well though |
| 18:10 | gfredericks | unless reducers are cooler than I thought |
| 18:12 | gfredericks | basic structure is (->> ten-thousand-things (mapcat make-lots-more-things) (map slow-fn) (apply max-key f)) |
| 18:12 | gfredericks | the seq coming out the mapcat would be too big for memory |
| 18:13 | jkj_ | gfredericks: there is ofcourse the possibility to try to be smart and just organize the data so that the lookup is faster... look up only map keys? |
| 18:14 | jkj_ | guess the map had some tree magic built in |
| 18:14 | akurilin | Quick sanity check: if I want to idiomatically get a gzipped json and parse it into a clojure map, is this flow reasonable: input stream of URL -> GZIPInputStream -> BufferedReader -> cheshire's parse-stream ? |
| 18:14 | dakrone | akurilin: sounds reasonable |
| 18:15 | jkj_ | then i just have to hammer the map by fetching OH8GLV, OH8GL, OH8G, OH8, OH, O to find the largest .startsWith |
| 18:15 | akurilin | Seems to work, but I've just never worked with java before so I'm wondering if I'm constructing too much stuff there :) |
| 18:15 | akurilin | dakrone, great, thank you! Do you have a notification on the cheshire keyword? :) |
| 18:15 | dakrone | akurilin: I do :) |
| 18:16 | akurilin | dakrone, not complaining, appreciate you supporting those really handy libraries. |
| 18:16 | jkj_ | akurilin: http://pastie.org/8350035 |
| 18:16 | dakrone | akurilin: no problem, glad they're working for you! |
| 18:17 | lunk | akurilin: not just a rtfm comment, but the java tutorials on streams are extremely valuable to read through *multiple* times, http://docs.oracle.com/javase/tutorial/essential/io/ |
| 18:17 | jkj_ | not cheshire there, but prolly works the same |
| 18:17 | akurilin | jkj_, thanks, good to have for reference! |
| 18:17 | jkj_ | (:import java.util.zip.GZIPInputStream) |
| 18:18 | akurilin | lunk, appreciate that. I've managed to avoid delving into java too much so far, but it'd be fun to find out what all the different tools out there are. |
| 18:18 | akurilin | lunk, plus that will explain the class hierarchy I was looking at just now. |
| 18:18 | lunk | akurilin: in this case, even if you don't consume a stream in a 'tight-buffered-loop', that doc set is very useful for understanding how to properly think about data streams in the jvm ecosystem |
| 18:19 | lunk | akurilin: ignore the class heirarchy for now, understand the concepts and you'll be golden |
| 18:19 | akurilin | lunk, that sounds really useful, I will! |
| 18:20 | akurilin | lunk, yeah it wasn't obvious what needed to wrap what in what order at a first glance, so that'll explain it hopefully. |
| 18:21 | lunk | akurilin: ultimately, it's just bits on a wire, you either read them or write them to the stream, the rest is gravy |
| 18:32 | clj_newb_2345 | is there a way to do the following with clojurescript? http://emacsrocks.com/e11.html |
| 18:32 | clj_newb_2345 | this is fucking insane |
| 18:33 | clj_newb_2345 | so basicallyI need a way to connect to the clojurescript repl from emacs |
| 18:37 | nDuff | clj_newb_2345: see piggieback |
| 18:40 | aaelony | has anyone tried nrepl + emacs on Mint Linux? I'm getting an nrepl window but no repl... |
| 18:41 | justin_smith | aaelony: I fixed this by using the elpa nrepl, after purging all my compiled elisp |
| 18:41 | justin_smith | same problem happened on latest ubuntu |
| 18:41 | justin_smith | I posted a so answer about how I fixed it, one moment |
| 18:42 | justin_smith | http://stackoverflow.com/questions/18208947/emacs-clojure-lein-and-nrepl-nrepl-buffer-shows-up-blank/18218870#18218870 |
| 18:42 | justin_smith | aaelony: ^ that link describes the fix |
| 18:42 | technomancy | justin_smith: using melpa before? |
| 18:43 | justin_smith | technomancy: no, I was actually using a git checkout of nrepl before |
| 18:43 | technomancy | huh |
| 18:43 | justin_smith | something weird happened when I rsynched from my old debian machine to the new ubuntu one, and everything broke |
| 18:44 | technomancy | yeah, elisp bytecode is not backwards-compatible |
| 18:44 | justin_smith | *git checkout of nrepl.el |
| 18:44 | technomancy | between versions of emacs |
| 18:44 | justin_smith | yeah, purging the elisp was probably the main thing |
| 18:44 | technomancy | the .elc files anyway |
| 18:44 | justin_smith | I am surprised it does not catch the incompatible elc versions and gripe or something |
| 18:45 | technomancy | yeah =\ |
| 18:45 | justin_smith | the pernicious thing is that if you have elc files loaded that were from a previous emacs, and you compile things with the new emacs, the breakage propagates |
| 18:45 | justin_smith | which leads to really weird hard to reproduce bugs |
| 18:45 | technomancy | further proof that AOT is the worst |
| 18:45 | justin_smith | yup |
| 18:46 | rasmusto | technomancy: so you recommend ^:skip-aot then? :P |
| 18:46 | aaelony | justin_smith: thanks, that's super. Will check out your write up... |
| 18:46 | justin_smith | imagine if gcc let you use libs from the wrong glibc, and generated output that would not run with the old glibc, or the new one, without complaining at any point |
| 18:47 | technomancy | rasmusto: there's an entry at the end of the FAQ explaining the way to go for uberjars |
| 18:47 | technomancy | profile isolation |
| 18:47 | justin_smith | aaelony: I think the process I descibe there may be overkill, finding and eliminating all .elc files may be sufficient |
| 18:47 | mtp | justin_smith: it did that during the 5-to-6 transition iirc :p |
| 18:47 | justin_smith | heh |
| 18:47 | rasmusto | technomancy: okay, I'll give it a read. Thanks :) |
| 18:47 | justin_smith | well never mind then |
| 18:47 | technomancy | rasmusto: should probably blog it or something to get the word out though |
| 18:48 | rasmusto | technomancy: I noticed it from the 3.0.0 deprecation warning, and when c.t.n.r/refresh-all would break after a `lein clean` |
| 18:48 | technomancy | protocols? |
| 18:48 | clojurebot | http://whiley.org/2012/02/29/a-problem-of-decoupling/ describes a problem solvable by protocols |
| 18:49 | technomancy | shut up clojurebot |
| 18:49 | rasmusto | no, nothing like that |
| 18:49 | rasmusto | lein check gives me a few reflection warnings, but nothing looks that scary |
| 18:51 | aaelony | justin_smith: that worked!! many thanks :) |
| 18:57 | justin_smith | aaelony: NP; that bug hounded me for ages, once I found a fix I wanted to put it up somewhere where others would find it |
| 19:03 | jkj_ | gfredericks: http://pastie.org/8350155 |
| 19:04 | jkj_ | gfredericks: i guess i might have found the solution |
| 19:12 | jkj_ | the calculation is just three hours now! |
| 19:15 | doomlord_ | can jit/inference techniques make a dynamically typed language like clojure match a statically typed language, or is that going to be a permanent issue with performance |
| 19:16 | technomancy | doomlord_: static types just let you avoid runtime reflection, which is pretty rare in Clojure code |
| 19:17 | justin_smith | technomancy: doesn't it also allow for example aggressive unboxing? |
| 19:17 | doomlord_ | on the one hand what with templates and HM type inference, i can imagine some sort of compiler/runtime producing performant code.. but is this a practical reality |
| 19:17 | technomancy | static typing helps you catch errors but doesn't provide much of a speed boost except for pathological code |
| 19:17 | doomlord_ | ^^(on the one hand, having seen templates and HM type inference..) |
| 19:17 | technomancy | justin_smith: yeah, but it already does a fair bit of inference there afaiu |
| 19:18 | justin_smith | the jvm can unbox in jit you mean? |
| 19:18 | technomancy | I don't think it can do more without giving up on certain interop features, but I could be wrong |
| 19:18 | technomancy | no, the clojure compiler infers primitive local types in some cases I think |
| 19:19 | justin_smith | that makes a lot of sense, java is pretty pathological in its level of boxing to start with, so you would want a different target vm before the unboxing had a point to it I guess |
| 19:19 | justin_smith | also immutability and unboxing are kind of mutually exclusive |
| 19:20 | justin_smith | I think |
| 19:20 | mercwithamouth | do any of you guys work through Purely Functional Data Structures?? Is it appropriate for lisp? |
| 19:20 | technomancy | tagged fixnums would get us a pretty significant perf boost for numerics |
| 19:21 | technomancy | but I don't think that's coming in any sort of reasonable timeframe |
| 19:23 | allenj12 | Hey is there a way you can test if a optional argument is passed to a function? does check true work? |
| 19:23 | technomancy | allenj12: I think you'd have to provide a default value with :or to distinguish between nil and not provided. |
| 19:24 | allenj12 | technomancy: hmm alright ty |
| 19:31 | clj_newb_2345 | nDuff: piggieback looks awesome; thanks |
| 19:39 | allenj12 | https://www.refheap.com/18930 can someone help me witht his? i asked the question in comments towards the bottom |
| 19:40 | callen | allenj12: lines 42-45 should just be a case. |
| 19:40 | callen | and don't have trailing tail parentheses. bad style. |
| 19:42 | callen | allenj12: 66-69 is another case |
| 19:42 | allenj12 | callen: a k ill fix that but is that causing the problem? i think its complaining about the main/a-star functions with prev-move |
| 19:50 | tufflax | allenj12: show the actual error message you get. Also, it looks like the paren at line 76 should not be there |
| 19:51 | tufflax | Also, def inside functions should be avoided |
| 19:51 | tufflax | use let instead |
| 19:51 | allenj12 | tufflax: kk ill show error message and fix those after i repaste |
| 19:52 | tufflax | allenj12: def creates a binding in the namespace, so using it inside functions have side effects |
| 19:52 | tufflax | has* |
| 19:53 | callen | allenj12: don't def in a function. |
| 19:54 | allenj12 | callen: yea changing that now |
| 19:55 | allenj12 | https://www.refheap.com/18930 here is the error |
| 19:56 | allenj12 | ill fix the defs |
| 19:57 | tufflax | allenj12: the error is not really hard to understand. prev-move does not exist in get-queue |
| 19:58 | allenj12 | tufflax: wow my bad for some reason i thought it was because of something else |
| 19:59 | tufflax | allenj12: also, why have you used & in the args list of a-star? You seem to want to only pass in 1 prev-move |
| 20:00 | allenj12 | tufflax: that was there originally cause when i first call a-star there would be no prev-move however i decided to just make a keyword to fill it in and see if that worked |
| 20:02 | tufflax | allenj12: the arg after & becomes a list (or seq, whatever) that collects the additional args, but you don't treat it that way. You can use (seq something) to check if it has items in it |
| 20:02 | tufflax | if something is the rest of the args (after &) |
| 20:03 | allenj12 | tufflax: i thought & was optional arguments |
| 20:03 | allenj12 | tufflax: does it do that through a sequence |
| 20:03 | tufflax | allenj12: yes, the optional args are collected in a seq/list |
| 20:03 | tufflax | even if there is only 1 |
| 20:04 | allenj12 | tufflax: ahh k so to check if it was passed i can test if the sequence is empty right? |
| 20:04 | tufflax | yes |
| 20:04 | allenj12 | tuffkax: ah! kk cool |
| 20:06 | tufflax | allenj12: you can also do something like this |
| 20:06 | tufflax | (defn special-sum ([] (+ 10 10)) ([x] (+ 10 x)) ([x y] (+ x y))) |
| 20:07 | justin_smith | core.async question: any tips on why each instance of pool-block in this code serves exactly one request, and dies right before printing the "wrapping up" line? |
| 20:07 | justin_smith | https://www.refheap.com/18932 |
| 20:07 | tufflax | At the bottom of the following page, there is a "+ source", click the + for a good example http://clojuredocs.org/clojure_core/clojure.core/= |
| 20:07 | tufflax | allenj12: ^ |
| 20:09 | allenj12 | tufflax: kk thanks gottt run machine learning grpup here ty |
| 20:32 | justin_smith | regarding my question above, it seems it was because I was not reading from the channels passed in, and a temporary workaround was to create the channels with a buffer so the go blocks could continue on after writing to each channel |
| 20:42 | brehaut | talios: nice readme ;) |
| 20:46 | talios | brehaut: shoosh - I need to add one ok :) |
| 20:47 | brehaut | talios: longer version: what is frege? |
| 20:48 | talios | brehaut: http://www.frege-lang.org/ - haskell for the jvm |
| 20:48 | brehaut | oh right. choice |
| 20:48 | talios | well, 99% haskell, its not 100% identical |
| 20:49 | brehaut | sure |
| 20:49 | brehaut | 100% haskell is hard |
| 20:52 | talios | and blammo - I hit a bug in my own plugin, gah :) |
| 20:59 | amehta | clojure stable release is 1.5.1 according to clojure.org. what version of clojurescript should i be using? |
| 21:01 | dnolen | amehta: 1859 works pretty well, 1889 if you like living closer to the bleeding edge |
| 21:01 | amehta | thanks david |
| 21:17 | _scape | so I have something like this: ({"bearing_Cylinder.001" {"v" [["v" "0.032977" "0.716507" "0.166762"] ["v" "0.033133" "0.716507" "0.165748"] ... ]}}) and I want to remove the "v" element on the vectors, I am doing all of these for loops with first and rest functions and basically getting no where. any hints? |
| 21:18 | tufflax | _scape: clojure.walk |
| 21:18 | _scape | more importantly, there are other elements such as f and n, all are first in the vector |
| 21:18 | _scape | ok, let me check it out |
| 21:18 | _scape | thx |
| 21:26 | dnolen | core.match 0.2.0-rc7 going out, I think finally killed all remaining sources of AOT issues, likely last release candidate |
| 21:31 | dnolen | er, I mean rc6 |
| 21:37 | arkh | dnolen: any recommended beginner reading for someone who's never worked with (lazy) pattern matching? |
| 21:37 | dnolen | arkh: it's not really lazy pattern matching, that's just describing the algorithm |
| 21:37 | dnolen | arkh: it's pattern matching same as you find in SML, OCaml, Scala, Haskell - but more flexible like Racket |
| 21:39 | arkh | dnolen: I haven't worked with any of those but I'll google it up. From what I've read on your repo it looks like something I could use though - thanks! |
| 21:40 | dnolen | arkh: it's like destructuring but you can use it for conditional logic |
| 21:42 | brehaut | dnolen: im curious, does core.match allow the same sort of totality analysis that a static language like haskell etc can provide? |
| 21:42 | brehaut | (ie, it knows if you have cases for every value of the matched thing) |
| 21:42 | dnolen | brehaut: not really because there's no restrictions about types |
| 21:42 | dnolen | brehaut: however it throws if you miss a case |
| 21:42 | brehaut | thats what i suspected |
| 21:43 | brehaut | oh nice |
| 21:43 | dnolen | brehaut: we could actually detect certain kinds of mistakes but it was experimental and I ripped it out when I overhauled core.match a couple months ago. |
| 21:44 | brehaut | its not like clojure is overly concerned with total functions anyway |
| 21:44 | brehaut | dnolen: i guess that would be better supported with information from say core.typed and the analyzer ? |
| 21:44 | dnolen | brehaut: however i bet you could overlay that restriction via core.typed |
| 21:44 | dnolen | brehaut: heh, we had the same thought |
| 21:45 | dnolen | brehaut: yes I think core.typed could do it's own type level analysis restriction what core.match is allowed to express |
| 21:47 | brehaut | dnolen: it will be interesting to see how a the affordances of typed matcher compares to haskell etc |
| 21:49 | brehaut | dnolen: from memory core.match doesnt depend need to upon concrete types as much as say haskell, right? |
| 21:53 | talios | brehaut: oh look - a readme! https://github.com/talios/frege-maven-plugin - and another new release, maybe I should do $work work tho :) |
| 21:53 | brehaut | talios: hah excellent |
| 21:55 | dnolen | brehaut: for optimization? |
| 21:55 | brehaut | dnolen: expressivity |
| 21:56 | brehaut | dnolen: i havent used haskell in a while, but i recall it being difficult to match on non concrete types |
| 21:56 | dnolen | brehaut: yes core.match is more expressive, Haskell doesn't really support matching on non-concrete types |
| 21:56 | brehaut | dnolen: F# has its active patterns to allow some dynamism in its matches; being more dynamic is of course something clojure generally has over these langs |
| 21:57 | brehaut | dnolen: right, so with that expressivity, but as type safe via core.typed. *waves hands* that'd be great |
| 22:19 | munderwo | So I have composure installed via lien. and the ring/ring-core is in my lein deps :tree list. But when from my repl I do (:use ring.util.response) or (:require ring.util.response) its says that CompilerException java.lang.ClassNotFoundException: ring.util.response, compiling:(NO_SOURCE_PATH:1:1) |
| 22:20 | munderwo | Am I missing something on how lein etc builds the class path? or is ring.util.response not in ring-core? |
| 22:23 | munderwo | I've also just installed [ring "1.2.0"] just to make sure I've got it all, and its still giving the same error. |
| 22:38 | kurisumasu | munderwo: (use 'ring.util.response)? |
| 22:39 | munderwo | kurisumasu: yeah, i've worked out that I can do that syntax in the repl, and that in my module it gets converted to (:use ring.util.response). Im still not sure what the ' is there for |
| 22:40 | kurisumasu | munderwo: (use & args) is a function. So when you call it as `(use ring.util.response)` it'll first try to resolve what "ring.util.response" means before giving it to the `use` function. |
| 22:41 | kurisumasu | munderwo: when you use the apostrophe you're quoting it, which basically means telling Clojure "hey, don't try to resolve this. Give it literally to the `use` function and they'll know how to resolve this name, ya?" |
| 22:42 | kurisumasu | munderwo: some things in Clojure are functions (like use), and they resolve their arguments before passing them to the function. Some are macros, where the macro gets to decide how they want to handle any argument, like the "(ns (:use x))" one. |
| 22:42 | munderwo | right. I'm a totally n00b at this point. just trying to work out how everything is working. Ahhhh, right. and thats what you have to do in the repl because ring.util.response isn't a symbol that can be resolved to something? |
| 22:43 | kurisumasu | munderwo: don't worry. Everyone starts as a beginner ;) |
| 22:43 | munderwo | where as when I'm using (ns) its a list of things it needs to do... |
| 22:44 | munderwo | Yeah I'm in that phase where everything is confusing, so I just need to work through that until I start getting whats going on :) |
| 22:45 | kurisumasu | munderwo: in the repl, `ring.util.response` is a symbol itself. But when you call the use function passing it as argument, Clojure will try to find what that symbol means, and it will fail because ring.util.response hasn't been loaded yet. |
| 22:46 | munderwo | yeah, right that makes sense. So this is the problem I'm having https://www.refheap.com/9c4ee1bec6666980eee5cc71b |
| 22:46 | kurisumasu | munderwo: the (ns) macro will eventually call the `use` function. In fact, you could do so in your module. The macro is just there to make your life easier, and make things more declarative :) |
| 22:48 | kurisumasu | munderwo: so, when you say `(:require x :as y)` you're telling Clojure that `x` and `y` are the same thing. But those things are namespaces (collections of functions and values), not a function per se. You have to use `y/name-of-the-function` to refer to the function itself. |
| 22:49 | kurisumasu | munderwo: in this case, you can use (:use ring.util.response), and it'll make anything in the `ring.util.response` namespace available in your module directly. |
| 22:50 | munderwo | right, which is what I had before, but somebody on the irc was like, its bad form to use 'use'. so Im just trying to get it working with :require. and I can understand why you would use use. (way to many uses of the work use in that last sentence :) |
| 22:51 | munderwo | so could I do (:require ring.util.response/response :as response) |
| 22:52 | kurisumasu | munderwo: you'd usually want to minimise the number of (use x) you have because you basically get all of that inside your current namespace. But you can use (:use x :only (y z)) to grab only `y` and `z` out of `x`. |
| 22:52 | akurilin | What are people's thoughts on Liberator? Would love to hear the opinion of folks who have used it over the long term. |
| 22:53 | munderwo | kurisumasu: so if you were to import this ring function how would you do it? I tried (:require ring.util/response :as response) and also (:require ring.util.response/response) |
| 23:02 | allenj12 | Hey so i changed some Defs to lets in my functions and now im getting some errors https://www.refheap.com/18935 i explained more in comments |
| 23:04 | kurisumasu | munderwo: for these ring things I'll usually just place them in the use section, so (:use ring.util.response) |
| 23:05 | munderwo | ok. cool. thanks! |
| 23:06 | sinistersnare | allenj12: lets denote scope using the parens |
| 23:06 | sinistersnare | so you need to wrap the whole scope of the let using the right paren |
| 23:07 | sinistersnare | wait you did that :p i think |
| 23:07 | sinistersnare | ill shutup |
| 23:07 | allenj12 | sinistersnare: lol i was about to say i was rly confused then |
| 23:08 | allenj12 | sinistersnare: turns out i missed a parens but i put that back in and same problem |
| 23:08 | sinistersnare | oh :p sorry! |
| 23:09 | allenj12 | sinistersnare: its ok lol any other possible problems |
| 23:12 | allenj12 | yea the error im getting is "Dont know how to create ISeq from: clojure.lang.Symbol ? in general what does that mean? |
| 23:12 | brehaut | allenj12: it means you probably have a symbol in your 'ns declaration where you should have a list or vector |
| 23:13 | sinistersnare | yeah, usually means some sort of missing thing! |
| 23:14 | allenj12 | brehart: o shit no brackets fml lol |
| 23:15 | brehaut | the 'haut' is pronounced 'oh'; i blame the french |
| 23:17 | talios | wait, and here I was thinking you were hot. |
| 23:17 | talios | I need to rethink my stalking. |
| 23:17 | brehaut | haha |
| 23:18 | talios | I keep thinking Andrew Braveheart, hero of functional developers in the southern northern lands |
| 23:18 | brehaut | well this has gone to a weird place |
| 23:19 | talios | and? :) |
| 23:19 | brehaut | lol |
| 23:21 | brehaut | talios: he works with you ? |
| 23:21 | talios | yup yup - javascript/coffeescript side of things |
| 23:21 | brehaut | i did not realise |
| 23:22 | talios | see the things you're learn if you came and worked for us? |
| 23:23 | brehaut | talios: coffeescript and java legacies? |
| 23:24 | sinistersnare | can anyone figure out how to add local jars to my clojure project? |
| 23:24 | talios | currently - the sooner we can get some clojure in the better ( or scala, or frege ;p ) |
| 23:26 | sinistersnare | is this what :resource-paths is for? to the sample project.clj! |
| 23:26 | TEttinger | sinistersnare, I kinda hacked it by extracting into a directory and naming that directory with that, :resource-paths |
| 23:26 | talios | that would be resource files, you really want to put those jars into a local repository |
| 23:26 | talios | i would think |
| 23:26 | sinistersnare | ive tried a local repository, its such a pain though |
| 23:26 | sinistersnare | i was having the hardest time figuring out the maven commands =\ |
| 23:27 | allenj12 | my clojure writing style is horrible there parens everywhere! |
| 23:27 | sinistersnare | i hope lein adds a feature for this :) |
| 23:27 | sinistersnare | i hear there was one, but it was deprecated... |
| 23:27 | kurisumasu | allenj12: you can try Clochure, instead :) |
| 23:27 | kurisumasu | allenj12: it'll certainly reduce the amount of parenthesis, at least. |
| 23:28 | allenj12 | kurisumasu: i just need practice :) ill look that up tho |
| 23:28 | s4muel | http://imgs.xkcd.com/comics/lisp_cycles.png |
| 23:28 | allenj12 | kurisumasu: new to functional languages and clojure |
| 23:29 | mgaare | allenj12: (pst, it's important you realize that clochure is a joke) |
| 23:29 | kurisumasu | allenj12: ahaha. It was a joke :) |
| 23:29 | allenj12 | LMAO |
| 23:29 | allenj12 | just looked at it |
| 23:30 | kurisumasu | allenj12: Most Lisp dialects really do have problems with parenthesis, actually, because the syntax isn't rich enough like Smalltalk and such languages. Clojure is at least a little bit better in that it uses different symbols for maps/vectors. |
| 23:31 | kurisumasu | The only Lisp dialect I know of that features a different syntax is Dylan — no parenthesis there. |
| 23:31 | allenj12 | kurisumasu: yea i need to get use to it i think i might clean up my a-star function by making it two functions https://www.refheap.com/18937 |
| 23:31 | allenj12 | huh |
| 23:31 | allenj12 | lol |
| 23:32 | mgaare | kurisumasu: I dunno, is {:this :that} and [:this :that] so vastly superior to (hash-map :this :that) and (vector :this :that) |
| 23:32 | allenj12 | wait cant i check if an optional paramter is passed with empty? |
| 23:33 | allenj12 | like (defn two [& too] (if empty? too) (println check)) |
| 23:34 | mgaare | allenj12: yeah that should work (although only if you have defined check somewhere) |
| 23:34 | kurisumasu | allenj12: oh, we usually place the closing parenthesis in the same line as the last expression in a form, rather than leaving them on a line of their own. Just a stylistic thing. |
| 23:34 | allenj12 | kurisumasu: can u give me an example |
| 23:34 | tufflax | allenj12: you are using the wrong syntax |
| 23:35 | tufflax | allenj12: just like with let before :P |
| 23:35 | tufflax | maybe you should look up some examples before trying :p |
| 23:35 | kurisumasu | mgaare: perhaps. I feel like they help identifying patterns quicker though, but the only other major Lisp I play with is Emacs Lisp, which is a little... |
| 23:35 | allenj12 | tufflax: omg sry i did before my focus on school has been all scatterd so i prolly misremembered |
| 23:36 | sinistersnare | can anyone see if anything is wrong with my project.clj for lein? https://gist.github.com/sinistersnare/6665526 i think i did :main correclty... |
| 23:36 | allenj12 | tufflax: wait whats wrong with my if statements? |
| 23:37 | allenj12 | tufflax: im looking at examples |
| 23:37 | s4muel | mgaare: consider (let [{:keys [one two]} m]) . That syntax would be nasty with all parentheses, i think. You'd have to call some special function to destructure. |
| 23:37 | tufflax | allenj12: I think you meant (if (empty? too) (println check)) |
| 23:37 | mgaare | s4muel: yeah, destructuring is awesome |
| 23:38 | tufflax | tufflax: the syntax was valid, but nonsensical :P |
| 23:38 | allenj12 | tufflax: o yea sry i thought u ment in my code |
| 23:38 | tufflax | hm no in your message :P |
| 23:38 | mgaare | allenj12: whoa that a-star function is making me dizzy ;) |
| 23:39 | allenj12 | mgaare: me to i know its a mess |
| 23:39 | allenj12 | mgaare: and ofcourse its not working properly |
| 23:39 | mgaare | allenj12: you know you can do more than one thing in a let at a time |
| 23:39 | allenj12 | mgaare: what do u mean? |
| 23:40 | tufflax | allenj12: start indenting your code properly. Also, nice trails of parens, looks like some sort of art |
| 23:40 | kurisumasu | allenj12: (let [a 1 b 2] (+ a b)) |
| 23:40 | kurisumasu | allenj12: unlike other lisps, Clojure's `let` will resolve [a 1] first, then [b 2], so you can use `a` in the second expression. |
| 23:41 | mgaare | kurisumasu: other lisps don't do that? |
| 23:41 | kurisumasu | mgaare: they have `let*` for that, but `let` doesn't guarantees ordering |
| 23:41 | mgaare | good to know! |
| 23:41 | allenj12 | kurisumasu: interesting ill change that but first i wanna see why when i run this its not working proprly it keep making a back move |
| 23:41 | mgaare | our beloved let is much like haskell's monadic do notation |
| 23:42 | sinistersnare | it says that it cannot find my main class cljdx.core |
| 23:42 | sinistersnare | which is the namespace for my file core.clj |
| 23:43 | brehaut | mgaare: let is only do notation for the identity monad |
| 23:43 | sinistersnare | well thats when i make a standalone jar out of it, when i run it, it gives me some error about not finding jars, but i think thats because it needs to be packaged into a jar |
| 23:43 | brehaut | mgaare: for and doseq is do notation for lazy and strict list respecitively |
| 23:44 | mgaare | brehaut: they are aren't they... pretty cool |
| 23:57 | yeoj___ | i'm a beginner, and i find i do a lot of work lately getting my namespaces correct. Is there a way to clear all namespace information in emacs without stopping/starting the nrepl again? |
| 23:57 | yeoj___ | i have conflicts contstantly for things that i've already loaded |
| 23:57 | tufflax | yeoj___: this might help https://github.com/clojure/tools.namespace |
