2014-03-19
| 00:00 | ddellacosta | noprompt: okay |
| 00:01 | ddellacosta | noprompt: sorry to keep you waiting. So yeah, one of things I think would be useful would be to have a lookup function or functions for routes |
| 00:07 | ddellacosta | $seen seangrove |
| 00:07 | lazybot | seangrove was last seen quitting 4 hours and 50 minutes ago. |
| 00:08 | ddellacosta | hmm, how do I send messages to folks? |
| 00:08 | amalloy | $help mail |
| 00:08 | lazybot | amalloy: Send somebody a message. Takes a nickname and a message to send. Will alert the person with a notice. |
| 00:08 | ddellacosta | amalloy: thanks |
| 00:09 | noprompt | ddellacosta: i think by having IRouteValue and adding the "raw" route as a third element in the vector is good enough for now w/o breaking the current API |
| 00:10 | ddellacosta | noprompt: okay, gotcha--so, basically, leave the locate-route modifications roughly as is, but move the route-spec fn to its own IRouteValue protocol? Seems reasonable |
| 00:29 | ddellacosta | noprompt: wondering what I should call the function that returns the actual route value...I have "route-spec" now, but maybe something like "get-original" or "original-route" or..."route-value?" |
| 00:29 | noprompt | ddellacosta: right. |
| 00:29 | ddellacosta | noprompt: the one inside the IRouteValue protocol I mean |
| 00:30 | noprompt | (defprotocol IRouteValue (route-value [route] "Return the original route value")) |
| 00:32 | noprompt | ddellacosta: basically, i'm trying to make sure we avoid adding too much functionality at one. |
| 00:32 | noprompt | ddellacosta: if that's enough to help you solve your problem while not assuming too much i think that's good. |
| 00:33 | noprompt | ddellacosta: by creating a separate protocol and returning the "raw" route it's a win for now and the future. |
| 00:33 | ddellacosta | noprompt: absolutely, and I definitely am on the same page with that--my goal with this is to "disturb" the codebase as little as possible. |
| 00:33 | ddellacosta | noprompt: alright, will make those modifications and push |
| 00:33 | noprompt | ddellacosta: lol did you see the shit i pushed yesterday?! |
| 00:33 | ddellacosta | noprompt: no, which? Will check it out |
| 00:33 | noprompt | ddellacosta: i added like 100 lines just so we could serialize/deserialize vectors/maps etc. for query strings. |
| 00:34 | noprompt | lol, it doesn't break anything and no one will notice but it's there. |
| 00:34 | dissipate | noprompt, what is there? |
| 00:35 | ddellacosta | noprompt: oh, yeah, I did see that now that you mentioned it. Nice stuff. :-) |
| 00:35 | ddellacosta | dissipate: ^ |
| 00:36 | noprompt | ddellacosta: somehow we've ended up in this space where query params turns out to be useful. |
| 00:37 | ddellacosta | noprompt: you know, it's interesting how the use-cases shift when you are doing client-side routing actually. Although maybe you weren't being tongue-in-cheek... |
| 00:38 | noprompt | ddellacosta: it's a weird nasty place. |
| 00:38 | ddellacosta | noprompt: hahaha |
| 00:38 | noprompt | the client sucks. but it's getting better all the time. |
| 00:38 | ddellacosta | noprompt: "the client..." web clients in general? |
| 00:39 | noprompt | ddellacosta: i just meant the front end. it's a mixed bag. |
| 00:39 | ddellacosta | noprompt: definitely agreed on both counts if that's what you mean. |
| 00:39 | ddellacosta | noprompt: yeah. Absolutely. But having done this since around 1998 I can say definitively we are both better off and worse off, depending on how you look at it. |
| 00:40 | noprompt | ddellacosta: considering that we might've had a chance at scheme i really don't know. |
| 00:40 | ddellacosta | noprompt: there are things you can do now that were inconceivable ten years ago. And the compatibility story has gotten a *lot* better, all my complaints about IE aside. However, the amount of complexity involved is astounding, when you take web development as a whole. |
| 00:41 | ddellacosta | noprompt: you mean, as opposed to what JS turned out to be? |
| 00:41 | noprompt | ddellacosta: well, legend has it, that javascript was supposed to be a scheme like language but netscape changed it at the last minute. |
| 00:42 | dissipate | noprompt, and we have been stuck with it ever since. :( |
| 00:42 | noprompt | you know, because java. |
| 00:42 | danielszmulewicz | Has Datomic memory requirements changed? I'm trying to upgrade to the latest version (0.9.4609), but I get an out of memory error. I have 1GB of memory on my VPS where datomic (0.8.something) worked flawlessly. |
| 00:42 | ddellacosta | noprompt: right, I remember that B. Eich was big into scheme, wasn't he...and Self? |
| 00:42 | ddellacosta | noprompt: well, good thing he was able to include at least some parts of it, conceptually |
| 00:43 | dissipate | ddellacosta, yes, and a guy named Douglas Crockford tried to convince us that there are actually 'good' parts of javascript |
| 00:43 | ddellacosta | dissipate: well, when you have to use a language, there have to be ways to cope |
| 00:43 | noprompt | ddellacosta: from what i've read. i've never met the man though. he could be a serial rapist serving time in prison and a robotic stunt double programmed to speak/write/tweet on his behalf is filling his role as a human until his release. |
| 00:44 | noprompt | dissipate: yeah but he gave us the term "dog balls" (function () {}()) <-- dog balls |
| 00:44 | dissipate | ddellacosta, and there in lies the problem. javascript should have never existed. it should all be byte code, and no i am not talking about the JVM |
| 00:45 | ddellacosta | hahaha |
| 00:45 | ddellacosta | you guys |
| 00:45 | noprompt | dissipate: yes and it would have been "cool" if money never existed to create classes and social problems but hey, what can you do? |
| 00:45 | dissipate | noprompt, that's hawt. :O |
| 00:46 | noprompt | lol |
| 00:46 | dissipate | noprompt, well, i suppose the USSR tried that. but even they still had money. |
| 00:47 | noprompt | ddellacosta: i've only been programming for around 4.5 years so i can't speak to the horrors prior to my arrival on the scene. |
| 00:47 | noprompt | css is pretty horrible i'll say that. |
| 00:47 | ddellacosta | noprompt: I remember Netscape 4 |
| 00:47 | ddellacosta | noprompt: and DHTML |
| 00:47 | ddellacosta | noprompt: I barely escaped with my life |
| 00:47 | noprompt | i remember it too, but probably not the way others did. |
| 00:48 | dissipate | noprompt, the horrors continue |
| 00:50 | noprompt | ddellacosta: one day i'm gonna have enough money saved up to go to japan. :-/ |
| 00:51 | ddellacosta | noprompt: well, if I'm here you can crash at my place. :-) |
| 00:51 | noprompt | ddellacosta: i *really* wanna shop for vinyl therer. ;_; it's too much on ebay. |
| 00:51 | ddellacosta | noprompt: oh, seriously, the Japanese are nuts about that stuff |
| 00:51 | ddellacosta | noprompt: like in an awesome/scaryw ay |
| 00:51 | noprompt | ddellacosta: those guys sell me a record for like 20 bux in yen and then it get's here with a price tag in yen for half the price. |
| 00:51 | ddellacosta | *scary way |
| 00:52 | noprompt | ddellacosta: there's so much good 60s/70s pop from japan. |
| 00:52 | noprompt | <3 it |
| 00:52 | dissipate | how is the clojure scene in japan? |
| 00:53 | noprompt | i only open iterm to spawn another instance of emacs |
| 00:59 | ddellacosta | dissipate: not much of one |
| 01:00 | dissipate | ddellacosta, well, i guess japan has bigger problems than programming languages. their entire race is pretty much dying out. |
| 01:00 | jph- | dissipate, slow population growth, not dying out |
| 01:01 | dissipate | jph-, slow as in negative? |
| 01:01 | noprompt | "race" lol |
| 01:01 | jph- | dissipate, im not sure what the latest figures are, but it could be in negatives atm |
| 01:01 | dissipate | jph-, it is indeed negative |
| 01:02 | ddellacosta | noprompt: I'm a bit confused as to how to do this, as the only place I can get ahold of the original route is in compile-route |
| 01:25 | jph- | im looking at some clojure 1.4.0 code |
| 01:25 | jph- | and i see (defn somefunc [_] ...) |
| 01:26 | jph- | what does _ mean? or has that been superceded since 1.4? |
| 01:28 | SegFaultAX | jph-: If that's exactly what was written, it's probably just a placeholder for the arg list. |
| 01:29 | SegFaultAX | Or rather, the parameter list. |
| 01:29 | jph- | so like |
| 01:29 | jph- | (defn f [& args] ...) equivalent? |
| 01:29 | jph- | this is the exact line: (defn get-iso [c _] (:iso-code c)) |
| 01:30 | SegFaultAX | No. [& args] is variadic (it collects all unbound args into a single list) |
| 01:30 | jph- | i've not encountered _ before so i'm not sure how it's used |
| 01:30 | SegFaultAX | So [a b c & rest] binds the first three arguments to parameters a b and c, then everything after that to args. |
| 01:31 | SegFaultAX | _ isn't meaningful in any way, it's universally used as the "I don't care about this value" name. |
| 01:31 | jph- | oh |
| 01:31 | jph- | got it |
| 01:31 | jph- | devnull of parameteres |
| 01:32 | SegFaultAX | If you saw a function whose parameter list was literally [_], that makes no sense at all. |
| 01:32 | jph- | yep |
| 01:32 | jph- | itd be (fn [] ...) |
| 01:32 | SegFaultAX | Right. And the fn would be side-effecting. |
| 01:33 | jph- | just not sure what has changed between clojure 1.4 and 1.5 |
| 01:33 | jph- | whether stuff i've looked at is deprecated or not |
| 01:56 | amalloy | SegFaultAX: why would a function whose parameter list is [_] make no sense? |
| 01:56 | amalloy | it's perfectly reasonable to write something like (defn always-3 [_] 3) for passing to map |
| 01:56 | amalloy | like, of course you can use constantly, but defining a named function is fine too |
| 01:56 | SegFaultAX | Maybe if constantly didn't exist. |
| 01:57 | SegFaultAX | Well ok, it makes sense in a very limited context I suppose. |
| 02:00 | SegFaultAX | amalloy: Are you saying it's common to have functions that take a single argument which is never actually used? |
| 02:00 | amalloy | no, i would expect it to be more common to use constantly. but to say that it doesn't make sense at all? that's very different from uncommon |
| 02:02 | SegFaultAX | I mean a nullary function that is purely side-effecting is fairly common. But a unary function that never references its argument? |
| 02:06 | xuser | I have seen (let [x (fn) _ (println x) x (some other fn)]... |
| 02:07 | SegFaultAX | Well that's strange, why put it in the let at all? Let forms have implicit do-blocks. |
| 02:07 | SegFaultAX | Oh rebinding it, yea. |
| 02:16 | amalloy | SegFaultAX: it's a unary function because it's being passed to something else that expects a unary function |
| 02:16 | amalloy | you wouldn't define it that way if you never passed it around |
| 02:17 | xuser | amalloy: how long have you being doing clojure? |
| 02:17 | amalloy | for like...4 years? |
| 02:17 | amalloy | ish |
| 02:19 | xuser | old school ;) |
| 03:09 | seangrove | ,(let [x (atom {}) y (atom {})] (swap! x assoc :y y)) |
| 03:09 | clojurebot | {:y #<Atom@118712b: {}>} |
| 03:09 | seangrove | ,(let [x (atom {}) y (atom {})] (swap! x assoc :y y) (swap! y assoc :x x) |
| 03:09 | clojurebot | #<RuntimeException java.lang.RuntimeException: EOF while reading> |
| 03:09 | seangrove | ,(let [x (atom {}) y (atom {})] (swap! x assoc :y y) (swap! y assoc :x x)) |
| 03:09 | clojurebot | {:x #<Atom@135371: {:y #<Atom@cb3966: {:x #<Atom@135371: {:y #<Atom@cb3966: {:x #<Atom@135371: #>}>}>}>}>} |
| 03:10 | seangrove | I feel like I pulled a hiredman |
| 03:10 | petehunt | what did i just see |
| 03:11 | petehunt | my emoticons are confusing me :P sec |
| 03:11 | seangrove | Heh, x => y => x => y... |
| 03:12 | seangrove | also... |
| 03:12 | petehunt | so it worked |
| 03:12 | petehunt | or are they pointing to the wrong versions of x and y |
| 03:12 | seangrove | ,(let [a [(atom nil)]] (reset! (first a) a)) |
| 03:12 | petehunt | or am i not making any sense |
| 03:12 | clojurebot | [#<Atom@45d316: [#<Atom@45d316: [#<Atom@45d316: [#<Atom@45d316: [#<Atom@45d316: #>]>]>]>]>] |
| 03:13 | seangrove | Seems like clojure allows it, it just sensibly stops printing the structure after a bit |
| 03:13 | seangrove | Hadn't really come across this before |
| 03:14 | petehunt | seangrove: so how would you model a mutual friendship in an om app |
| 03:14 | petehunt | what would your state atom look like |
| 03:15 | seangrove | {:users [{:id :pete :name "Sam"}{:id :aaron :name "Aarom}] :friendships [[:pete :aaron]]} ? |
| 03:16 | seangrove | woops, [[:sam :aaron]] |
| 03:17 | petehunt | so you would refer to them by id |
| 03:17 | seangrove | Or if I needed fast lookups {:users [{:id :sam :name "Sam" :friends [:aaron]}{:id :aaron :name "Aarom} :friends [:sam] :friendships [[:sam :aaron]]} |
| 03:17 | petehunt | got it |
| 03:17 | seangrove | Yeah, I think so |
| 03:17 | seangrove | Just guessing off the top of my head though, and it's way past my bedtime |
| 03:18 | seangrove | I'm off, best of luck out there ;) |
| 03:50 | sm0ke | when i do a lein with-profiles repl, i get 'nrepl not in classpath' error? |
| 03:50 | sm0ke | why is that? |
| 03:51 | sm0ke | lein with-profiles myprofile repl |
| 03:59 | sm0ke | +myprofile |
| 04:50 | borkdude | Is there a way to get all possible keyword paths from a nested map (as in json-like thing), like: {:a {:b "d" :c "f"} :b nil} -> [[:a :b] [:a :c] [:b]] ? |
| 05:11 | szymanowski | hello, is there a way to add a keyval to an array-map and be sure that it doesn't become a hash-map? |
| 05:17 | mpenet | szymanowski: I don't think so. You can always use a LinkedHashMap instead if that's absolutely necessary |
| 05:18 | szymanowski | thank you, i don't know LinkedHashMap, i will look at it |
| 05:18 | clgv | szymanowski: no. it will automatically change if a certain threshold is reached |
| 05:19 | clgv | szymanowski: you can use a sorted-map though if you require a certain order defined by a comparison predicate |
| 05:21 | szymanowski | yes i've tried with sorted-map but it doesn't really fit my needs very well, maybe https://github.com/amalloy/ordered |
| 05:21 | clgv | szymanowski: ah you want to have insertion order |
| 05:21 | szymanowski | yes |
| 05:21 | clgv | well then go ahead and use amalloy's lib |
| 05:22 | szymanowski | yes it looks nice |
| 05:33 | jph- | and the penny drops for defrecord/defprotocol |
| 05:33 | jph- | this stuff is sweet |
| 05:43 | mskoud | Trying to use eval (know its evil) using a ring webserver. When run in the repl it ok and in the correct namespace, but when run through webserver the eval evaluates in the clojure.core namespace. How do i avoid that? |
| 05:47 | clgv | mskoud: despite you probably dont want to hear this. you are very likely doing something wrong by using "eval" |
| 05:48 | clgv | mskoud: you can fix it if you put an "(in-ns 'my.desired.target.ns)" in the code you pass to `eval` |
| 05:48 | hyPiRion | yeah, a `(do (in-ns 'my-ns) ~body) should work |
| 05:49 | clgv | mskoud: could you please report here when that part of the code backfired at you? ;) |
| 05:50 | mskoud | im trying to interpolate a string which can contain sexprs |
| 05:51 | clgv | must be a huge advantage compared to templating otherwise I'd use the latter |
| 05:52 | rstandy | what libraries do people use to implement authentication in clojure web applications? |
| 05:53 | hyPiRion | friend |
| 05:54 | jph- | rstandy, im just using noir session |
| 05:55 | rstandy | hyperboreean, jph-: thanks |
| 05:55 | rstandy | will check that |
| 05:56 | rstandy | jph-: does noire already manage authentication then? |
| 05:56 | rstandy | noir, I mean |
| 05:56 | jph- | if you look at a basic luminus template (lein new luminus blah +site) it does the utmost basics |
| 05:56 | rstandy | jph-: ah ok |
| 05:56 | jph- | the session stuff is relatively mature |
| 05:57 | jph- | but it's up to you to do the registration and login verification |
| 05:57 | jph- | you could drop in various hashing algorithms, or maybe something fancier |
| 05:57 | rstandy | jph-: ok, got it |
| 05:58 | rstandy | I would like to use something to extends for implementing authentication over LDAP or via oauth |
| 05:58 | clgv | rstandy: I used "friend" |
| 05:58 | jph- | i'd start with friend if you need something more full-featured |
| 05:58 | rstandy | clgv, hyPiRion: do you use it in production? |
| 05:58 | clgv | rstandy: for webform-based authentication it did not need much customization to get it running |
| 05:59 | hyPiRion | rstandy: no, not in production. |
| 05:59 | hyPiRion | Well, unless you mean my private websites-production. |
| 05:59 | rstandy | hyperboreean: ok, thanks |
| 05:59 | clgv | rstandy: since I am working at a university "use in production" likely does not have the same meaning as for you. but yeah, it is in production |
| 06:00 | rstandy | clgv: do you have a certain number of people using it daily? |
| 06:01 | rstandy | clgv: is the login page accessibile from the web? Just to see an real example of friend auth |
| 06:01 | clgv | rstandy: since it is no classical website. no. 2-3 people use it when the application is running |
| 06:02 | rstandy | clgv: ok, thanks for the info |
| 06:03 | rstandy | oh, nice, there is a friend-demo at herokuapp.com |
| 06:04 | rstandy | will try it |
| 06:04 | clgv | yeah right, there are demos for several authentication workflows |
| 06:06 | rstandy | clgv: really interesting library, from a very smart guy, I see (cemerick) |
| 06:09 | clgv | rstandy: yeah. though some people complained in IRC that it would be to complicated to get started with it. but I do not agree with that |
| 06:20 | rstandy | clgv: will definitely take a look at friend (which is also intriguing for its clever name) |
| 06:24 | myguidingstar | hi all, is there any way to call a function from project's dev dependencies from a leiningen plugin? |
| 06:25 | xsyn | myguidingstar: look at https://github.com/zcaudate/vinyasa |
| 06:26 | xsyn | or you can use the full path |
| 06:26 | myguidingstar | let's see |
| 06:26 | xsyn | http.async.client/function |
| 06:26 | xsyn | as an example |
| 06:34 | mskoud | clgv: works in repl (eval `(do (clojure.core/in-ns 'cms.core) (+ 1 3))) but returns "java.lang.IllegalStateException: Can't change/establish root binding of: *ns* with set" when run through ring server. |
| 07:04 | honza | does anyone have a post with how to set up piggieback + vim-fireplace with the clojurescript repl? |
| 07:14 | clgv | mskoud: (binding [*ns* 'cms.core] ...) could be an alternative |
| 07:15 | clgv | mskoud: but you see that you are already shooting yourself in the foot, right? |
| 07:16 | jph- | how do i case the class of a var? (case (class var) java.math.BigDecimal "yay!" "nope") doesnt seem to work |
| 07:17 | clgv | jph-: you cant literally with case |
| 07:17 | jph- | what's the right way to do it? |
| 07:18 | clgv | jph-: it is the same issue as with static constants |
| 07:18 | jph- | i know i can do (instance? ...) |
| 07:18 | jph- | but that's not case |
| 07:18 | clgv | jph-: yeah, (cond (instance? java.math.BigDecimal v) ...) |
| 07:18 | jph- | oh yeh i'd forgotten about cond |
| 07:19 | clgv | jph-: you could also use the helper macro that was suggested for static constants ( I think that was on the mailing list lately) |
| 07:21 | clgv | jph-: depending on whether that dispatch happens in a hotspot you might want the case behaviour instead of cond which expands to nested ifs |
| 07:43 | mskoud | clgv: yes maybe. trying to make some content able to pull info from some data in the code. |
| 08:28 | voldyman | whats the best method of getting xml data from a url? |
| 08:29 | jcromartie | depends on a lot of factors |
| 08:29 | voldyman | tried slurp but it converts '<' to < etc. |
| 08:29 | jcromartie | slurp does not do that |
| 08:29 | voldyman | it does for me. |
| 08:29 | jcromartie | ,(slurp "http://www.w3schools.com/xml/note.xml") |
| 08:29 | clojurebot | #<SecurityException java.lang.SecurityException: denied> |
| 08:30 | jcromartie | well anyway |
| 08:30 | jcromartie | it doesn't |
| 08:30 | voldyman | jcromartie: you are right, only half of the tags are escaped |
| 08:30 | jcromartie | can you share the URL to the XML document? |
| 08:31 | voldyman | https://itunes.apple.com/us/rss/topsongs/limit=10/explicit=true/xml |
| 08:32 | voldyman | oh fish, solved the problem, i think |
| 08:34 | voldyman | (clojure.data.xml/parse "<a><b>a</b><b>c</b></a>") |
| 08:34 | voldyman | ,(clojure.data.xml/parse "<a><b>a</b><b>c</b></a>") |
| 08:34 | clojurebot | #<ClassNotFoundException java.lang.ClassNotFoundException: clojure.data.xml> |
| 08:36 | voldyman | error in clojure.data.xml/parse https://www.refheap.com/61821 |
| 08:41 | hyPiRion | ,(require 'clojure.data.xml) |
| 08:41 | clojurebot | #<FileNotFoundException java.io.FileNotFoundException: Could not locate clojure/data/xml__init.class or clojure/data/xml.clj on classpath: > |
| 08:50 | jstew | I think that clojurebot doesn't like it because it's an external lib |
| 08:51 | hyPiRion | yeah |
| 08:51 | hyPiRion | not bundled with vanilla clojure |
| 10:10 | Bronsa | ,(defrecord x [a]) |
| 10:10 | clojurebot | sandbox.x |
| 10:10 | Bronsa | ,(x. 1 nil {:a 2}) |
| 10:10 | clojurebot | #sandbox.x{:a 1, :a 2} |
| 10:10 | Bronsa | voila` |
| 10:11 | jcromartie | What do most of y'all use for blogging platforms? Is Tumblr ever a good choice for code-related blogs? I don't see many technical posts there. |
| 10:12 | jcromartie | I should probably host it myself :) I have no excuse |
| 10:12 | edbond | jcromartie, I vote for github pages |
| 10:12 | jcromartie | edbond: yeah, not a bad idea |
| 10:14 | jcromartie | rockin' |
| 10:19 | jstew | Tumblr would be awesome for a tech related blog. Make animated gifs of nrepl. |
| 11:04 | luxbock | I have two protocols that implement largely the same functionality |
| 11:04 | luxbock | at first I wrote them as normal functions for both cases, but then I decided a protocol would simplify things |
| 11:04 | luxbock | but now I'm wondering how I should organize my namespace around this new change |
| 11:05 | luxbock | err. I have just one protocol that is implemented by two different records I should say |
| 11:06 | luxbock | is it standard practice to throw all of that stuff in a separate my-proj.impl.X namespace and then just build an outward facing API from those bits later? |
| 11:07 | mpenet | luxbock: it's not unusual yes |
| 11:07 | luxbock | alright, guess that's what I'll do then |
| 11:07 | luxbock | glad I found clj-refactor.el |
| 11:17 | shep-werk | luxbock: thanks for mentioning that, now I'm going to use it :-) |
| 11:36 | jcromartie | let's say I want to do x y and z |
| 11:36 | jcromartie | and I want to stop and return the first case where some pred returns a truthy value for any of those |
| 11:36 | jcromartie | I say "do" because these are side-effecty functions |
| 11:37 | jcromartie | they all return a map of either {:success …} or {:error ...} |
| 11:37 | jcromartie | condp? |
| 11:37 | jcromartie | no |
| 11:40 | dnolen_ | (some #(%) [x y z]) |
| 11:40 | jcromartie | ah ha, yes, and wrap up the side effects in fns |
| 11:40 | jcromartie | thanks |
| 11:41 | jcromartie | except I really want first filter |
| 11:41 | jcromartie | but yes |
| 11:42 | jcromartie | I was *about* to write a macro :) |
| 11:42 | jcromartie | like, filter-forms |
| 11:45 | cemerick | hyPiRion: are you using double-check post-test.check-merge, or an older SNAPSHOT? |
| 11:48 | jcromartie | dnolen_: how stupid is this? https://gist.github.com/anonymous/9644556 |
| 11:50 | dnolen_ | jcromartie: I don't really have an opinion, looks fine if that's what you want to do |
| 11:52 | dnolen_ | jcromartie: though if I think about it not sure why you can't just use `or` |
| 11:52 | jcromartie | good call |
| 11:59 | dsay | using shoreleave http-rpc for remote call. response is returned. |
| 11:59 | dsay | perform reset! in callback function on a defined atom but when use atom it doesn’t have new value. what need to do to set var with remote response? |
| 12:01 | clgv | I have a seldom event S and very frequent events F that operate on a reference. it seems that the transaction of S never succeed while F events are running. after I stop the F events another event S succeeds. how can I fix that? |
| 12:02 | augustl | so, I extracted jdk8 to a folder. How can I make "lein" use it? Tried setting JAVA_HOME to $HOME/local/jdk1.8.0 to no avail. |
| 12:03 | clgv | augustl: linux/unix? |
| 12:03 | augustl | just found LEIN_JAVA_CMD :) |
| 12:03 | augustl | clgv: yeah |
| 12:03 | clgv | augustl: google for jdk and "update-alternatives" |
| 12:03 | clgv | augustl: if you want to use it for more then just one try |
| 12:05 | shep-werk | clgv: As a Gentoo user, I feel I should be indignant :-) |
| 12:05 | rasmusto | gentoo -O3 erryday |
| 12:05 | augustl | clgv: I don't want to change my system java :) |
| 12:05 | augustl | LEIN_JAVA_CMD=$HOME/local/jdk1.8.0/bin/java seems to have worked |
| 12:05 | shep-werk | rasmusto: oh, are we going to get in a CFLAGS war? Oh yeah! |
| 12:06 | jjl` | anyone got a good example of as-> usage? it's very hard to google for |
| 12:06 | rasmusto | shep-werk: those days are over for me, I'm in arch now |
| 12:07 | clgv | shep-werk: stop using gentoo to save our planet ;) :P |
| 12:07 | shep-werk | rasmusto: well, we are still above the Ubuntu rabble, right? /s |
| 12:07 | shep-werk | clgv: Why would I want to turn off my space heater? It's cold here! |
| 12:08 | clgv | shep-werk: there are heaters with a better efficiency :P |
| 12:08 | shep-werk | my heater has the side effect of producing nice terminal text |
| 12:08 | rasmusto | ,(-> 3 inc dec (as-> b [b (- b)])) |
| 12:08 | clojurebot | [3 -3] |
| 12:09 | rasmusto | jjl`: does that clarify it? |
| 12:09 | jjl` | rasmusto: not really |
| 12:09 | rasmusto | (doc as->) |
| 12:09 | clojurebot | "([expr name & forms]); Binds name to expr, evaluates the first form in the lexical context of that binding, then binds name to that result, repeating for each successive form, returning the result of the last form." |
| 12:10 | rasmusto | "expr" is what gets passed in with the threading macro |
| 12:10 | hyPiRion | cemerick: I'm using 5.6-SNAPSHOT. (5.7-SNAPSHOT weren't out when I tried to compile) |
| 12:10 | clgv | jjl`: as-> is most useful when you have different postions where to insert the result of the previous calculation |
| 12:10 | rasmusto | name is a name you pick, and forms are expressions that use that named var: [b (- b)] for example |
| 12:10 | jcromartie | hmm… don't rely on lazy operations, e.g. map, for fine-grained control over side effects |
| 12:11 | jjl` | clgv: *nod* i've just done a macroexpand in repl to try and make sense of it. but i'm not completely sure i get it |
| 12:12 | rasmusto | ,(macroexpand `(-> 3 inc dec (as-> b [b (- b)]))) |
| 12:12 | clojurebot | (let* [sandbox/b (clojure.core/dec (clojure.core/inc 3)) sandbox/b [sandbox/b (clojure.core/- sandbox/b)]] sandbox/b) |
| 12:13 | clgv | ,(as-> 10 x (range x) (reduce + x) (/ x 10)) |
| 12:13 | clojurebot | 9/2 |
| 12:13 | jjl` | i think i get it |
| 12:13 | hyPiRion | (-> a (b c) (d e f) g (h)) is the same as (h (g (d (b a c) e f))) |
| 12:14 | cbp | jjl`: http://blog.fogus.me/2009/09/04/understanding-the-clojure-macro/ |
| 12:15 | jjl` | i don't have a problem with -> (which i find very useful) |
| 12:15 | cbp | ah sorry then |
| 12:15 | jjl` | it's just as-> that i'm struggling with. but i think i've got a grasp |
| 12:16 | cemerick | hyPiRion: OK. If you have any troubles with it (aside from the namespace renaming :-/), let me know. Probably going to cut a 0.5.7 release this week. |
| 12:16 | cemerick | (first "real" release) |
| 12:17 | rasmusto | jjl`: hyPiRion's example shows what as-> can do by itself, mine was more showing that -> and as-> can work together if you need a named binding partway through the threading macro |
| 12:17 | rasmusto | jjl`: er s/hyPiRion/clgv |
| 12:17 | hyPiRion | cemerick: will do – no worries |
| 12:18 | jjl` | rasmusto: yeah, i think your example solves a problem ive had down the line, where one of the things that needed threading needed threading to another point in the sexp |
| 12:18 | jjl` | naturally i rewrote it as a let and suffered the ugliness |
| 12:18 | cbp | jjl`: the point of as-> is to remove nesting -> and ->>. So you can define where the placeholder is instead of depending on first or last position |
| 12:19 | jjl` | *nod* |
| 12:21 | honza | does anyone have a post with how to set up piggieback + vim-fireplace with the clojurescript repl? |
| 12:30 | jcromartie | is there a better way to write this side-effecty loop thing? https://gist.github.com/anonymous/a51f8d487fbdeca53cbe |
| 12:31 | bbloom | jcromartie: doseq? |
| 12:31 | jcromartie | doseq can't short circuit |
| 12:31 | jcromartie | this is pure bikeshedding :) |
| 12:31 | bbloom | ah, how about reduced ? |
| 12:31 | jcromartie | slash golf |
| 12:31 | rasmusto | :while isn't short-circuity? |
| 12:31 | bbloom | ,(doc doseq) |
| 12:31 | clojurebot | "([seq-exprs & body]); Repeatedly executes body (presumably for side-effects) with bindings and filtering as provided by \"for\". Does not retain the head of the sequence. Returns nil." |
| 12:32 | bbloom | (doc for) |
| 12:32 | clojurebot | "([seq-exprs body-expr]); List comprehension. Takes a vector of one or more binding-form/collection-expr pairs, each followed by zero or more modifiers, and yields a lazy sequence of evaluations of expr. Collections are iterated in a nested fashion, rightmost fastest, and nested coll-exprs can refer to bindings created in prior binding-forms. Supported modifiers are: :let [binding-form expr ...], :while test, :when te |
| 12:32 | jcromartie | hm |
| 12:32 | jcromartie | ok |
| 12:32 | bbloom | yeah, i think let+while will do what you want |
| 12:32 | bbloom | rasmusto: thanks, i always forget doseq does taht |
| 12:33 | jcromartie | no, doseq returns nil |
| 12:33 | jcromartie | I want to return the first value for which :error is truthy |
| 12:33 | bbloom | jcromartie: use reduced |
| 12:33 | jcromartie | without evaluating any more |
| 12:33 | jcromartie | chunked seqs are right out |
| 12:33 | jcromartie | hm |
| 12:34 | bbloom | (reduce (fn [_ task] (let [result (task)] (when (:error result) (reduced result))) nil tasks) |
| 12:34 | jcromartie | woah, an escape from reduce? |
| 12:34 | bbloom | (doc reduced) |
| 12:34 | clojurebot | "([x]); Wraps x in a way such that a reduce will terminate with the value x" |
| 12:34 | jcromartie | when did that become a thing? |
| 12:34 | bbloom | (doc reduced?) |
| 12:34 | clojurebot | "([x]); Returns true if x is the result of a call to reduced" |
| 12:34 | bbloom | ,(meta #'reduced) |
| 12:34 | clojurebot | {:ns #<Namespace clojure.core>, :name reduced, :arglists ([x]), :column 1, :added "1.5", ...} |
| 12:34 | bbloom | 1.5 :-) |
| 12:34 | Raynes | jcromartie: Clojure has a changelog. You should poke around in there every once in a while :P |
| 12:34 | Raynes | It's exciting. |
| 12:35 | clgv | after the releases is sufficient ;) |
| 12:36 | jcromartie | Raynes: yes, thanks |
| 12:36 | jcromartie | :| I've just been cruising since 1.2 |
| 12:36 | cbp | jcromartie: you can use some |
| 12:37 | jcromartie | cbp: are you sure some won't chunk? |
| 12:37 | jcromartie | yeah, some *looks* safe |
| 12:37 | jcromartie | but it's not explicit |
| 12:38 | jcromartie | some appears to be the winner :) |
| 12:38 | jcromartie | (some #(let [result (%)] (and (:error result) result)) tasks) |
| 12:38 | cbp | jcromartie: it uses first & next, no chunked-cons so i dont see why it should |
| 12:39 | Raynes | And you can easily find out if it will |
| 12:39 | Raynes | Throw some side effects at it. |
| 12:39 | cbp | ,(some #(do (print %) (when (:error %) %)) [1 2 3 {:error 4} 5]) |
| 12:39 | clojurebot | 123{:error 4}{:error 4} |
| 12:40 | jcromartie | but that's not an explicit guarantee, just an implementation detail |
| 12:40 | augustl | is there a way to set java-cmd in a leiningen project without hardcoding it into the project.clj? |
| 12:41 | cbp | some is supposed to short circuit so i dont see any reason for it to do any more work than it should by chunking |
| 12:41 | clgv | augustl: the question contradicts itself |
| 12:41 | technomancy | augustl: might be able to set it in .leinrc |
| 12:41 | augustl | technomancy: trying to make leiningen use java 8 extracted to a folder, so I don't have to install it as my system java to play around with it :) |
| 12:42 | technomancy | augustl: why not just put it on your $PATH? |
| 12:42 | augustl | technomancy: that's all leiningen needs? |
| 12:42 | Raynes | jcromartie: If you're concerned about that, you're probably going to need to get Rich to put it in writing that it'll never change, or copy the source of some so that you know it'll never change. :P |
| 12:42 | Pate_ | any Clojurists from South Africa? |
| 12:42 | technomancy | augustl: sure |
| 12:43 | augustl | setting PATH worked fine, great :) |
| 12:43 | rasmusto | famous last words |
| 12:46 | augustl | booyah https://gist.github.com/augustl/9645871 |
| 12:46 | augustl | the beauty of JS, easily accessible right from Clojure ;) |
| 12:49 | technomancy | haha |
| 12:49 | technomancy | nice |
| 12:49 | sdegutis | Yeah, I was thinking of how they should have named the Java utility to run scripts "java-script". |
| 12:50 | augustl | sdegutis: haha, a missed opportunity |
| 12:50 | sdegutis | They named it "jrunscript" :/ |
| 12:50 | austinh | How can I test that a value is a MapEntry in Clojurescript? |
| 12:50 | augustl | there's a bin named jjs though :) |
| 12:51 | augustl | austinh: isa? |
| 12:51 | jjl` | 'beauty of js'. sorry, got me giggling |
| 12:52 | jjl` | i'm amused by the idea of running clojurescript on the jvm though |
| 12:52 | augustl | will attempt to make it easy to create single page web apps with fb react and render initial page from the server, without much extra work |
| 12:52 | augustl | and without the need to run a separate node process for that |
| 12:53 | augustl | jjl`: full circle :) What's the use case? |
| 12:53 | tbaldridge | It'll be interesting to see how that whole thing works out in relation to Graal and Truffle. Last I heard the JS Truffle VM is quite a bit faster than nashorn |
| 12:53 | jjl` | augustl: oh i don't have a good use case, i'm just amused :) |
| 12:54 | austinh | augustl: Thank you. |
| 12:54 | jjl` | augustl: i'm daft enough to have tried porting perl to the jvm so amusement from craziness comes easily |
| 12:54 | augustl | FUD time.. I remember reading somewhere that v8 is optimized for crashing. After all, if a chrome tab crashes, just reload, no big deal. So comparing that to nashorm might be interesting. |
| 12:55 | augustl | jjl`: perlj.am would be the domain name for that project |
| 12:55 | jjl` | php is certainly optimised for short runtimes (memory leaks etc.) which is actually a valid design decision if you know youre scripts are supposed to execute in a second or so |
| 12:56 | jjl` | i'm pretty sure nashorn is optimised for long runtimes. that's the only way they'll get the full benefit of hotspot |
| 12:56 | augustl | yeah, seens like nashorn is oracle saying "we're good at doing dynamic languages now" |
| 12:56 | jjl` | actually i think it's their way of saying "yeah, okay, rhino is a bit crap" |
| 12:56 | technomancy | augustl: nashorn is "see, invokedynamic isn't just for jruby!" |
| 12:57 | augustl | :D |
| 12:57 | augustl | jjl`: isn't rhino a mozilla project though? |
| 12:57 | jjl` | because jython uses it as well? :) |
| 12:57 | jjl` | yes, rhino is mozilla. but it also means there was no great jvm js implementation |
| 12:57 | jjl` | i think they're just hopping on the bandwagon more than anything |
| 12:58 | jjl` | "oh yeah, javascript is cool. we've always thought it. look here! we wrote one!" |
| 12:58 | jjl` | i think it'll get remarkably interesting once clojurescript is self-hosting |
| 12:59 | augustl | there's also a node.js implementation for nashorn btw |
| 12:59 | augustl | which uses libuv, the cool part of node |
| 12:59 | jjl` | the other interesting thing is vert.x |
| 13:00 | jjl` | http://vertx.io/ |
| 13:00 | tbaldridge | And then Truffle just goes off the deep and and says "let's do a tracing JIT on the JVM" |
| 13:00 | tbaldridge | *deep end |
| 13:00 | augustl | https://avatar-js.java.net/ |
| 13:00 | tbaldridge | http://lafo.ssw.uni-linz.ac.at/papers/2014_CGO_OneVMToRuleThemAll.pdf |
| 13:00 | augustl | vertx is a completely separate implementation of the event loop |
| 13:00 | augustl | avatar is just a wrapper around the very same library that node uses internally |
| 13:01 | augustl | not sure which is a better idea.. Would make sense to do a node.js implementation on top of vert.x too |
| 13:01 | jjl` | that would certainly help get people ported over |
| 13:04 | dnolen_ | jjl`: nashorn is not optimized for long runtimes, I've been following along since very relevant for Clojure(Script) narrative. They seems very interested start up fast, optimize fast - like the other JS engines. |
| 13:04 | jjl` | interesting |
| 13:05 | dnolen_ | tbaldridge: pretty sure that isn't true w/ respect to Truffle / Nashorn |
| 13:05 | dnolen_ | tbaldridge: invokeDynamic rewritten for Nashorn far as I know |
| 13:05 | jjl` | i think the day is fast approaching that i'll have to give up programming and become a farmer. when every tech stack has javascript |
| 13:06 | dnolen_ | tbaldridge: and Nashorn is now as fast as SpiderMonkey and w/in shooting distance of JSC/V8 |
| 13:06 | jjl` | it can't be long until someone writes a build system in javascript designed for building java, that takes the form of a DSL |
| 13:06 | hiptobecubic | It's really a shame that javascript is such an abomination |
| 13:07 | austinh | augustl: Actually, are you sure about using isa? to test MapEntry in cljs? I can't get it to work. |
| 13:07 | augustl | austinh: hmm, not sure what a MapEntry is |
| 13:09 | locks | hiptobecubic: ur just jelly |
| 13:15 | dnolen_ | austinh: there is no concrete MapEntry type in CLJS, just the IMapEntry protocol |
| 13:16 | dnolen_ | austinh: when seq'ing maps you just a get a seq of vectors, and vectors satisfy IMapEntry |
| 13:16 | dnolen_ | can test IMapEntry with (satisfies? IMapEntry foo) |
| 13:16 | austinh | dnolen_: Thanks! |
| 13:16 | tbaldridge | dnolen_: that could be, the talk I was thinking of was both biased (presented by a lead Truffle researcher) and older (~1yo) |
| 13:17 | dnolen_ | tbaldridge: yeah Nashorn was a dog a year ago |
| 13:25 | seangrove | Nashorn definitely seems like ti could open up some very cool possibilities |
| 13:55 | jstew | Anyone know if I can run code after a ring response completes? (using compojure). |
| 13:57 | jstew | I generate some temp files, and want to clean them up. |
| 13:57 | technomancy | jstew: middleware can do that |
| 13:59 | jstew | technomancy: If I remove the file in middleware, won't it be gone by the time I try to serve it as the response? |
| 13:59 | jjl` | you misunderstand middleware |
| 13:59 | jjl` | your function receives the next function in the chain and you call it |
| 13:59 | jstew | indeed I may :) |
| 13:59 | jjl` | so you can do things both before and after |
| 13:59 | amalloy | jstew: are you trying to serve the temp files themselves as the response? |
| 13:59 | jstew | amalloy: Correct. |
| 14:00 | jjl` | oh, like sendfile() or something? |
| 14:00 | jjl` | which plugin are you using to send it? |
| 14:00 | amalloy | so he probably can't (straightforwardly) clean them up in a middleware - ring needs to be able to see the files in order to serve them, after his handler/middleware stack is done |
| 14:00 | jjl` | because it could just be a case of ordering your middleware if it's one that will just slurp the file and pass the content back |
| 14:01 | amalloy | jjl`: he doesn't need a plugin, really - ring can accept java.io.File as the :body |
| 14:01 | amalloy | and slurping it isn't much good if it's too big to fit in ram |
| 14:01 | jstew | That's what I am doing. Serving the file directly and setting content-Disposition in the headers |
| 14:02 | jjl` | amalloy: in that case, it's a bigger problem than it first appears :) |
| 14:02 | amalloy | jstew: in most (all?) modern operating systems, you can delete a file while you have it open, and handles to it will stay valid |
| 14:02 | amalloy | so you can open an InputStream on the file, then delete it, then pipe that stream to ring |
| 14:02 | jstew | Ahh... Yes, that will work. Thank you! Forgot about that. |
| 14:02 | amalloy | (ring is happy to accept streams) |
| 14:02 | jjl` | this only works of course if you haven't taken a flock() |
| 14:03 | amalloy | java probably doesn't even let you |
| 14:03 | jjl` | hah |
| 14:03 | amalloy | right, it doesn't. since not every OS in the world supports that |
| 14:04 | jjl` | this is, incidentally, why perl's core file handling stuff is awful. it *does* support VMS |
| 14:05 | `szx | (into [] lazyseq) vs (vec lazyseq)? surprisingly, google didn't have the answer |
| 14:06 | amalloy | `szx: amazingly, there's no easy answer. vec seems like it *should* be right, and i'd use it because it's most readable, but i think into/[] is actually a bit faster |
| 14:07 | `szx | amalloy: alright, so vec it is till i run into performance issues |
| 14:07 | amalloy | `szx: and when you do run into performance issues, there will be better places to fix them :) |
| 14:08 | jcromartie | into is definitely faster |
| 14:09 | jcromartie | IMHO into is a bit of an underrated gem of the collections interface |
| 14:12 | cbp | into is faster because it uses transients |
| 14:13 | stuartsierra | austinh, augustl: remember `isa?` compares 2 classes. `instance?` asks if an object is of a particular class. |
| 14:15 | amalloy | jcromartie, cbp: into is *usually* faster. it is definitely not the case that it's always faster |
| 14:16 | austinh | stuartsierra: Thanks! |
| 14:16 | cbp | amalloy: I imagine into gets faster the more items there are |
| 14:17 | amalloy | for example, consider (let [a (into-array (range 30))] (vec v)) |
| 14:17 | amalloy | vec has optimizations for array inputs |
| 14:17 | amalloy | into has to seq over the array |
| 14:17 | cbp | ah |
| 14:18 | amalloy | (in the example above is about 10 times faster than into) |
| 14:19 | amalloy | in practice, into will "always" be faster, but it just doesn't really matter |
| 14:19 | amalloy | i suggest using vec, because it's more specialized, and eventually someone will get around to implementing the same transient optimizations for it that work with into, as well as the other vector-specific ones it already has |
| 14:21 | technomancy | yeah, imo into makes most sense when you're pouring into something that's not empty |
| 14:22 | technomancy | you use it to pour into {} too, but only because there's nothing like vec for maps for whatever reason |
| 14:23 | Bronsa | technomancy: my guess is simply because `map` was already taken |
| 14:23 | cbp | ran out of names :-D |
| 14:25 | rasmusto | ,array-map |
| 14:25 | clojurebot | #<core$array_map clojure.core$array_map@1e9bfd3> |
| 14:26 | technomancy | yeah, but you have to apply it |
| 14:26 | rasmusto | oh yeah, how is array vs hash map picked? |
| 14:27 | jjl` | apply is my usual frustration with -> |
| 14:27 | technomancy | jjl`: (partial apply f) =) |
| 14:27 | technomancy | don't actually do that |
| 14:27 | jjl` | quite |
| 14:28 | jjl` | come to think of it, i'm not sure why i haven't written a wrapper that switches the arguments |
| 14:28 | cbp | rasmusto: I believe the cutoff is around 8 items |
| 14:28 | jjl` | that would solve it |
| 14:29 | rasmusto | cbp: okay, so nothing to do with the types of keys used or anything |
| 14:29 | cbp | rasmusto: no I don't think so |
| 14:33 | Averell | Is recur a kludge or motivated by no TCO on the JVM? Or am i paranoid? |
| 14:33 | bbloom | ~recur |
| 14:33 | clojurebot | Cool story bro. |
| 14:33 | bbloom | ~tco |
| 14:33 | clojurebot | yes please |
| 14:33 | bbloom | lol |
| 14:33 | bbloom | Averell: this has been discussed many times here |
| 14:33 | TimMc | Averell: Paranoid? Who's out to get you? |
| 14:33 | Averell | any faq? |
| 14:34 | Averell | by natural suspicion |
| 14:34 | Averell | my* |
| 14:34 | bbloom | Averell: in short, TCO is useful for many problems, but far from critical b/c you can always use a trampoline |
| 14:34 | bbloom | Averell: recur is tail RECURSION optimization, which is different than general TCO |
| 14:35 | tbaldridge | Averell: TCO isn't supported on the JVM....so there that, but also recur covers almost everything so I don't see much of a problem |
| 14:35 | Averell | yes, but lets say there was, would recur still exist? |
| 14:35 | amalloy | technomancy: at one point i did (def ap (partial partial apply)), to make it easier to squeeze apply forms into -> chains |
| 14:36 | bbloom | Averell: if the JVM supported proper tail calls, it's likely that clojure would have still needed something like recur, since clojure was simultaneously developed for the CLR, which also doesn't have general TCO |
| 14:36 | justin_smith | it's nice to have something that throws an error if not in the tail position |
| 14:36 | tbaldridge | probably for backwards compatibility. But also I think recur is really nice since it fails to work if its used in a non tail call position |
| 14:36 | amalloy | i don't remember if i ever actually used it, though - it looks knida awkward |
| 14:36 | bbloom | Averell: nor does javascript, for the clojurescript backend |
| 14:36 | tbaldridge | bbloom: sure it does, just not on x86-64....or x86....one of those, not sure |
| 14:36 | bbloom | Averell: so yes, host constraints led to the addition of the "recur" form, but it's still useful in much the same was as static type checking or runtime assertions are useful |
| 14:37 | technomancy | Averell: also: most uses of tco in scheme are better served by laziness in clojure |
| 14:37 | bbloom | tbaldridge: the CLR *can* do TCO, but doesn't promise it |
| 14:37 | amalloy | i imagine if we did have general tail calls we'd have something like (tail f x args), meaning "act like (f x args), but throw an exception if you can't do this without using stack space" |
| 14:38 | amalloy | or rather, don't compile if you can't do this |
| 14:39 | bbloom | amalloy: and that's practically exactly what i do in my interpreter: https://github.com/brandonbloom/eclj/blob/03bbe44dfb355930ff341b742b3b74c78325ca5d/src/eclj/core.clj#L583-L586 |
| 14:39 | augustl | anyone here happen to know how you provide your own classes to a jdk8 nashorn runtime? :) |
| 14:39 | bbloom | amalloy: i implement normal recursion with the y-combinator, but i implement recur as an effect that enforces tail position on the continuation |
| 14:40 | cbp | hmm, if I have some manner of (zipmap (keys m) (vals m)) I would have to rewrite that for clojure 1.6? |
| 14:41 | jcromartie | cbp: why? aren't keys and vals consistent? |
| 14:41 | bbloom | tbaldridge: clarifying: the CLR supports explicit tail calls, but the JIT may or may not perform tail call OPTIMIZATION for non-tail calls. the JVM supports neither explicit nor optimized tail calls |
| 14:41 | jcromartie | http://dev.clojure.org/jira/browse/CLJ-1302 |
| 14:41 | jcromartie | this is in 1.6 |
| 14:42 | bbloom | tbaldridge: or i guess not for "non-tail calls", i mean for implicit tail calls. ie normal calls in tail position |
| 14:42 | bbloom | big difference as far as debuggers are concerned... |
| 14:42 | Averell | why is that apparently so much harder than detecting tail position? |
| 14:42 | rasmusto | (doc keys) |
| 14:43 | clojurebot | "([map]); Returns a sequence of the map's keys." |
| 14:43 | cbp | jcromartie: there were some hashing changes and in that ticket Stuart Sierra says you should not rely on it |
| 14:43 | bbloom | Averell: reality is more complex than the lambda calculus? :-) |
| 14:43 | Averell | i suppose :) |
| 14:43 | bbloom | Averell: exception handling, especially finally blocks, for example |
| 14:43 | jcromartie | "The absence of this property in the docs is correct. You should not rely on this." |
| 14:43 | jcromartie | (other stuart) |
| 14:44 | cbp | er |
| 14:44 | cbp | yes other stuart |
| 14:44 | cbp | :-P |
| 14:44 | bbloom | Averell: the comp sci community genuinely didn't understand the nature of exception handling until the last few years |
| 14:44 | jcromartie | but rich hickey says "keys order == vals order == seq order" |
| 14:44 | tbaldridge | jcromartie: cbp: "keys order == vals order == seq order" |
| 14:44 | jcromartie | i.e. /thread |
| 14:44 | cbp | Oh but at the end Rich says that yeah |
| 14:44 | amalloy | yeah, i'm amazed to hear that from stuart |
| 14:44 | cbp | *joy* |
| 14:44 | jcromartie | WHAT AN IDIOT, GOD, PFFFFT |
| 14:45 | amalloy | bbloom: does the CS community understand it now? |
| 14:45 | stuartsierra | I remember this discussion now. Stuart Holloway was incorrect. The order of keys and vals is the same. |
| 14:46 | bbloom | amalloy: much better than they did, that's for sure |
| 14:48 | bbloom | Averell: question answered? happy to provide further references |
| 14:52 | dnolen_ | Averell: the reason Clojure and ClojureScript don't offer TCO is very simple - it makes interop problematic. |
| 14:53 | Averell | no, i'm good. thanks :) |
| 14:53 | Averell | if you have any links to better understand exception handling, that sounded interesting. |
| 14:55 | bbloom | Averell: i gave a talk on algebraic effects on monday night, it should be on youtube later this week. follow @papers_we_love on twitter. some background materials are linked to from here: https://github.com/papers-we-love/papers-we-love/issues/9 |
| 14:55 | bbloom | Averell: in short, exceptions are now best understood in terms of delimited continuations |
| 14:56 | bore | Hi guys. Just a quick one for a clojure newbie. How do I replace the last character of the String most efficiently? |
| 14:57 | bore | do I need to dip into java? |
| 14:57 | technomancy | can't you just use lasts? |
| 14:57 | technomancy | last |
| 14:58 | maxthoursie | (apply str (butlast some-string)) |
| 14:58 | maxthoursie | wouldn't say it's efficient |
| 14:58 | technomancy | oh, misread |
| 14:58 | cbp | Is there an alternative to compiling protocol buffers from clojure than lein-protobuf? |
| 14:58 | cbp | Maybe I should do it manually.. |
| 14:59 | bbloom | (doc re-replace) |
| 14:59 | clojurebot | Cool story bro. |
| 14:59 | maxthoursie | bore: (clojure.string/replace "foo" #"o$" "b") |
| 14:59 | bbloom | ah, d'oh not re-replace |
| 14:59 | bbloom | bore: yeah what maxthoursie said |
| 15:00 | jcromartie | cbp: I'd think your protocol buffers should be a separate project |
| 15:01 | cbp | jcromartie: That's a good idea. But separate from the need to compile them |
| 15:02 | bore | well what maxthoursie said sounds a bit inefficient and I'll need to call str twice. Once with apply to get string back and then once to add the replacement character |
| 15:02 | bore | otherwise makes sense |
| 15:03 | maxthoursie | bore: the second one |
| 15:03 | justin_smith | matching on #".$" should be fast, strings in the jvm are counted |
| 15:03 | maxthoursie | bore: use clojure.string/replace |
| 15:03 | bore | So far I have (str (.substring foo 0 (dec (count foo))) bar) |
| 15:04 | bore | aha OK that sounds more like it |
| 15:04 | bore | thanks a lot |
| 15:05 | maxthoursie | bore: if you want it faster i guess you should store it in a mutable way so you can just replace the character directly |
| 15:05 | maxthoursie | bore: but if you can aviod that -> cleaner code |
| 15:05 | maxthoursie | strings are the only thing that java got right (wrt immutability) |
| 15:06 | bore | no I think copying into byte buffer would make things worse - I do not intend to reuse the base string much |
| 15:06 | bore | but thanks for the idea |
| 15:06 | bore | I'll go with regex solution |
| 15:06 | bore | it strikes a good balance |
| 15:07 | maxthoursie | fails on the 'aviod regexes when you can' guildline though :) |
| 15:09 | justin_smith | I guess you could use a StringBuilder or something if you wanted to super optimize it |
| 15:10 | justin_smith | https://www.refheap.com/61951 .substring is a little faster than the re solution |
| 15:12 | justin_smith | https://www.refheap.com/61951 updated to show the def of input |
| 15:13 | Averell | is count walking the full string? |
| 15:13 | justin_smith | no, strings are counted |
| 15:13 | justin_smith | they store their size |
| 15:14 | Averell | oh that's what that meant. duh. |
| 15:14 | gtrak | count uses seq functions, pretty sure |
| 15:15 | gtrak | so it doesn't matter if strings are counted |
| 15:15 | justin_smith | oh, so you want .length |
| 15:15 | gtrak | eat the cost of the reflection warning :-) |
| 15:16 | justin_smith | the book "clojure high performance programming" claims count is O(1) on counted collections |
| 15:16 | gtrak | aha, actually I see a special case for CharSequence |
| 15:16 | pyrtsa | gtrak: count delegates to RT/count on the JVM: https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/RT.java#L528-L558 |
| 15:16 | jcromartie | clojure.core/count is optimized for CharSequence |
| 15:16 | jcromartie | oh yeah |
| 15:16 | gtrak | so count should be fine |
| 15:16 | jcromartie | :) |
| 15:17 | jcromartie | how many people were just on that file |
| 15:17 | gtrak | it's almost as if they thought of that |
| 15:17 | justin_smith | gtrak: according to this book, it is O(1) on list, hashmap, arraymap, vector, queue, treemap |
| 15:17 | justin_smith | and if I am reading that source right, it agrees |
| 15:18 | gtrak | err.. I think it could be wrong for list |
| 15:18 | clojurebot | excusez-moi |
| 15:18 | gtrak | I remember reading it delegating to the rest, if that's a cons then you've got problems |
| 15:18 | gtrak | but in practice it's probably negligible |
| 15:18 | Averell | hehe. so at least it works for very short lists. |
| 15:19 | pyrtsa | count is O(n) for lists. Unless there's a tail that implements Counted. Then it's O(n_head). |
| 15:20 | gtrak | the class of the head is not a testament to the rest of it |
| 15:20 | pyrtsa | :) |
| 15:21 | pyrtsa | gtrak: But indeed it is when the head is Counted. ;) |
| 15:21 | gtrak | err, hmm, actually looks like persistentlist manages the count for every case. |
| 15:21 | gtrak | as long as you start there when you cons you should be fine |
| 15:22 | augustl | isn't (defn ^:static -myMethod [] ...) the way to create static methods for a :gen-class ns? |
| 15:22 | pyrtsa | Oh, PersistentList does implement Counted. But lazy seqs don't. |
| 15:22 | gtrak | conj, rather.. grr |
| 15:23 | gtrak | if you call cons, you break it, if you conj onto a list, you're fine |
| 15:23 | pyrtsa | Funnily, the result of (range n) does not implement Counted. |
| 15:24 | justin_smith | I guess because lazy things in general do not |
| 15:24 | justin_smith | it seems like it would be so trivial though! |
| 15:24 | pyrtsa | Yeah, but it could be implemented with an own class. |
| 15:24 | justin_smith | also, (nth (range) n) is O(n) though one could naively imagine it being O(1) |
| 15:26 | Averell | but nth on a counted thing would be O(1)? |
| 15:27 | pyrtsa | nth for Indexed will be O(1) or O(log n) or something like that, I guess. |
| 15:27 | Bronsa | Averell: lists are counted, nth is still O(n) |
| 15:29 | pyrtsa | Oh, I wasn't quite accurate with that. I meant that Counted doesn't affect nth. Again, the impl is found in RT.java. |
| 15:30 | Averell | static public. it's distracting :) |
| 15:42 | Averell | Man browsing that is kinda cool. although the language suddenly loses some mystique, just seeing it there, how simple stuff is. |
| 15:56 | hyPiRion | Most of it is simple |
| 15:57 | hyPiRion | But when you attempt to answer the question "is this symbol/keyword legal?" by reading the Reader, you're gonna have a bad time |
| 15:59 | Averell | hehe. i already have a bad time moving out of simple collections, but still much more approachable than i thought. |
| 16:00 | akurilin | Does anybody have a solution to this: I have 2 local libs my project depends on, I'm using "checkouts" folder in development. When building the uberjar though I need to be sure that the git repositories are up to date and that I ran lein install or I might be shipping the uberjar with outdated libs |
| 16:00 | gtrak | i don't think checkouts has much to do with artifacts |
| 16:01 | gtrak | unless you're relying on bad AOT behavior |
| 16:01 | akurilin | Checkouts lets you not notice that the local .jar repo is not up to date, right? Everything will work just fine until you build and ship |
| 16:01 | gtrak | checkouts will add the subdirs to the classpath is my understanding |
| 16:02 | akurilin | That's correct. |
| 16:02 | akurilin | lein uberjar uses the repo though. |
| 16:02 | gtrak | right. |
| 16:03 | gtrak | you probably want to mvn install the dependency jars, it's worth the trouble. |
| 16:03 | akurilin | What does that mean? |
| 16:04 | gtrak | 'lein install', just like you said |
| 16:04 | gtrak | you need actual releases and maven artifacts for the deps |
| 16:05 | gtrak | if you don't want to go through the trouble, ie you are the author of those checkouts, why not just add the source to your project's source-paths? |
| 16:05 | tbaldridge | pyrtsa: in ClojureScript I think it does |
| 16:05 | gtrak | maybe I'm misunderstanding what you mean by 'local libs'... |
| 16:06 | tbaldridge | pyrtsa: bleh, stale irc window |
| 16:06 | gtrak | is this clojure source in a clojure project that's not released anywhere? |
| 16:06 | akurilin | yes it is |
| 16:06 | akurilin | it's just shared across multiple projects |
| 16:06 | gtrak | ah, yea. well.. maybe a more convenient way to share it is :source-paths :-) |
| 16:07 | akurilin | Admittedly the lein install approach is a pita |
| 16:07 | gtrak | like this: https://github.com/technomancy/leiningen/blob/master/sample.project.clj#L244 |
| 16:07 | gtrak | throw "../my-lib/src" in there |
| 16:07 | gtrak | and don't worry about it |
| 16:08 | gtrak | unless you actually want to start releasing the shared code |
| 16:08 | akurilin | Let me try that, that sounds really useful :) |
| 16:08 | gtrak | then you can remove it from checkouts, too |
| 16:09 | jcromartie | I'm really finding a lot of places where "for" is much more readable than "map" |
| 16:09 | jcromartie | fun fact: (if [pred x] …) will allways be true :) |
| 16:10 | gtrak | lol yes |
| 16:10 | rasmusto | for + :when/:let/:while is a lot nicer than map + filter/take-while/??? |
| 16:10 | lazybot | rasmusto: Oh, absolutely. |
| 16:11 | gtrak | ~for + :when/:let/:while |
| 16:11 | clojurebot | Pardon? |
| 16:13 | rasmusto | nrepl.el question: why does it not switch modes when I C-c M-j? |
| 16:14 | rasmusto | oh, it was something in my config... hm |
| 16:15 | justin_smith | rasmusto: using emacs now? woah |
| 16:15 | brehaut | is there anything analgous to sql subselect in core.logic ? |
| 16:15 | rasmusto | justin_smith: trying it out, vim is giving me pains recently |
| 16:16 | rasmusto | justin_smith: more specifically, BLOCKING highlighting of matching parens causing many ms of lag |
| 16:16 | justin_smith | weird |
| 16:17 | zspencer | Is there a clojure idiomatic way of injecting a dependency? |
| 16:17 | rasmusto | Had to start dumping out profiling information, and yeah, thousands of hits to the highlight function |
| 16:17 | technomancy | akurilin: I'd recommend setting :target-path to "target/%s" to avoid that problem |
| 16:17 | technomancy | akurilin: your checkout deps should be stripped from the classpath when creating the uberjar, but class files compiled when they were on the classpath could still be included with the default :target-path setting |
| 16:18 | technomancy | akurilin: see the last Q on https://github.com/technomancy/leiningen/blob/master/doc/FAQ.md |
| 16:19 | seangrove | Ah, I miss vars sometimes in cljs |
| 16:20 | jjl` | you've paid for paradise, put up a parking lot? |
| 16:22 | TimMc | jjl`: I can't tell if you're punning, or if that's how you think the lyrics go. |
| 16:25 | gtrak | It took me 3 years to hit a problem where metadata was the right solution, but it made a hard thing pretty trivial. |
| 16:26 | gtrak | avoided changing a ton of code around. |
| 16:27 | akurilin | technomancy: so you'd not recommend the source-paths approach? |
| 16:27 | gtrak | had to validate a data structure, but the information that was needed to do it was no longer there, so it was either attach it via metadata, or pass around an extra argument everywhere or change the structure and everything that uses it. |
| 16:27 | akurilin | or is that still cool? |
| 16:27 | technomancy | akurilin: checkouts is just a wrapper around source-paths and a few other things to begin with |
| 16:28 | technomancy | akurilin: but yeah, changing source-paths won't actually fix the problem if you have classfiles bleeding over across profile boundaries |
| 16:34 | akurilin | technomancy: so to be clear, would I still want to run lein install there on my deps, or are they somehow going to be built for me during lein uberjar |
| 16:34 | akurilin | ? |
| 16:34 | gtrak | that looks like a different problem than the one you're trying to avoid |
| 16:34 | gtrak | that's how to avoid AOT in general except when you're making uberjar |
| 16:35 | technomancy | akurilin: checkout deps are off the classpath when uberjarring no matter what, so yeah. stuff is going to come out of ~/.m2. |
| 16:36 | akurilin | Ok got it, thanks |
| 16:36 | technomancy | right, so ensuring what goes into the jar is up-to-date is bigger than just this |
| 16:36 | gtrak | technomancy: I just suggested he avoid a project boundary for code-sharing if he's not committed to doing releases. |
| 16:37 | technomancy | but you can't get that right if you're leaking across profiles anyway |
| 16:37 | gtrak | and mess with sources paths instead |
| 16:37 | akurilin | So the idea is that if I use source-paths then I simply can't use profiles correctly? |
| 16:37 | gtrak | totally unrelated |
| 16:37 | technomancy | akurilin: sorry, to be clear checkouts are only enabled through profiles |
| 16:38 | technomancy | the base profile is what turns checkouts on |
| 16:38 | technomancy | but only the uberjar profile is active when cerating an uberjar |
| 16:38 | technomancy | so you're using profiles whether you're doing it explicitly or not |
| 16:40 | akurilin | So the point is that once you add a source-path directive to project.clj, you always have that path on regardless of profile, riht? |
| 16:41 | akurilin | it becomes an extension of the project folder |
| 16:41 | gtrak | profiles can do anything, but pretty much |
| 16:42 | gtrak | the uberjar/checkouts stuff is irrelevant, it's a source folder for your project. |
| 16:42 | gtrak | you could write a profile that kills it, but that wouldn't make much sense :-) |
| 16:42 | technomancy | oh, sure |
| 16:42 | technomancy | or you could add checkouts to the defproject directly |
| 16:42 | akurilin | Well the impression I'm getting is that checkouts is a "magical folder", as in it's ignored when building the uberjar |
| 16:43 | akurilin | as opposed to something you'd add with source-path |
| 16:43 | technomancy | :checkout-deps-shares [:source-paths] will make it active during uberjar if that's what you want |
| 16:44 | gtrak | checkouts implies you want something with artifacts, I thought. |
| 16:45 | technomancy | well if you're using it for something other than "Convenience during development" then you're beyond using it for what it's intended for =) |
| 16:47 | technomancy | anyway, everyone should use profile isolation regardless, and if you also do this :source-paths thing it might also solve different problems |
| 16:47 | gtrak | hacking checkouts to give custom behavior on libraries you don't control is probably bad. |
| 16:48 | gtrak | if you control the deps, then you don't need checkouts except for convenience, and maybe it's more convenient to do other things. |
| 16:48 | gtrak | to avoid having to install/deploy that stuff in the first place. |
| 16:49 | gtrak | we had a massive shared library that was shared across 3 products, and the artifact was AOT'd, which meant that they all had to use the same clojure version among other complexities. |
| 16:49 | gtrak | I think we could have avoided some problems by doing it this way. |
| 16:51 | technomancy | the simplest thing to do is just not create deployable artifacts on the same machine you do dev on |
| 16:51 | gtrak | the fact that clojure AOT tries to AOT the world means you'd get all the dependency classes in the artifact, so we filter those. |
| 16:51 | technomancy | poof; no more staleness problems |
| 16:51 | gtrak | all to distribute a proprietary shared jar without source in it. |
| 16:52 | jjl` | TimMc: it was never one of my favourites. they're probably wrong but i'm willing to accept the more flattering alternative |
| 16:53 | gtrak | technomancy: that doesn't solve this problem. |
| 16:53 | jjl` | TimMc: ohh, now i look them up it makes a lot more sense |
| 16:53 | gtrak | rather, the one we had. |
| 16:55 | technomancy | gtrak: ah yeah, well AOTing a library is never going to be pretty |
| 16:55 | gtrak | http://dev.clojure.org/jira/browse/CLJ-322 |
| 16:56 | technomancy | wow, this was once on the list for 1.6?? |
| 16:56 | lazybot | technomancy: Definitely not. |
| 17:08 | hugod | pprint doesn't seem to print metadata, even if *print-meta* is true |
| 17:11 | hugod | I wonder if fipp does… |
| 17:12 | bbloom | hugod: nope, but it should be pretty easy to add |
| 17:12 | bbloom | hugod: take a look at edn.clj |
| 17:13 | hugod | bbloom: ok, will have a look |
| 17:16 | TimMc | jjl`: :-) |
| 17:18 | mikerod | When processing unevaluated Clojure forms with a goal is to fully-qualify all Symbols that are non-local |
| 17:18 | mikerod | with a goal of fully-qualifying** |
| 17:18 | mikerod | is there any good strategy? :) |
| 17:19 | stuartsierra | tools.analyzer? |
| 17:19 | mikerod | Looks like it becomes quite a trick to detect locally Symbol bindings that the unevaluated code may produce. |
| 17:19 | mikerod | stuartsierra: I haven't looked much at it before. I suppose I should look a bit. |
| 17:21 | stuartsierra | Within a macro, you can get the current local symbols from &env I think. You can look up global symbols with `resolve`. |
| 17:21 | mikerod | If you just blindly qualify any Symbol that can be qualified in the current *ns*, I think the only issue is it would ignore local shadowing of those Symbols. |
| 17:22 | mikerod | Yeah, &env is an option I explored. I couldn't come up with a reasonable way to use it though when I haven't actually evaluated the binding forms. So there really is no "env" at the time, unless I reproduce it. |
| 17:23 | mikerod | For example, if I have some code like '(let [x 5] (for [y [1 2]] x y z); I want to see '(let [x 5] (for [y [1 2]] x y ns.name/z) returned |
| 17:24 | mikerod | What the code does doesn't matter, it is assumed to be useful later in another context. |
| 17:24 | mikerod | I just want to store it off with qualified Symbols. It is simple I think, if there are no locals shadowing global Symbols. |
| 17:33 | mikerod | I believe that locals only come from let* and letfn* when it comes down to the special forms right? |
| 17:34 | mikerod | Nvm I think I answered myself. I'll keep digging. :P |
| 17:35 | justin_smith | mikerod: don't forget fn / (defn) bindings |
| 17:37 | mikerod | justin_smith: yep, just thought about them. then I thought, maybe I would have to consider recur and catch as well. I don't know for sure, but they stood out as possibilities. |
| 17:38 | justin_smith | loop (recur would bind for either fn or loop) |
| 17:39 | mikerod | hmm yeah |
| 17:40 | rasmusto | vabC-cC-r for evaling a region? or is there a better way to eval an inner expression in emacs |
| 17:40 | rasmusto | (this is an evil-mode hack btw) |
| 17:44 | dsrx | rasmusto: if the point is at the end of a sexp C-x C-e will evaluate just that sexp in cider |
| 17:44 | rasmusto | (blah (foo bar)|) + C-X C-e ? |
| 17:44 | dsrx | yeah |
| 17:45 | dsrx | well, idk if C-X will do anything different than C-x :p |
| 17:45 | rasmusto | is there a way to do it from (blah (foo| bar)) ? |
| 17:45 | rasmusto | typo :) |
| 17:48 | aaron_ | hi |
| 17:49 | fu86 | I want to use enlive to fetch the content from an url which is not html |
| 17:50 | fu86 | enlive-html/get-resource seems the right way, but I don't know how to use it :( |
| 17:50 | brehaut | fu86: look at clj-http for general purpose http requests |
| 17:51 | brehaut | from memory enlive-html/get-resource will always try to consume the resourse as html |
| 17:51 | fu86 | thanks |
| 17:51 | fu86 | i've tried to use an "custom" loader which simply return the content but it does not work |
| 17:52 | fu86 | i will use clj-http for such stuff, thanks |
| 17:57 | amalloy | rasmusto: (blah (foo| bar)) there's not really a good way to define how far up you want to go - should it eval (foo bar), or (blah (foo bar)), or the context surrounding that, or...? |
| 17:57 | amalloy | or indeed just foo |
| 17:57 | amalloy | instead, use C-x C-e as-is, and if you want to go up N levels first use paredit-forward-up N times (or with a prefix argument of N) |
| 17:59 | amalloy | i have paredit-{forward,backward}-{down,up} on the C-M-{WESD} keys, although in practice i mostly use backward-up |
| 18:02 | rasmusto | amalloy: hm, ok. |
| 18:11 | rasmusto | ahh, emacs has cider now |
| 18:11 | rasmusto | something must have pulled it in as a dep |
| 18:15 | cbp | magnars: ping |
| 18:39 | seangrove | dnolen_: Any from the Om instrumentation, I'd like dynamically swap the render method on a component before passing it to om/build* - e.g. to have a preview render method and an actual render method that otherwise uses the same methods for everything else |
| 18:39 | seangrove | dnolen_: Any suggestions on how I'd go about that? |
| 18:44 | storme | hello |
| 18:44 | storme | anyone use yokogiri before? |
| 18:45 | Raynes | I bet devn has. |
| 18:50 | storme | I'm trying to make an image scraper with clojure |
| 18:51 | storme | would yokogiri be useful for this? |
| 18:51 | SegFaultAX | Haha, is that a thing? yokogiri? |
| 18:51 | storme | yep lol |
| 18:51 | storme | <-- coming from ruby |
| 18:51 | SegFaultAX | That's an awesome name. |
| 18:51 | storme | so not sure what clojure libs are out there |
| 18:52 | SegFaultAX | storme: Are you scraping html? |
| 18:52 | storme | yep |
| 18:53 | SegFaultAX | enlive can do that. I haven't used yokogiri so I don't know how good it is. |
| 18:53 | storme | ah thanks |
| 18:57 | rasmusto | "defun" indentation isn't working for me in clojure-mode, is there something I have to set? |
| 19:01 | rasmusto | ugh, I had lisp-indent-offset set somewhere :< |
| 19:05 | Cr8 | I usually use Jsoup for scraping |
| 19:06 | Cr8 | It can do css-like selectors, and its types mostly participate in the java collection interfaces, so you can seq over them and such |
| 19:13 | devn | storme: oh hello there |
| 19:14 | devn | yokogiri is really, really simple. enlive and all of those others things can do scraping, but I got annoyed with how much other stuff they brought to the table that I simply didn't need |
| 19:14 | devn | yokogiri is named after yoko harada, who did the JRuby implementation for Nokogiri |
| 19:15 | storme | so for an image scraper, would you recommend yokogiri over enlive? |
| 19:15 | devn | storme: i mean, you're asking me if i'd recommend it? I wrote it, so yeah! It's great! :) |
| 19:15 | devn | it's pretty darn easy |
| 19:15 | devn | in fact, let me dig up a snippet for you |
| 19:16 | storme | oh haha |
| 19:26 | devn | hmph, thanks for asking about it storme -- i just was whipping up an example for you, and found that I'm getting an exception from org.apache.httpclient |
| 19:30 | storme | devn working for me |
| 19:34 | devn | yeah, i think i just had a blocked outgoing connection for emacs |
| 19:36 | SegFaultAX | 1.6 is RC... oh snap. |
| 19:44 | devn | storme: https://gist.github.com/devn/9654084 |
| 19:45 | devn | storme: hopefully that speeds things along for you. |
| 19:45 | SegFaultAX | devn: That's slick. Nokogiri is such a useful library, thanks for making a Clojure variant. :) |
| 19:46 | devn | yeah dude, no problemo. it's just a wrapper around a couple libs. There's a lot Nokogiri will do that yokogiri won't at the moment |
| 19:47 | devn | contributions would be much appreciated |
| 19:47 | SegFaultAX | I'm assuming that's the node building side? |
| 19:47 | SegFaultAX | It seems like the query side is pretty fleshed out. |
| 19:47 | devn | yeah, i mean, on the other hand |
| 19:47 | devn | i think tools like enlive and laser and so on already have that |
| 19:47 | SegFaultAX | Yup. |
| 19:48 | devn | they all sort of seemed to have more complicated screen scraping stories though IMO |
| 19:48 | devn | i just wanted a thingy to query really, so *shrug* |
| 19:48 | muhoo | scraping with emlive ia pretty painless |
| 19:48 | storme | thanks a lot devn! |
| 19:49 | devn | yeah no problem, also, welcome :) |
| 19:50 | devn | muhoo: yeah? It was a long time ago, i remember like 3 tutorials were written for it |
| 19:51 | devn | which to me was not something i was very interested in, since i just wanted to, as i said above, do some basic screen scraping |
| 19:51 | muhoo | it worked for me |
| 19:52 | devn | like, the fact that tutorials needed to be written to understand how to get a page and find an element was kind of crazy |
| 19:53 | devn | muhoo: actually, i take that back i guess. i think i just wanted to use xpath and css out of familiarity |
| 19:54 | devn | muhoo: i'd rather not need to find a fn called 'first-child' when i already know how to do that with xpath |
| 19:54 | devn | but meh, to each their own. i should give enliven another look |
| 19:57 | justin_smith | first-child, not to be confused with firstborn-child - (promise firstborn-child) |
| 20:14 | TimMc | justin_smith: It really is too bad promise doesn't take any arguments. |
| 20:17 | xeqi | seangrove: have you done any perf check on java8's new js engine for server side rendering? |
| 20:18 | seangrove | xeqi: No, dnolen has been battering away on it for the past month though |
| 20:24 | justin_smith | TimMc: (let [offering (promise)] (deliver offering firstborn-child)) |
| 20:25 | arrdem | hahaha |
| 20:26 | arrdem | the gods of pointer arithmetic find your offering pleasing |
| 20:26 | arrdem | your application may run |
| 20:26 | arrdem | for now |
| 20:26 | TimMc | :-D |
| 20:26 | TimMc | &(let [p (promise)] (deliver p p)) |
| 20:26 | lazybot | java.lang.StackOverflowError |
| 20:26 | brehaut | wait wat? surely only if its dereffed? |
| 20:27 | brehaut | &(let [p (promise)] (deliver p 1)) |
| 20:27 | lazybot | ⇒ #<core$promise$reify__6153@3827c3: 1> |
| 20:27 | brehaut | oh right |
| 20:28 | TimMc | It's just print silliness. |
| 20:28 | TimMc | &(pst) |
| 20:28 | lazybot | java.lang.RuntimeException: Unable to resolve symbol: pst in this context |
| 20:28 | gfredericks | ,(deftypo Foo [] java.util.Map clojure.lang.IDeref) |
| 20:29 | clojurebot | #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: deftypo in this context, compiling:(NO_SOURCE_PATH:0:0)> |
| 20:29 | gfredericks | ,(deftype Foo [] java.util.Map clojure.lang.IDeref) |
| 20:29 | clojurebot | sandbox.Foo |
| 20:29 | cbp | lol deftypo |
| 20:29 | gfredericks | ,(Foo.) |
| 20:29 | clojurebot | #<IllegalArgumentException java.lang.IllegalArgumentException: Multiple methods in multimethod 'print-method' match dispatch value: class sandbox.Foo -> interface clojure.lang.IDeref and interface java.util.Map, and neither is preferred> |
| 20:29 | TimMc | brehaut: Interestingly, the overflow is in java.util.regex.Pattern$BmpCharProperty.match or so. |
| 20:29 | justin_smith | that is an awesome typo |
| 20:29 | gfredericks | deftypo is the weirdest feature of clojure |
| 20:29 | rasmusto | is that a core.logic thing? deftypeo? |
| 20:29 | TimMc | That's from the Intercal side of Clojure's history. |
| 20:29 | gfredericks | I did core.logic for a month and now I can't stop putting o's at the end of things |
| 20:30 | Bronsa | here's a funny one |
| 20:30 | Bronsa | ,(deftype x [a]) |
| 20:30 | clojurebot | sandbox.x |
| 20:30 | Bronsa | ,(x. 1 nil {:a 2}) |
| 20:30 | clojurebot | #<CompilerException java.lang.IllegalArgumentException: No matching ctor found for class sandbox.x, compiling:(NO_SOURCE_PATH:0:0)> |
| 20:30 | Bronsa | uh |
| 20:30 | Bronsa | derp |
| 20:30 | Bronsa | ,(defrecord x [a]) |
| 20:30 | clojurebot | sandbox.x |
| 20:30 | Bronsa | ,(x. 1 nil {:a 2}) |
| 20:30 | clojurebot | #sandbox.x{:a 1, :a 2} |
| 20:31 | TimMc | oooo |
| 20:31 | arrdem | o0oooooh |
| 20:31 | TimMc | Basis and extra attrs? |
| 20:31 | Bronsa | yeah |
| 20:31 | justin_smith | (leto [wiki "http://dune.wikia.com/wiki/Leto_Atreides_I"] wiki) |
| 20:31 | arrdem | justin_smith: har har |
| 20:31 | gfredericks | is there a ticket for that one? |
| 20:31 | cbp | ,(update-in (x. 1 nil {:a 2}) [:a] inc) |
| 20:31 | clojurebot | #sandbox.x{:a 2, :a 2} |
| 20:32 | Bronsa | gfredericks: not that I know of. |
| 20:32 | gfredericks | I guess it would be pretty low priority |
| 20:32 | Bronsa | gfredericks: there's to say that that ctor is an implementation detail |
| 20:32 | gfredericks | I didn't even know about it |
| 20:32 | Bronsa | __meta + __extmap |
| 20:32 | rasmusto | ,((x. 1 nil {:a 2}) :a) |
| 20:32 | clojurebot | #<ClassCastException java.lang.ClassCastException: sandbox.x cannot be cast to clojure.lang.IFn> |
| 20:32 | rasmusto | ,(get (x. 1 nil {:a 2}) :a) |
| 20:32 | clojurebot | 1 |
| 20:33 | Bronsa | watch this |
| 20:33 | gfredericks | fun fact: you can't add metadata to a lazy seq without forcing it |
| 20:33 | Bronsa | ,(map->x (x. 1 nil {:a 2})) |
| 20:33 | clojurebot | #sandbox.x{:a 1} |
| 20:33 | arrdem | Bronsa: this is absurd. thanks for sharing :D |
| 20:33 | Bronsa | oh damn |
| 20:33 | arrdem | gfredericks: that doesn't make sense... |
| 20:33 | Bronsa | that returns #user.x{:a 2} in my repl |
| 20:33 | Bronsa | w/e |
| 20:33 | rasmusto | probably not a big deal |
| 20:34 | gfredericks | arrdem: it makes good sense after digging into the impl for a while |
| 20:34 | cbp | ,*clojure-version* |
| 20:34 | clojurebot | {:interim true, :major 1, :minor 6, :incremental 0, :qualifier "master"} |
| 20:36 | Bronsa | arrdem: I "discovered" this behaviour while investigating for CLJ-1323 |
| 20:36 | Bronsa | that's a funny one too |
| 20:36 | gfredericks | arrdem: basically you don't want two objects holding onto the same thunk |
| 20:36 | arrdem | Bronsa: I'm so jelly of the fun you have playing with the core |
| 20:36 | gfredericks | cuz that would require a lot more coordination |
| 20:36 | arrdem | Bronsa: one day I'll catch up, but not this semester :c |
| 20:37 | arrdem | gfredericks: hum... that makes sense but I'd have to read it myself. |
| 21:02 | amalloy | gfredericks: you also can't conj onto a lazy-seq without forcing it, as i recall |
| 21:02 | amalloy | forcing the first element, anyway; of course you don't have to force the whole thing |
| 21:11 | hyPiRion | gfredericks: that would just force the first element, right? |
| 21:11 | hyPiRion | unless that's exactly what amalloy said. |
| 21:12 | amalloy | well, i was saying it about conjing; you're saying the same thing about metadata, which is also true |
| 21:12 | hyPiRion | Yeah, I was unsure whether you only meant conjing or both |
| 21:26 | Steveo_ | where can I find the ant colony code that rich refers to in his concurrency talk? |
| 21:27 | gfredericks | amalloy: that one is not obvious to me |
| 21:28 | gfredericks | ,(let [a (atom false) l (lazy-seq (swap! a not) nil)] [(first (conj l 3)) @a]) |
| 21:28 | clojurebot | [3 true] |
| 21:28 | gfredericks | ,(let [a (atom false) l (lazy-seq (swap! a not) nil)] [(first (cons 3 l)) @a]) |
| 21:28 | clojurebot | [3 false] |
| 21:28 | gfredericks | weeeeird |
| 21:29 | cbp | o_O |
| 21:29 | amalloy | gfredericks: because cons just creates a Cons object, but conj dispatches to .cons on the collection |
| 21:29 | amalloy | and lazy-seq delegates most of its operations to (.seq this) |
| 21:30 | gfredericks | ooh okay |
| 21:30 | gfredericks | so the lazy-seq impl is just lazy |
| 21:30 | amalloy | yes. it definitely could be better. instead it's lazily not being lazy |
| 21:31 | rasmusto | heh |
| 21:31 | amalloy | gfredericks: they're laughing lazily. only if you force them will you hear the sound |
| 21:32 | arrdem | haha |
| 21:32 | brehaut | gfredericks: no need to be so strict |
| 21:33 | gfredericks | by "glaring at" of course I meant "forcing" |
| 21:35 | storme | (reduce + (range 16 128)) |
| 21:35 | storme | ,(reduce + (range 16 128)) |
| 21:36 | clojurebot | 8008 |
| 21:36 | storme | ,(/ (reduce + (range 19 173)) 11) |
| 21:36 | clojurebot | 1337 |
| 21:37 | gfredericks | ,(take 5 (for [x (range) y (range x) :let [z (range y x) s (seq (str z))] :when (= s (reverse s))] [y x z])) |
| 21:37 | storme | ,(str (/ (reduce + (range 19 173)) 11) (reduce + (range 16 128))) |
| 21:37 | clojurebot | "13378008" |
| 21:37 | clojurebot | Execution Timed Out |
| 21:37 | storme | ,(str (/ (reduce + (range 19 173)) 11) " " (reduce + (range 16 128))) |
| 21:37 | clojurebot | "1337 8008" |
| 21:37 | gfredericks | ,(take 2 (for [x (range) y (range x) :let [z (range y x) s (seq (str z))] :when (= s (reverse s))] [y x z])) |
| 21:37 | clojurebot | Execution Timed Out |
| 21:37 | gfredericks | oh whops |
| 21:38 | gfredericks | ,(take 5 (for [x (range) y (range x) :let [z (reduce + (range y x)) s (seq (str z))] :when (= s (reverse s))] [y x z])) |
| 21:38 | clojurebot | ([0 1 0] [0 2 1] [1 2 1] [0 3 3] [1 3 3]) |
| 21:38 | gfredericks | ,(take 5 (for [x (range) y (range x) :let [z (reduce + (range y x)) s (seq (str z))] :when (> z 10) :when (= s (reverse s))] [y x z])) |
| 21:38 | clojurebot | ([5 7 11] [4 8 22] [3 9 33] [2 10 44] [0 11 55]) |
| 21:38 | amalloy | gfredericks: what on earth |
| 21:38 | gfredericks | storme made me wonder how often (range x y) is a palindrome :P |
| 21:39 | gfredericks | ,(take 5 (drop 50 (for [x (range) y (range x) :let [z (reduce + (range y x)) s (seq (str z))] :when (> z 10) :when (= s (reverse s))] [y x z]))) |
| 21:39 | clojurebot | ([9 34 525] [33 34 33] [0 35 595] [1 35 595] [5 35 585]) |
| 21:39 | storme | lol gfredericks |
| 21:39 | gfredericks | ,(take 5 (drop 50 (for [x (range) y (range 10 x) :let [z (reduce + (range y x)) s (seq (str z))] :when (> z 10) :when (= s (reverse s))] [y x z]))) |
| 21:39 | clojurebot | ([31 45 525] [32 45 494] [44 45 44] [12 46 969] [34 46 474]) |
| 21:39 | amalloy | ah, i see |
| 21:39 | gfredericks | ,(reduce + (range 34 46)) |
| 21:39 | clojurebot | 474 |
| 21:40 | gfredericks | ,(take 5 (drop 50 (for [x (range) y (range 10 x) :let [z (reduce + (range y x)) s (seq (str z))] :when (> z 1000) :when (= s (reverse s))] [y x z]))) |
| 21:40 | clojurebot | ([53 102 3773] [80 102 1991] [74 103 2552] [27 104 5005] [16 106 5445]) |
| 21:40 | gfredericks | ,(take 5 (drop 50 (for [x (range) y (range 10 x) :let [z (reduce + (range y x)) s (seq (str z))] :when (> z 10000) :when (= s (reverse s))] [y x z]))) |
| 21:40 | clojurebot | ([91 192 14241] [85 194 15151] [26 197 18981] [50 197 18081] [86 197 15651]) |
| 21:40 | gfredericks | so much math I'm making clojurebot do |
| 21:40 | storme | ,(:require [yokogiri.core :as html]) |
| 21:40 | clojurebot | #<CompilerException java.lang.ClassNotFoundException: yokogiri.core, compiling:(NO_SOURCE_PATH:0:0)> |
| 21:40 | storme | is clojurebot open source? |
| 21:41 | arrdem | $source |
| 21:41 | lazybot | Source not found. |
| 21:41 | arrdem | ~source |
| 21:41 | gfredericks | $google clojurebot |
| 21:41 | lazybot | [hiredman/clojurebot · GitHub] https://github.com/hiredman/clojurebot |
| 21:41 | storme | ty |
| 21:41 | arrdem | clojurebot: source |is| $google clojurebot |
| 21:42 | gfredericks | ,(println "$google clojurebot") |
| 21:42 | clojurebot | $google clojurebot\n |
| 21:42 | arrdem | gfredericks: bots ignore each-other, remember? |
| 21:42 | storme | ,(println "I'm a bot") |
| 21:42 | clojurebot | I'm a bot\n |
| 21:42 | arrdem | well.. ours do. |
| 21:42 | rasmusto | ,(print "$google clojurebot") |
| 21:42 | clojurebot | $google clojurebot |
| 21:42 | rasmusto | ##(print "$google clojurebot") |
| 21:42 | lazybot | ⇒ $google clojurebotnil |
| 21:42 | storme | ,(println "$google clojurebot") |
| 21:42 | clojurebot | $google clojurebot\n |
| 21:43 | storme | ,(println "$google lazybot") |
| 21:43 | clojurebot | $google lazybot\n |
| 21:43 | arrdem | storme: what did I say about the bots ignoring each other... |
| 21:43 | storme | ah, missed that arrdem |
| 22:26 | amalloy | lazybot ignores clojurebot |
| 22:26 | amalloy | but not vice versa, as i recall |
| 22:26 | amalloy | &(symbol ",1") |
| 22:26 | lazybot | ⇒ ,1 |
| 22:27 | amalloy | oh, right. he prefixes his output so that clojurebot doesn't interpret it |
| 22:27 | rasmusto | print a backspace character |
| 22:40 | xeqi | echo ,(println "&(+ 1 2)") |
| 22:40 | lazybot | ,(println "&(+ 1 2)") |
| 22:40 | clojurebot | &(+ 1 2)\n |
| 23:19 | gfredericks | $mail clojurebot ,(println "hello") |
| 23:19 | lazybot | Message saved. |
| 23:19 | gfredericks | $mail clojurebot ,(def sent true) |
| 23:19 | lazybot | Message saved. |
| 23:20 | gfredericks | ,sent |
| 23:20 | clojurebot | #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: sent in this context, compiling:(NO_SOURCE_PATH:0:0)> |
| 23:33 | TimMc | gfredericks: clojurebot would have to ask for mail. |
| 23:52 | amalloy | oh my god. now lazybot is going to be /msging clojurebot all the time |
| 23:52 | amalloy | i'm glad i added $unmail for admins |