#clojure log - Oct 10 2014

The Joy of Clojure
Main Clojure site
Google Group
List of all logged dates

0:41 brainproxy: is there a enlive selector for :type ? e.g. suppose I want to do a transform on comment nodes

0:52 johann: brainproxy: ya try using the :comment keyword

0:56 brainproxy: johann: how? (html/select resource [:comment]) ... that doesn't seem to work

0:57 johann: brainproxy: can you show me a snippet of the html you are working with?

0:57 brainproxy: it's just some html which has <!-- comments --> in various places

0:57 nothing unusual about it at all

0:58 johann: oh! i thought you were looking at xml with <comment></comment> tags

0:58 brainproxy: so, for example, if I swap :div for :comment, I get what I would expect

0:58 ah, I see

0:58 well what I want is to be able to select nodes which aren't tags per se

0:58 johann: ya interesting

0:59 brainproxy: they're in the sequences returned by enlive

0:59 {:type :comment ...}

1:00 johann: but your issue is that you can't access it with the select function

1:00 right?

1:00 brainproxy: are with html/at, etc.

1:00 johann: but when you parse the entire document it shows it

1:00 brainproxy: is there a glob selector for any and all nodes?

1:01 then I could write my own transformer and choose to do something if (:type node) is :comment

1:01 otherwise, just dump node back out

1:01 johann: ya what you want is html/html-resource i think

1:02 brainproxy: yep, using that

1:02 I mean a selector that will match any node

1:02 within the body of html/at usage, etc.

1:03 johann: so you basically want just the clojure data of the html?

1:03 sorry if i'm not helping! its late over here :)

1:04 squeedee: You know what I am finding harder than understanding functional programming.. functional unit tests.

1:04 brainproxy: no I have that also

1:04 :)

1:04 squeedee: I seem to only know how to define imperitive unit tests

1:05 johann: squeedee: always some resistance when taking on new modes of thought!

1:07 squeedee: i very quickly grasped how to think about the decomposition of a problem into pure functions. No problem with composition using currying, lookups, and by pivoting on state.

1:07 justin_smith: squeedee: functional unit tests are easier because with truly functional code you don't need mocking or anything else: just check if a given input gives the correct output

1:07 squeedee: I'm not saying i'm suddenly great at functional, but I find it pretty easy to grok..

1:07 the tests, not so much..

1:07 justin_smith: what's hard about the tests?

1:08 johann: ya it'd be cool for you to show an example

1:08 squeedee: Fact is my unit testing habbits in imperative style are what brought me here

1:08 yeah ok lemme commit my rubbish

1:09 by brought me here, i mean, my unit testing made me more and more 'functional' even in my imperative languages

1:09 johann: brainproxy: i'm still a little unsure of what you are trying to accomplish. if you can be patient with me and maybe try to reword it i can dedicate some good willed googling in your direction ;)

1:11 squeedee: what do you mean by that? like the scope of your tests closed in to the smallest possible units (in most problem spaces i mean) and voila its function composition?

1:13 squeedee: johann not voila

1:13 it made me appreciate functional composition more. It also made me appreciate data-is-king

1:13 https://github.com/squeedee/lsystem/blob/master/test/lsystem/core_test.clj

1:14 johann: ya the data is king thing is still sort of marvelous to me

1:15 squeedee: johann: my tests wanted minimal dependencies for simplicity, and wanted to do very little, so rather than a ball of state with a ball of behaviours around it, i wanted functions that were not tied to the design of the object

1:15 so a lot of my objects became 'command' style objects.

1:16 which is a really loose way of saying 'function'

1:16 johann: yeah right

1:16 i've been working with quil, which is a clojure api over processing, so i'm dealing with a little oo lately

1:17 brainproxy: johann: (html/at resource **sel-for-any-node** (fn [n] (if (= (:type n) :comment) something n)))

1:17 johann: and i finally got that like an object is this well intended behavior that is a big pain to use outside of its creation-context

1:17 squeedee: anyway, it made sense to me what rich was talking about when i listened to him, and what folks were saying elsewhere in the functional camp. But tests, not so sure about them.

1:18 johann: what's your uncertainty?

1:19 squeedee: Tho to be certain, this is my first clojure app, and i know im just stood at the starting line

1:19 well I'm used to my tests driving my design.

1:19 justin_smith: looking at that test code, I see a bunch of stateless functions, where you provide an input and verify the output. This is the right way to do clojure testing.

1:20 johann: squeedee: what's your workflow right now? are you using a repl?

1:21 squeedee: i guess that makes sense, but what i found was that I made a choice for all the tests at once, not an incremental, refactored choice as i tried to make all conditions pass

1:22 for this i moved out of the repl because i wanted to try test driving it. So I'm using LT and lein autoexpect

1:22 johann: kk

1:22 squeedee: I use eval in LT a lot

1:22 johann: ya i think having a short feedback loop to evaling smaller parts changes the role of tests a little

1:23 squeedee: dunno about that, works well in ruby, have a repl there

1:23 johann: squeedee: i'm playing with enlive right now

1:23 justin_smith: squeedee: side note, your current tests could be much more concise using (clojure.test/are [x y] (= x (iterate test-rules y)) [\a] [\a] (vec "ab") (vec "ab") ...)

1:24 brainproxy: johann: so the correct way to express **self-for-any-node** is [html/any-node]

1:24 johann: wait i meant brainproxy for that last tag my bad

1:25 justin_smith: squeedee: http://clojuredocs.org/clojure.test/are

1:30 squeedee: justin_smith: thanks for the tip, not sure how i feel about it tho. Looks like 'examples' in other frameworks i've seen.

1:31 justin_smith: looks like expectations isn't compatible with clojure.test anyway, I didn't realize

1:31 squeedee: sorry i mean 'theories'

1:33 justin_smith: it has this: http://jayfields.com/expectations/templating.html

1:34 although thats when you change the subject, not the parameters

1:38 vIkSiT: 'lo all

1:38 has anyone used clojurescript secretary routes here?

1:38 in conjunction with something like luminus?

1:38 how do you ensure that routing from client side hpappens, while everything else is served from luminus?

1:38 eg, /home should use scretary

1:39 but /api/getHomeData should use luminus

2:55 learn2code: hi, i come here to learn clojure

2:55 TEttinger: learn2code, you picked the right place

2:56 learn2code: but this is the first time I am here

2:56 and in IRC

2:56 i dont know how IRC acutally works

2:56 TEttinger: there's some good stuff here http://clojure-doc.org/articles/content.html#essentials

2:56 it's a chatroom

2:56 there are many people here, but most aren't talking due to time zone differences

2:57 learn2code: yeah, it's so quiet now

2:57 TEttinger: during the day in the US it's much busier.

2:58 learn2code: yeah, i am in US

2:58 I will try it tmr morning

2:58 TEttinger: is there a particular type of application you want to make with clojure?

2:58 learn2code: yes, I do have

2:58 TEttinger: web apps benefit a lot from it

2:58 I make games

2:58 learn2code: i want to build a small app

2:59 just keep polling ebay

2:59 TEttinger: there's probably an API that can push updates?

3:00 learn2code: and notify me whenever a new listed shows up and the price is acceptable to me

3:00 the input will be the <name> <price>

3:00 yeah, I know

3:00 clj-ebay

3:00 i am using it

3:00 TEttinger: good!

3:01 learn2code: for keep polling the server I use at-at

3:01 the point here is

3:01 how to connect them together

3:01 and handle the commandline to read arguments

3:05 TEttinger: at-at is great, yeah

3:05 there is a lib for better command line arg parsing

3:07 https://github.com/clojure/tools.cli

3:08 learn2code: thanks TEttinger

3:09 TEttinger: no prob

3:11 learn2code: how long have you been wokring with clojure

3:12 do you know any interesting app with good doc in Clojure that I can find on Github

3:16 TEttinger: oh, sorry learn2code , I was in another channel.

3:18 I'd say I've been working with clojure, on and off, for... maybe 3-4 years. it took me a long time to learn it and I kept getting impatient. it's easier to learn now, more resources

3:18 http://4clojure.com (or was it .net) is good

3:18 learn2code: lol

3:18 i am looking at it now

3:19 TEttinger: I'd say leiningen itself may be a good option for a well-documented app on github

3:19 but I haven't read its source

3:20 there's documented source for lein here http://leiningen.org/reference.html

3:21 learn2code: nice

3:22 TEttinger: you'll likely encounter a lot of symbols that are confusing early on, there's a good guide out there...

3:23 learn2code: yeah, I collected a cheatsheet for me already

3:24 TEttinger: http://yobriefca.se/blog/2014/05/19/the-weird-and-wonderful-characters-of-clojure/

3:26 learn2code: wow

3:26 thanks TEttinger

3:26 TEttinger: no problem! glad to help

3:27 learn2code: i am going to sleep

3:27 is there anyway to meet you again

3:27 in IRC?

3:27 the same nick?

3:27 TEttinger: I'll likely be here in #clojure most days, same nick

3:27 learn2code: nice

3:27 g9

3:28 TEttinger: I have an IRC client that auto-joins several channels, #clojure is one

3:29 learn2code: you built by yourself or had from somewhere else

3:30 that's exactly what I want

3:35 Rhainur: learn2code: most IRC clients do auto-joining

4:13 yekKhaste: hi every body!

4:13 I want to start learning clojure

4:14 but I have no exprince with lisp or functional programming

4:14 Where is best place to start?

4:15 johann: yekKhaste: hi!

4:15 yekKhaste: obviously this is a matter of personal taste, but this is a really nice resource for beginners http://www.braveclojure.com/

4:17 yekKhaste: johann, Thank, but what next?

4:17 where can I learn more about clojure

4:17 johann: i'd suggest starting off with the clojurekoans just so you can get introduced to the syntax and core ideas

4:17 http://clojurekoans.com/

4:18 yekKhaste: And is it possible to build android app with clojure?

4:18 I hate java, but I like make android app

4:19 johann: ya theres a mailing list of people who use clojure on android

4:19 yekKhaste: johann, Thank a lot!

4:19 johann: no problem!

4:20 if you need help setting your environment or anything let me know

4:20 you should also check out lighttable. http://lighttable.com/

4:22 heres the mailing list for clojure-android if you want to ask them any questions https://groups.google.com/forum/#!forum/clojure-android

4:49 kenrestivo: i vaguely remember seeing a macro somewhere that'd turn [foo bar] into {:foo foo, :bar bar}

4:49 now i can't find it tho

4:52 scottj: kenrestivo: flatland.useful.map/keyed

4:52 kenrestivo: aha! thanks, i had a feeling it'd be a flatland

5:01 donbonifacio: is there a way to reload all code in the repl?

5:03 schmir: donbonifacio: https://github.com/clojure/tools.namespace

5:03 kungi: donbonifacio: yes

5:03 donbonifacio: try tools.namespace, but think about global state first

5:08 donbonifacio: thanks :) I don't have global state. On my workflow I edit files in vim and after a while the repl stays out of date, and I'd like a sync without restart

5:10 kungi: donbonifacio: then tools.namespace.repl/refresh probably is what you need

5:10 donbonifacio: If you write a webapp you can also use a ring middleware to reload your files

7:56 conorbev: I just came across the function alter-var-root. Am I missing something or is it understandable that I'm surprised it doesn't end in ! ? It seems pretty similar in function to swap! on an atom.

8:03 noncom: conorbev: idk.. who knows..

8:07 conorbev: OK :)

8:17 skratl0x1: can anyone look at this one and suggest some improvements? http://stackoverflow.com/questions/26299169/updating-nested-structures-in-clojure

8:23 noncom: skratl0x1: well, idk, but try looking at a solution using (reduce) or reducers .. or even (loop)...

8:23 just implement them and see how they work..

8:27 skratl0x1: noncom: aren't reducers mostly for multi-core applications? this one will run in cljs, but if reducers can simplify it / make it more readable, then why not, but how?

8:28 schmir: skratl0x1: I do think this is a bit too much magic. and I fail to see why the "by-hand" approach leads to ugly code

8:29 noncom: skratl0x1: try first doing this with (reduce).. at least you will see the difference

8:30 skratl0x1: schmir: because the application of this function is much more clear than reading through (into {} (for.. (for.. (into {} (for... (for ...

8:30 this is more generic approach to the problem, it makes the code much more readable

8:36 schmir: skratl0x1: (map-values (fn [list-of-geoms] (map inc-buffer list-of-geoms)) sample) IMHO is much more readable. you just have to implement map-values and inc-buffer.

8:36 (unless you really need the generic solution)

8:37 I think you should use a keyword instead of *

8:37 dmi3y: Guys, I'm not able to use clj-time in my project for some reason: https://www.refheap.com/91532 Does anyone know what am I doing wrong?

8:38 skratl0x1: schmir: yeah, I made it use the * for its common use as a wildcard

8:39 schmir: anyway, the question was, given the API that I want, how to make the function /better/

8:40 I don't think paths (including wildcards) or selectors is any kind of anti-pattern

8:41 not to mention that the f actually has to take the current path too

8:50 clgv: dmi3y: looks like a dependency conflict. what does "lein deps :tree" tell you?

8:53 dmi3y: @clgv yup, my deps tree looks scarry: https://www.refheap.com/91536

8:54 clgv thanks for the tip

8:55 clgv: dmi3y: [org.elasticsearch/elasticsearch-hadoop "2.0.1"] wins of clj-time with respect to joda-time

8:55 s/of/over/

8:56 dmi3y: so it probably pulls in an incompatible joda-time

8:56 dmi3y: so, can I just exlude joda-time from elasticsearch-hadoop deps?

8:58 clgv: dmi3y: without knowing that lib I can't tell. you can try - provided you have a good test suite you can be confident to a certain degree whether it works or doesn't work

8:58 dmi3y: clgv: got it, thank you

8:58 clgv: btw. why does "lein deps :tree" always print "warn" as first line?

8:59 schmir: skratl0x1: the function creates elements for non-existent keys

9:01 skratl0x1: (update-each-in {} [:xxx * :foo] first)

9:01 clgv: skratl0x1: depending on your scenario something like clojure.walk/postwalk-replace might be handy

9:09 gfredericks: reiddraper: I've been pondering the problem of "test.check only generates tiny numbers" and was thinking maybe the size should bound the log of the numbers generated instead of their actual value?

9:10 I'm also trying to tackle gen-double is how I got on that topic

9:11 "just use a bigger size" is problematic because collections

9:12 maybe we could add generators for "small numbers" that work like the current ones, for whenever that's useful

9:12 or just assume people know how to call Math/log

9:27 clgv: gfredericks: what is the concrete problem?

9:38 mdeboard: global concrete shortage

9:39 I seriously can't believe I'm getting porn spam on irc

9:39 rolfb: mdeboard: spam on irc? never experienced

9:40 unsolicitied pm's with links?

9:40 mdeboard: yeah

9:40 Bronsa: it's more frequent than one might expect

9:40 mdeboard: and it's ONLY in the clojure channel

9:40 Only from bots in the clojure channel, that is

9:40 ken_barber: yeah, we get lots in #puppet, pm or in chan

9:40 mdeboard: I'm in several other high traffic channels, never get bots

9:40 rolfb: wow

9:40 ken_barber: oh speak of the devil, its trisnti pinging me now

9:40 mdeboard: yeah

9:40 that was what prompted

9:50 gfredericks: clgv: in the default usage of test.check you'll pretty much never generate numbers outside of the range -200 ... 200

9:50 clgv: gfredericks: ah right, because of the shrinking strategy

9:50 gfredericks: no

9:50 because of sizing

9:51 clgv: gfredericks: yeah is that not related to shrinking as well?

9:51 gfredericks: in that large numbers could prompt enormous shrinks?

9:52 clgv: ok, so you want to add generators with a maximum range that grows much faster or that does not grow at all?

9:52 gfredericks: grows much faster

9:52 exponentially faster probably

9:52 clgv: just trying to understand the issue ;)

9:53 gfredericks: i.e., it should not require customization to generate arbitrary numbers in the range of Long/{MIN,MAX}_VALUE

9:53 clgv: ok those are probably usefull for algorithms on numbers or differently stated: numbers that do not correspond to collection sizes or similar

9:53 ok, sounds pretty useful

9:53 gfredericks: yeah that's why I was thinking both sizing strategies could be useful

9:54 I'll probably just start by adding some generators to test.chuck

9:54 clgv: chuck the groundhog? :P

9:54 * gfredericks jumps on bicycle

9:55 * clgv usually sits on a chair when programming ;)

10:16 verma: what's the opposite of cider-jack-in? Like I want to kill the repl stuff that's running

10:16 CookedGryphon: cider-quit

10:17 verma: nice

10:17 yay

10:17 (inc CookedGryphon)

10:17 lazybot: ⇒ 3

10:18 clgv: probably you'd sometimes prefer to kill it instead of quit ;)

10:20 verma: clgv there doesn't seem to be a cider-kill

10:20 also, sometimes the repl window goes away

10:20 how do I bring it back/

10:21 clgv: verma: that was just a joke on "kill" ;)

10:21 verma: oh :P

10:22 clgv: but yeah I love my "killall -9 java" :D

10:22 ;)

10:22 verma: Yes, I've been needing it more often past few hours

10:41 perplexa: hi, can i ensure the order of execution of statements?

10:42 Bronsa: perplexa: are you asking if it's guaranteed that in (do 1 2 3) 1 will be executed before 2 before 3?

10:42 perplexa: Bronsa: yes exactly

10:42 Bronsa: yes

10:42 perplexa: i have (a) (b) and (c), which need to get executed in order, b relies on a, and c should be executed last, but since b takes some time c always is first

10:42 which then breaks b

10:43 Bronsa: how would i do that?

10:43 also, trisnti is sending spam/worms/whatever - should remove from channel :P

10:44 clgv: perplexa: woah evil chains of side effects - try to be sure if that is really necessary

10:44 Bronsa: perplexa: I'm not sure I understand what you're asking anymore. it looks like you are doing side-effecty async stuff

10:44 perplexa: clgv: connecting to zookeeper, fetching data and dropping connection, yes, it's necessary i think. and since i only need to do it once, i don't want to leave the connection open

10:45 clgv: perplexa: if this is in parallel you can use promises to synchronize

10:46 perplexa: mh

10:46 clgv: each function can return a promise which is delivered as soon as its work is completed and then you could do (do @(a) @(b) @(c))

10:47 perplexa: ah ok so you need the data from (b) as result value?

10:47 perplexa: yep

10:47 clgv: then just adjust the idea above appropriately ;)

10:48 perplexa: reading up on promises

10:48 clgv: perplexa: humm a different question: doesnt the code initiating the connection block until you have one? and is there not a blocking method to wait for the data?

10:48 perplexa: thank you

10:48 clgv: perplexa: if so you really dont need the promises ;)

10:49 perplexa: clgv: let me check

10:49 clgv: perplexa: if conenction implements java.io.Closeable you could just do (with-open [conn (create-zookeeper-connection)] (get-data-blocking conn))

10:49 gilr00y: test test

10:50 perplexa: clgv: even when i use :async? false it tells me the connection has expired

10:50 the connection is open, that works ;/

10:51 clgv: perplexa: do you use clojure library for zookeeper or its java libs directly?

10:51 perplexa: fetching data usually takes a couple of secs and close always gets called first ;x

10:51 using liebke/zookeeper-clj

10:54 clgv: perplexa: wow last update 3 years ago - are you sure you want to use that?

10:54 perplexa: no but i didn't find anything else

10:55 verma: man, my cljs repls keep dying :(

10:58 any help appreciated: https://www.refheap.com/91544

10:58 perplexa: clgv: i guess i'll instead just establish the zk connection when the app starts and close when it terminates ;/

11:01 verma: I guess I am going to go restart my computer like some neanderthal :(

11:05 drorbemet: Hi, maybe you have any suggestions on how to procede. I am searching for the current solution for completion and navigation in clojure projects. I am using emacs prelude as a base, and I am used to smex and org-mode: Should I use helm-clojure? I think I am struggling with compatibility issues and overlapping of functionality of the following modes: smex, ido, orgmode, company, helm.

11:08 nwjsmith: drorbemet: cider probably https://github.com/clojure-emacs/cider

11:18 zot: is there a nice way to use (-> "resource.edn" io/resource … edn/read) instead of slurp + read-string?

11:18 drorbemet: nwjsmith: yes I am using cider. ... ok right I should try to stick to it's functions first, thanks

11:22 llasram: zot: Not built-in. I define a utility `edn-slurp` or such

11:27 zot: after much more googling, found this: (with-open [r (-> "ph-fs-c1.edn" io/resource io/reader java.io.PushbackReader.)] (edn/read r))

11:27 (that's my working version from the example)

11:29 llasram: zot: Yep, that basic structure should work fine (assuming your EDN files contain only a single top-level EDN form)

11:29 zot: they do ;) i was hoping something would return the PushbackReader directly, but that's still reasonably tight

11:29 si14: how hard can it be to use Friend with bidi?

11:30 llasram: zot: There is not, but that is IMHO for the best. Learn to embrace the platform :-)

11:35 verma: its just a bad day :( can't get any cljs repls to work

11:36 arrdem: justin_smith: I'm pretty sure on that SO question he was working with an elided (deftype Double [x y])

11:40 perplexa: ,(def x) (bound? x)

11:40 clojurebot: #'sandbox/x

11:40 perplexa: meh

11:40 gfredericks: ,(bound? x)

11:40 clojurebot: #<ClassCastException java.lang.ClassCastException: clojure.lang.Var$Unbound cannot be cast to clojure.lang.Var>

11:40 gfredericks: ,(bound? #'x)

11:41 clojurebot: false

11:41 perplexa: hmkay ;x

11:41 ty

11:46 reiddraper: gfredericks: yeah the whole number generation in test.check could use an overhaul, including generating larger numbers when generating scalars

11:46 * gfredericks is trying to write gen-bounded-int right this second

11:47 mmeix: I have a very basic newbie-question

11:47 https://www.refheap.com/91545

11:48 arrdem: mmeix: and is a macro not a function

11:48 mmeix: macros cannot be partially applied

11:48 mmeix: ok

11:48 arrdem: mmeix: note that or is a macro as well.

11:48 mmeix: so this is why (comp not or) will not work

11:48 arrdem: mmeix: what you want here is (fn [x y] (and x y)), which is a function you can throw around

11:48 justin_smith: arrdem: yeah, probably, just didn't want to take that for granted

11:48 arrdem: *correct

11:48 mmeix: coorect

11:49 justin_smith: 's k I'll take all the sweet karma you don't need more anyway

11:49 justin_smith: heh

11:49 mmeix: thanks

11:49 arrdem: mmeix: np

11:50 justin_smith: arrdem: also, that diagram is weird - it provides no way to distinguish between pointing to the head of a list and pointing to the element contained in the item at the head of the list

11:50 verma: any idea why this is happeneing: https://gist.github.com/verma/d3c2ac234f10aac62b0a

11:50 arrdem: justin_smith: agreed

11:50 verma: I feel my defrecord/defprotocol stuff is correct (I hope)

11:50 martinklepsch: what am I doing wrong if my (go .. (<! ... )) only does something the second time something should arrive?

11:50 arrdem: oh god that stack trace

11:50 verma: this is line 31 is my source though; https://gist.github.com/verma/d3c2ac234f10aac62b0a#file-renderer-cljs-L11

11:50 Bronsa: verma: you have a defn inside the defrecord

11:51 verma: oh what

11:51 Bronsa: https://gist.github.com/verma/d3c2ac234f10aac62b0a#file-renderer-cljs-L31

11:51 arrdem: (inc Bronsa)

11:51 verma: oh what the f**ck

11:51 lazybot: ⇒ 60

11:51 verma: (inc Bronsa)

11:51 lazybot: ⇒ 61

11:51 verma: man

11:51 :D

11:51 * arrdem actually lols

11:51 verma: dafuq did that come from

11:51 I am not even sure

11:52 I think I am splitting into two personalites, for real, one tries to fuck up with the other

11:53 Bronsa, everything is working now :) thanks

11:55 Bronsa: np

11:56 mmeix: follow-up question

11:56 I can say:

11:56 (defn nor [v w] (not (or v w)))

11:56 arrdem: sure

11:57 nkoza: mmeix: but nor will evaluate his arguments, you should implement it as a macro to short-circuit like or does

11:57 pyninja: what's the best way to distribute a clojure project as a command line script?

11:59 justin_smith: pyninja: uberjar - jars support embedded shebang

12:00 pyninja: justin_smith, i ran `lein uberjar` and I can do `java -jar target/...-standalone.jar` now. What's the easiest way to avoid having to type `java -jar` every time? What do you mean by embedding shebang?

12:01 justin_smith: https://gist.github.com/briandealwis/782862

12:02 this will never work on Windows, btw

12:02 and the linux version is actually via configuring binfmt-misc iirc

12:02 gfredericks: reiddraper: (->> (gen/sample gen-double 100) shuffle (take 10))

12:03 (-1.892723244743957E-168 -1.6700025521101543E-200 -2.386704789976519E-273 8.305320896742236E-221 -5.260968365987402E-108 6.572354280687105E285 -3.064843548259326E213 -3.2056301999684517E179 4.952015778406743E156 -Infinity)

12:03 pyninja: justin_smith, thanks, will try

12:03 gfredericks: I just fmapped it from a significant and exponent

12:03 I can't remember if there were any known downsides with that approach?

12:06 I just did a failing property that checks that a random double is < 42.0

12:06 it failed at 9.811099542902225E117 and shrunk to 64.0

12:09 reiddraper: gfredericks: yeah i think shrinking was kind of the open question with doubles

12:10 gfredericks: also not sure i follow that last example, both 9.8 and 64.0 it considered < 42.0?

12:10 mmeix: I have a follow-up question (please bear with basic stuff) https://www.refheap.com/91547

12:11 justin_smith: ,(or [false false]) ; mmeix

12:11 clojurebot: [false false]

12:11 justin_smith: ,(boolean (or [false false])) ; mmeix

12:11 clojurebot: true

12:12 justin_smith: it's a collection of falses, and collections are truthy

12:12 Bronsa: ,(some [false nil true])

12:12 clojurebot: #<ArityException clojure.lang.ArityException: Wrong number of args (1) passed to: core/some>

12:12 Bronsa: ,(some identity [false nil true])

12:12 clojurebot: true

12:12 mmeix: ah, i see ... what would be the remedy?

12:12 justin_smith: Bronsa: for style points, (some boolean [false nil true]) would be nice here :)

12:13 Bronsa: mmeix: use some/not-any/every? when you need to apply or/and

12:14 mmeix: aha ... this wan't obvious to me

12:14 Bronsa: justin_smith: meh, truthy and falsey are good enough for me :)

12:14 mmeix: so simple "or" and "and" only for two arguments?

12:15 Bronsa: mmeix: no

12:15 ,(or nil false true)

12:15 clojurebot: true

12:15 mmeix: right

12:15 Bronsa: mmeix: what you're trying to do there is to apply "or" to a collection of values

12:15 mmeix: I wanted to create a nor with variable argument number

12:15 Bronsa: and that's not possible because or is a macro

12:15 you have to use a function for that

12:16 mmeix: so this is a real followup to my older question

12:16 ok - thanks

12:16 Bronsa: clojure has a bunch of functions that usually can be used when you want to "apply" or/and

12:16 bbloom: puredanger: i noticed you've been working on the feature expressions design page, i wanted to point out some extra complications with portable errors, so i updated http://dev.clojure.org/display/design/Platform+Errors

12:16 mmeix: thanks all!

12:16 Bronsa: those are some, not-any?, every?, not-every, every-pred, some-fn

12:17 look them up, the docstrings are pretty nice

12:17 mmeix: great

12:17 thnx again

12:17 Bronsa: np

12:17 bbloom: puredanger: in short, js/Object isn't the superclass of all types, and it's possible to throw numbers, strings, etc that are not instances of Object, so you really do need a :default or comparable no-instanceof form

12:19 cka3yc: how to do realtimeoutput to console ? (clojure.java.shell/sh "ping" "google.com")

12:24 sh returns a map of

12:24 :out => sub-process's stdout (as byte[] or String)

12:25 but how can I read it line by line (before process is finished)?

12:30 anyone?

12:30 clojurebot: anyone is anybody

12:31 clgv: cka3yc: there is a library from raynes that provides the input/ouput streams to you

12:31 Raynes: das conch

12:31 clgv: Raynes: oh he is alive. long time no (irc) see ;)

12:31 Raynes: You don't want the input or output streams though.

12:31 I'm always here.

12:31 Just too busy to stare at the channel all the time these days.

12:31 Always around for pings though :)

12:32 clgv: did I just address you directly with "he" - guess I should be going home soon...

12:33 cka3yc: "You don't want the input or output streams though" why?

12:33 clgv: cka3yc: do you want to interact with the programm as in "simulate a commandline user"?

12:33 Raynes: Oh, sorry, forgot to actually be helpful.

12:34 Conch gives you a high level interface that can return a lazy sequence of lines while the process is running.

12:34 clgv: Raynes: yeah your signal to noise ratio has seen better times ;)

12:34 Raynes: I suggest you use that rather than reimplement it on top of the lower level streams.

12:34 clgv: Raynes: ah interesting, I guess I used a pretty old version of it

12:35 cka3yc: yes, I want to show output to user line by line when it comes in and not when subprocess is done

12:35 clgv: uh right. 0.2.4 ... havent touched that code in a while

12:36 cka3yc: (future (sh/stream-to-out p :out)) ?

12:39 justin_smith: ProcessBuilder isn't super hard to use either, and it provides very flexible streaming of stdout / stderr / stdin (into the process)

12:39 clgv: cka3yc: looks like what you need, from the name ;)

12:40 Raynes: It really isn't.

12:40 The README has detailed instructions for using conch to get a lazy seq of lines from a process.

12:40 clgv: didnt he want to stream output as it's available

12:40 Raynes: He wanted to stream line by line.

12:40 Or, that's what he said.

12:41 cka3yc: thanks, will go learn java then

12:41 clgv: ah ok, I thought that was just a way of saying "asap"

12:42 stuartsierra: Is there a Java process-invocation scheme that gives you NIO channels instead of streams?

12:48 justin_smith: stuartsierra: the default for ProcessBuilder is a ProcessBuilder.Redirect/PIPE, I wonder if that has any relation to nio.channels.Pipe?

12:48 probably not

12:52 stuartsierra: Looks like it only supports redirection to files or inheriting from the current process.

12:53 justin_smith: stuartsierra: it does InputStream / OutputStream

12:53 my usual technique is a thread that reads from the stream and prints when lines are complete

12:53 stuartsierra: Yes, those are blocking I/O. I was wondering if I could do async I/O with a process.

12:53 justin_smith: right

12:56 stuartsierra: Process.getInputStream returns a java.lang.UNIXProcess$ProcessPipeInputStream

12:58 Oh, but you *can* wrap a Channel around an InputStream.

12:58 technomancy: "This project is famous. If that many other packages are referencing it, it must be good and useful." https://www.versioneye.com/java/org.clojure:clojure/1.6.0

12:58 * technomancy nods sagely

12:59 technomancy: wait actually http://chainsawsuit.com/comic/2014/10/08/nod-with-feeling/

13:00 mdeboard: :|

13:02 justin_smith: OT: best company to buy a domain name from?

13:03 arrdem: I've heard good things about Ghandi and they package DNS with domains for dirt cheap

13:03 will probably be moving my stuff to them when my existing dyndns subscriptions get closer to expirint

13:04 clgv: technomancy: haha :D

13:05 gfredericks: oh god cider has gotten itself into some sort of funk where it errors every 1 second

13:07 CookedGryphon: gfredericks: sounds familiar

13:08 gfredericks: couldn't figure out how to disable cider in that buffer because it takes more than 1 second for me to type M-x fundamental-mode, so I just did cider-quit

13:08 * nullptr nods

13:09 nullptr: though, cider updates approximately hourly and sometimes just fetching all the latest bits fixes weird behavior for a while

13:10 * gfredericks makes his semiweekly threat to figure out how to use maven for emacs packages

13:10 arrdem: isn't that just doing a git local install and picking a tag?

13:11 gfredericks: git submodules probably?

13:12 technomancy: gfredericks: el-get

13:14 gfredericks: technomancy: lets me fix versions?

13:14 technomancy: gfredericks: yeap

13:14 gfredericks: also supports TLS

13:15 gfredericks: "The Latest Soup"

13:15 Leftover

13:15 Sandwich

13:15 the lettuce salad

13:16 technomancy: I will look into this thamks

13:16 (inc technomancy)

13:16 lazybot: ⇒ 144

13:16 gfredericks: square!

13:16 technomancy: gfredericks: it's not quite as polished, but it puts the user back in control

13:23 llasram: (inc technomancy)

13:23 lazybot: ⇒ 145

13:23 llasram: technomancy's too hip to be square

13:29 dwysocki: could somebody explain to me the limitations of Engelberg's instaparse's total parse mode?

13:29 I'm using it to parse a mini-java program, and I inserted an extra semicolon in the middle of the program, expecting it to parse up to there, and insert a :failure node

13:30 instead it returns only a :failure node, containing the entire program's text

13:30 so somehow I'm mis-interpreting the way it works

13:33 arrdem: dwysocki: by default instaparse will do total parsing, you need to enable partial parsing probaboy

13:33 *probably

13:38 gfredericks: test.check double generator https://github.com/gfredericks/test.chuck/commit/d330aadc335d3910de524d35d2e9a10ac4441002

13:39 reiddraper: gfredericks: woot

13:42 justin_smith: arrdem: http://noisesmith.org/ thanks for the advice

13:42 clgv: gfredericks: Chuck Norris counted to infinity. Twice!

13:42 arrdem: justin_smith: hope it works out for you

13:43 justin_smith: was .com taken? you could have gotten .us and been cool like technomancy :P

13:43 justin_smith: heh

13:43 * technomancy shakes his fist at the technomancy.org guy

13:43 justin_smith: arrdem: godaddy parked - and I plan to use it for my open source stuff, so org seemed more apropriate anyway

13:44 dwysocki: arrdem: I thought the point of total parses was to parse up to the point of failure, and insert a failure node containing whatever wasn't parsed

13:44 mping: hi

13:44 arrdem: dwysocki: no a total parse is all or nothing

13:44 technomancy: actually I had people think I created http://www.technomancy.org/google-suggest-venn which was actually kind of cool

13:44 arrdem: dwysocki: AFAIK

13:45 dwysocki: hacking on something else and not looking at the docs atm

13:45 dwysocki: arrdem: that seems counter to the example in the readme

13:45 arrdem: dwysocki: then ignore me

13:45 dwysocki: arrdem: lol ok

13:45 arrdem: dwysocki: :P

13:45 dwysocki: arrdem: thanks for helping anyway

13:46 I'm basically trying to parse up to the point of failure, and then, depending on context, parse what follows the failure with a different start rule

13:46 I thought total parse mode was the way to do it, but it doesn't seem to work as I thought :\

13:48 nwjsmith: Java Interop question: If I add a type hint ^Object for a call to .hashCode, is it problematic if the "actual" callee has it's own implementation of hashCode?

13:48 clgv: nwjsmith: type hint "^Object" is worthless

13:49 Bronsa: ,(defrecord x [] Object (hashCode [_] 1))

13:49 clojurebot: #<CompilerException java.lang.ClassFormatError: Duplicate method name&signature in class file sandbox/x, compiling:(NO_SOURCE_PATH:0:0)>

13:50 Bronsa: ,(deftype x [] Object (hashCode [_] 1))

13:50 clojurebot: sandbox.x

13:50 Bronsa: ,(.hashCode ^Object (x.))

13:50 clojurebot: 1

13:50 Bronsa: nwjsmith: ^ no issue

13:50 clgv: nwjsmith: methods on the JVM are virtual by default so you can not select a method by casting

13:51 dwysocki: does anybody know of any projects that actually use instaparse?

13:51 the documentation is nice, but I'd like to see some real examples

13:51 nwjsmith: clvv: k, so it's unreasonable to assume that I can ditch the reflection warnings for a function calling 'hashCode' on an unknown (at compile time) type

13:52 arrdem: dwysocki: https://github.com/arrdem/dogeon :P

13:52 dwysocki: lol

13:52 * arrdem no shame

13:52 nwjsmith: clgv: on reading my question, I'm answering my own question

13:52 dwysocki: arrdem: I'll look at it anyway

13:53 arrdem: dwysocki: how kind if you :P

13:53 dwysocki: oh, that reminds me

13:53 is there anything better than slurp for inputting the code to parse?

13:54 seems like there should be a lazier way to do it

13:59 bwreilly: there isn't a destructuring syntax for simply dumping all the keys in a map, is there?

13:59 something like

13:59 (let [{:keys :all} {:a "welp"}] a) => "welp"

13:59 arrdem: bwreilly: how could there be?

13:59 bwreilly: local bindings are static, a map is arbitrary

13:59 gfredericks: eval!

14:00 arrdem: gfredericks: shhhhh

14:01 gfredericks: also, walk the code and replaced all unresolved symbols with map lookups

14:02 which has a laughable amount of edge cases

14:02 so...do it

14:02 bwreilly: haha

14:02 gfredericks: method-missing.clj

14:02 arrdem: o gawd wai

14:06 gfredericks: WARNING: CIDER's version (0.7.0) does not match cider-nrepl's version ("0.7.0")

14:06 I think that's due to my whidbey problems actually

14:07 justin_smith: is there a "cider errors" twitter parody account yet?

14:07 gfredericks: justin_smith: I think it's comparing "0.7.0" to "\"0.7.0\""

14:07 justin_smith: but why would either version string have quotes in it?

14:08 gfredericks: mine does because whidbey

14:08 is malfunctioning

14:08 and double quoting everything

14:09 everything in my repl gets printed as if wrapped in pr-str

14:09 justin_smith: interesting

14:09 gfredericks: because whidbeys' middleware is pprinting and the builtin pr-str middleware is still there too

14:11 there's some haxy side-effects at the top level of the whidbey code but for some reason I haven't dug into yet that's sufficient for me

14:11 probably *any other* middleware mentions pr-values and that's all it takes

14:15 dwysocki: from what I've found on github and google

14:15 it doesn't look like anybody has tried writing a very big parser in instaparse

14:16 just nice little languages

14:16 does anybody think I should just switch to clj-antlr if I'm trying to a rather complicated parser?

14:16 pjstadig: https://github.com/pjstadig/clojure/commit/d8b5178f3836d5ff5ec19d3eda7aa7f79fceecee

14:17 dwysocki: the biggest issue I'm seeing right now with instaparse is its error reporting

14:17 it gives you one error, and ends

14:17 I need error recovery, which antlr supports

14:20 arrdem: pjstadig: nice!

14:20 pjstadig: arrdem: i'm offering it to anyone who wants to push it through the clojure contribution process

14:21 * arrdem mutters darkly

14:22 pjstadig: i've signed the CA so there shouldn't be any problem there (not that it's that much code anyway)

14:22 stuartsierra: pjstadig: looks relatively easy. Have you submitted a patch on JIRA?

14:23 tvanhens: anyone been able to read config variables in project.clj during the build phase of heroku?

14:23 technomancy: tvanhens: I would highly recommend against that

14:23 there are ways you can do it, but it's almost always a bad idea

14:23 puredanger: pjstadig: please submit!

14:23 tvanhens: how do you configure creds for things like s3 wagon?

14:23 technomancy: tvanhens: ah, yes credentials are a special case where it's OK

14:24 tvanhens: yeah its only for creds

14:24 for s3 wagon and datomic

14:24 technomancy: tvanhens: LEIN_USERNAME, LEIN_PASSWORD, and LEIN_PASSPHRASE are propagated automatically

14:24 for other things, see https://github.com/heroku/heroku-buildpack-clojure#configuration

14:25 tvanhens: awesome, thanks :)

14:31 edbond: why nil here? (async/<!! (async/onto-chan (chan 100) ["ABC" "DEF"]))

14:31 ,(async/<!! (async/onto-chan (chan 100) ["ABC" "DEF"]))

14:31 clojurebot: #<CompilerException java.lang.RuntimeException: No such namespace: async, compiling:(NO_SOURCE_PATH:0:0)>

14:32 tvanhens: technomancy, do you know of any examples where BUILD_CONFIG_WHITELIST is used? I tried to add my keys space seperated to that var and re-pushed but it lead to a weird file not found error

14:35 edbond: hm, (async/<!! (async/to-chan ["ABC" "DEF"])) works fine

14:41 drorbemet1: What is the line to start cider repl using emacs lisp (cider-jack-in "~/clojureproject") gives me a promt for a path because it is interactive

14:42 edbond: problem solved, onto-chan doesn't return chan ^_^

14:43 stompyj: What does #’ mean in clojure?

14:43 I’m not familiar with it, and it’s hard to search

14:44 jeremyheiler: stompyj: var

14:44 justin_smith: ,'#'+

14:44 clojurebot: (var +)

14:44 stompyj: aha

14:44 thanks!

14:44 edbond: stompyj, http://yobriefca.se/blog/2014/05/19/the-weird-and-wonderful-characters-of-clojure/

14:45 stompyj: danke

14:45 drorbemet1: I would like to start cider repl using emacs lisp on startup of emacs

14:45 jeremyheiler: stompyj: also http://clojure.org/reader

14:45 dthurn: q: are there versions of 'conj' that default to different types, like say a set, when the first arg is nil?

14:46 jeremyheiler: dthurn: you can use fnil to make one

14:46 pjstadig: stuartsierra: puredanger: http://dev.clojure.org/jira/browse/CLJ-1561 i guess

14:46 puredanger: thx

14:46 jeremyheiler: ,((fnil conj #{}) nil 1)

14:46 clojurebot: #{1}

14:47 dthurn: cool, thanks

14:47 justin_smith: very useful when used inside update-in

14:48 stuartsierra: dthurn: Note: if this is used in a performance-sensitive loop, define the `fnil` function once outside the loop.

14:49 I recall someone wrote a test.check extension for describing systems in terms of actions, anybody remember the name?

15:03 amalloy: bwreilly: http://stackoverflow.com/questions/9345056/in-clojure-how-to-destructure-all-the-keys-of-a-map

15:04 bwreilly: amalloy: thanks

15:04 arrdem: we don't have a "defmemfn", do we?

15:05 for "mem" being "memoize" not "member"

15:06 bwreilly: really a curiosity, I've been soured on the more elaborate means of destructuring recently around readability/visibility issues

15:11 puredanger: we released transducers for Java, JavaScript, and Ruby today if anyone is interested http://cognitect-labs.github.io/

15:11 you know, for those "other" languages :)

15:12 arrdem: puredanger: when does the Haskell edition come out?

15:12 puredanger: no plans for that one :) left for the reader….

15:13 arrdem: puredanger: so why have a custom function type that happens not to be clojure.lang.Fn?

15:13 * arrdem idly curious

15:14 puredanger: so it's not tied to Clojure?

15:14 re Haskell… https://twitter.com/bodil/status/513031949586141184 :)

15:15 arrdem: lols

15:15 puredanger: he did actually say that too :)

15:16 mdeboard: puredanger: you work at cognitect too?

15:16 puredanger: yes

15:16 grim_radical: "Because transducers-java may be incorporated into products or client projects, we prefer to do development internally and do not accept pull requests or patches."

15:16 :/

15:16 mdeboard: live by the sword

15:17 llasram: And Cognitect's policy toward open source software remains as insane as ever

15:20 dysfun: :(

15:22 amalloy: arrdem: no, there is no defmemfn. useful does actually have one, although i don't really recommend using it: you don't usually want your memoization cache permanently locked up in a var

15:23 it's better to memoize functions within some kind of scope so you can free them when you're done

15:24 arrdem: amalloy: yeah exactly that's bitten me before

15:24 *that has

15:27 amalloy: $mail danielcompton i released useful 0.11.3 for you

15:27 lazybot: Message saved.

16:02 technomancy: llasram: "technically open source"

16:06 pjstadig: technomancy: llasram: "transparent source"

16:06 llasram: (inc pjstadig)

16:06 lazybot: ⇒ 7

16:39 lvh: is there a particularly good lein template for doing om (cljs) + sente (clj srver side) development?

16:39 seems like you always have to set one or the other up manually

16:42 amalloy: "transparent source". i like that description

16:44 gfredericks: it's at least forkable right?

17:16 puredanger: bbloom: btw I did see your message above re portable catch all and will try to bring that out more on the page

17:17 bbloom: puredanger: cool thanks

17:18 brianwong: hey guys, i have a question about mocking and with-redefs-fn. I am trying to test a namespace, but i am trying to mock out a function that that namespace uses, not a function defined in the namespace being tested.

17:18 hopefully this example helps

17:18 https://www.refheap.com/91577

17:18 bbloom: puredanger: those damn js semantics, huh?

17:18 dogonthehorizon: Hey folks, I'm rolling my own DSL and I'd like to limit the use of certain forms to specific parent forms. e.g. I would only like to realize send-test if it's parent form is method-call: (method-call (send-test "something")). Is this possible via macros or some other language construct currently in core?

17:19 justin_smith: brianwong: yeah, that's what with-redefs is for

17:19 (doc with-redefs)

17:19 clojurebot: "([bindings & body]); binding => var-symbol temp-value-expr Temporarily redefines Vars while executing the body. The temp-value-exprs will be evaluated and each resulting value will replace in parallel the root value of its Var. After the body is executed, the root values of all the Vars will be set back to their old values. These temporary changes will be visible in all threads. Useful for mocking out functions durin

17:20 jeremyheiler: brianwong: you just need a reference to teh var that holds that function

17:20 justin_smith: jeremyheiler: with with-redefs you don't even need that

17:20 brianwong: all the examples seem to mock out functions defined in the namespace being tested

17:21 not functions that the namespace is calling out

17:21 justin_smith: brianwong: doesn't matter, that is arbitrary

17:21 brianwong: so what would the symbol be in my case?

17:21 ec2/run-instances?

17:21 justin_smith: yes

17:22 brianwong: arg

17:22 argh

17:22 i had it, i just forgot the #'

17:22 thanks. it worked

17:22 for with-redefs-fn

17:23 justin_smith: ,(doc with-redefs-fn)

17:23 clojurebot: "([binding-map func]); Temporarily redefines Vars during a call to func. Each val of binding-map will replace the root value of its key which must be a Var. After func is called with no args, the root values of all the Vars will be set back to their old values. These temporary changes will be visible in all threads. Useful for mocking out functions during testing."

17:23 justin_smith: why not just use with-redefs?

17:23 brianwong: im trying to read the difference right now

17:23 thanks

17:25 justin_smith: another tactic is to create a function that generates the arguments to ec2/run-instances, or takes ec2/run-instances as an argument. That way during testing you can verify that the correct argument list is generated, or pass in a replacement function directly without having to do any redefs.

17:26 brianwong: look at the source of with-redefs - I think with-redefs-fn is just there to implement the with-redefs macro

17:26 $source with-redefs

17:26 lazybot: with-redefs is http://is.gd/rdnNKM

17:28 rurumate: suppose you have this json in the clipboard: {"foo":12}. you want to paste it into the repl, but it needs escaping. it's not enough to surround it with quotes, you also have to replace " with \". Is there a fast/easy way to do this?

17:28 justin_smith: rurumate: what editor?

17:29 rurumate: emacs

17:29 justin_smith: there is an auto-string-escape mode

17:29 rurumate: neat, I'll check it out

17:29 justin_smith: string-edit.el M-x string-edit-at-point

17:29 it even works recursively :)

17:30 Balveda: "No cljson encoding for '{"_id" #<ObjectId 543848e84e19b68fa401e6a3>, "applicant" "JG", "appid" 1, "yob" 1999, "answer" "KPO"}'."

17:30 justin_smith: I bind it to C-c e

17:30 Balveda: What's the deal with this?

17:30 justin_smith: Balveda: what json lib is that using?

17:30 never mind, found it, cljson uses cheshire

17:31 the "_id" key's value is something the cheshire does not know how to encode

17:32 you can add new encoders so it knows how to handle various classes: http://dakrone.github.io/cheshire/cheshire.generate.html#var-add-encoder

17:32 but the reader will need to know the same rules if you want to create the same object type, of course

17:37 amalloy: or, more likely, tell your mongodb client not to give you the _id

17:37 or just remove it yourself before encoding the object

17:37 justin_smith: this is also an option, yes

17:39 amalloy: i saw "yob" in that mongodb map and thought maybe there was a swedish person applying for a job: http://search.dilbert.com/comic/Yust%20Sven

17:41 pandeiro: is there a trick to lower compilation time for projects with timbre?

17:44 Balveda: I was thinking the same thing

17:44 Thanks

18:00 dweave: hi yall. new to clojure, haven’t used a lisp since school a number of years ago. have a new project at work and i’m trying to refresh myself a bit. Is there any reason not to use sublime text when working with clojure? Not seeing a lot of love for sublime + clojure around the interwebz. I do see that there are clojure plugins though.

18:01 bbloom: dweave: use whatever you're comfortable with

18:01 rurumate: dweave: sublime is proprietary software afaik? also emacs is quite good

18:01 bbloom: dweave: don't try to learn a language and an editor at the same time

18:02 dweave: thanks bbloom

18:02 thats what i thought, wasn’t going to try to pick up emacs, but was wondering if something like light table had some advantage

18:02 doesn’t seem that way really tho

18:02 rurumate: I learned clojure and emacs at once. it worked fine and made sense to me

18:03 bbloom: dweave: you'll want to always work with a repl open and ready. eventually you'll get sick of copy pasting code back and forth, then you can investigate other tools / plugins / whatever

18:04 rurumate: I think editor choice is quite important question for beginners. And some editors may be more frustrating than others.

18:05 dweave: why is clojure so editor sensitive i don’t really get that

18:05 not understanding what emacs really brings to the table for clojure that it wouldn’t do so for other languages

18:05 rurumate: because, a repl is less fun without autocomplete, syntax highlighting, and paredit

18:06 dweave: for instance sublime has a repl

18:06 ed-g: does anyone know how to change the characters which are allowed for a variable in a defroutes path? for instace (defroutes foo (ANY "/bar/:baz" [baz] (handler baz))), I would like for baz to be able to contain "," but by default that character is not OK

18:06 bbloom: dweave: clojure (and many/most other lisps) are designed to support editing your program while it is running

18:06 dweave: and your editor can support you in communicating with your running program

18:06 dweave: hmm i see

18:06 thanks i have to do some more reading it soudns like

18:07 rurumate: there's a video on youtube called clojure paredit speedrun or something, it helped me get started

18:14 amalloy: ed-g: iirc it's something like (ANY ["/bar/:baz" {:baz #"[\w,]+"}]-

18:15 catern: dweave: emacs is the best editor anyway, it's just more accepted in clojureland

18:15 dweave: ha everyone I know that uses it admits they secretly hate it

18:16 i’ve always been a vim guy

18:16 bbloom: dweave: there are many vim users in clojure land. tpope's vim-fireplace works great

18:16 dweave: i myself am a fireplace user

18:17 catern: dweave: i also have been a vim guy, but I switched to emacs with evil-mode, vim emulation, because it is superior

18:17 amalloy: dweave: the worst thing about emacs is that if you mention you use a different editor catern will tell you emacs is the only good one

18:17 bbloom: the editor war is totally not necessary here

18:18 dweave: use what you're familiar with. don't try to learn two things at once

18:18 amalloy: or: do try, because it's fun. but don't do it because you feel you have to

18:18 dweave: haha

18:18 i’ll figure it out

18:18 bbloom: amalloy: learning a new editor is my definition of "not very much fun"

18:18 :-)

18:19 amalloy: bbloom: i had fun learning them both at once

18:19 so i always have to object when advice is given to the effect that it's unconditionally bad

18:19 catern: amalloy: that's the best thing about emacs! then you don't have to hear me harp about it

18:19 after you switch

18:19 amalloy: catern: that's what /ignore is for

18:19 catern: :(

18:20 bbloom: amalloy: i'm not saying it's unconditionally bad, i'm saying that it's potentially bad and people generally come in here feeling obligated to learn emacs to write lisp of any kind

18:20 amalloy: but, if you are coming from vim, evil mode or not, your fingers will hate you if you try to learn anything else along side another editor :-P

18:20 amalloy: yeah, i bet

18:21 dweave: language agnostic: what is the benefit of emacs in evil mode

18:21 ed-g: amalloy: thanks!

18:21 amalloy: i opened up eclipse for the first time in months today, and kept hitting C-x C-s to save

18:21 dweave: seems like the bindings are the editor in the case of vim and emacs

18:21 catern: i'm glad people editor-warred around me, that way i switched to the superior editor

18:21 dweave: certainly not!

18:21 ed-g: for those in the studio audience, full route syntax docs at https://github.com/weavejester/clout

18:22 dweave: catern do go on, to me I use vim because of the key bindings mainly

18:22 and because it looks cool

18:22 catern: dweave: consider http://upsilon.cc/~zack/blog/posts/2008/11/from_Vim_to_Emacs_-_part_2/ and, one second, eh, my comment here: http://www.reddit.com/r/vim/comments/2coqzs/how_do_you_work_with_vim/cjk7ndo

18:22 amalloy: vim has some really advanced editing capabilities

18:22 dweave: emacs doens’t look cool

18:22 technomancy: dweave: the command language is the main strength of vim, and the programming language is the main strength of emacs.

18:22 justin_smith: dweave: emacs has a number of extensions and features - many use emacs specifically because of things like org-mode or the interactive process integration

18:22 amalloy: (inc technomancy)

18:22 lazybot: ⇒ 146

18:22 technomancy: you may notice these two things are not mutually exclusive

18:23 catern: I really need to adapt that comment into a blogpost

18:23 justin_smith: dweave: I don't know of anyone who uses emacs because of the key bindings (unless it's just laziness "those are the bindings I know")

18:23 catern: even if it is wrong in some ways, I hate linking to it

18:23 johnwalker: ,(clojure.java.browse/browse-url "http://google.com")

18:23 clojurebot: #<ClassNotFoundException java.lang.ClassNotFoundException: clojure.java.browse>

18:23 johnwalker: darn

18:23 catern: justin_smith: well emacs key bindings are default or available many places

18:23 justin_smith: sure- but I don't actually like them

18:23 vi bindings are better

18:23 amalloy: "how do you work in vim?" "here is why i love emacs"???

18:23 catern: agreed

18:23 lazybot: amalloy: Yes, 100% for sure.

18:24 tjd: johnwalker: you tried.

18:24 amalloy: $google google

18:24 lazybot: [Google] http://www.google.com/

18:24 justin_smith: $google google google

18:24 lazybot: [Google] http://www.google.com/

18:24 catern: dweave: and emacs does look cool, hmm... I guess these are some good screenshots http://tuhdo.github.io/emacs-tutor.html

18:24 johnwalker: i did try.

18:25 justin_smith: catern: woah, someone who actually turns on nyan-mode

18:26 catern: i know someone who leaves it on, it's kinda amusing

18:26 justin_smith: my modeline tends to get pretty crowded, if I could have nyan on it's own line at the top I would totally go for it

18:27 dweave: anyway, as an emacs user, I'll suggest not using emacs unless you need the features, and if you need the features, something like evil is helpful because you get the keybindings you know / like.

18:34 catern: yes, emacs is something you grow up to

18:49 vIkSiT: hello all

18:49 has anyone here used secretary routes with compojure/luminus?

19:12 qbg: Has anyone adapted core.async to produce a transducer instead?

19:14 go! to be specific

19:25 dopamean_: hey #clojure. im pretty new to the language and im trying to wrap my head around promises. i wrote a small library that makes some http requests and it was suggested to me that the functions making the requests return a promise but based on the basic stuff ive read about promises im not sure i understand how to do that

19:25 it seems that id need one function to return a promise and another to deliver the value?

19:26 i also dont understand what happens if the http request fails

19:27 qbg: You could deliver some sort of error response

19:29 dopamean_: the people suggesting returning the promise seemed to imply that this would somehow avoid exceptions

19:36 qbg: I'm not seeing how

19:36 dopamean_: nor am i

19:37 but i dont really understand what a promise adds here anyway

19:37 so i dont really understand the suggestion at all

19:38 qbg: The code that makes the request doesn't have to block until it actually needs the result if you use a promise

19:41 dopamean_: im not really sure what that means. i think i need to look at more examples.

19:42 justin_smith: dopamean_: often a future is more applicable - unless you have some other custom worker pool you want to execute the task

19:43 technomancy: dopamean_: I'd stick with regular HTTP requests till you get more familiar with the language. it's difficult to see what a promise gets you outside very specific use cases that the client could simply wrap themselves.

19:43 justin_smith: also futures are simple to use, and raise any error that may have come up when you dereference them (ie. try in the repl (def a (future (/ 1 0))) ... wait a moment ... @a (now you get the exception returned)

19:44 )

19:45 dopamean_: @technomancy - right now if the request is successful my function returns a map with certain values. do you think it would make sense to catch the exception if the request fails and have it return a map with some sort of error value?

19:45 or would it be better to throw the exception

19:46 technomancy: dopamean_: returning an error value only makes sense if you have an option type

19:47 since clojure doesn't ship with pattern matching out of the box, that style is pretty rare

19:47 error values that aren't option types are the worst

19:47 dopamean_: good to know. i kind of figured as much and wasnt particularly fond of that idea

19:48 amalloy: just return Double/NaN whenever something goes wrong. it will never fail to surprise the user

19:48 technomancy: yeah, this isn't Google Go; we have a civilized approach to error handling =)

19:48 amalloy notwithstanding

19:48 dopamean_: lol

19:48 amalloy: clojurebot: amalloy |has| a sadistic approach to error handling

19:48 clojurebot: Ik begrijp

19:49 qbg: (System/exit 0)

19:49 *really* surprise the user :p

19:49 technomancy: ~guards

19:49 clojurebot: SEIZE HIM!

19:50 dopamean_: thanks for the advice guys. i really appreciate it.

19:51 technomancy: np

20:41 ed-g: Hello, I'm attempting to extend a protocol related to jdbc, so that I can pull PostgreSQL arrays as clojure vectors. I don't know how to access the relevent protocol. An example from the Internet has (extend-protocol jdbc/IResultSetReadColumn ...) however I don't have jdbc loaded; I'm using Postgres from yesql. Should I import jdbc directly from Java?

20:44 This is the article I'm referencing: http://hiim.tv/clojure/2014/05/15/clojure-postgres-json/ only I want to use (seq (.toArray *postgres-array-val*)) instead of json/read-str

20:45 dnolen_: loving the Chestnut http://swannodette.github.io/2014/10/10/magic/

20:51 nullptr: dnolen_: neat -- looks like it was updating from a save hook? you didn't have to eval your changes?

20:54 dnolen_: nullptr: yes, figwheel updates namespaces when you save - you can use defonce to protect app state

20:54 nullptr: i see ... seems like something that would start to lag as your app gets huge, no?

20:56 justin_smith: ed-g: yesql is pulling in clojure.java.jdbc, you can use the jdbc classes

20:56 ed-g: justin_smith: thanks! :-)

20:57 dnolen_: nullptr: I don't see how

20:57 nullptr: but you might not be familiar with the Om rendering model

20:57 nullptr: browsers can eval tens of thousands of lines of JS at ridiculous speeds

20:58 nullptr: i'm thinking more of the cljs->js translation, not the browser runtime (yes, familiar with om model)

20:59 dnolen_: nullptr: I don't follow why would you use cljs->js outside of interop?

20:59 nullptr: and even then you should just have the data in the right format

21:00 cljs->js is a sign of not accepting the truth of interop

21:01 nullptr: if i'm not using chestnut, i'm using lein cljsbuild, which is doing cljs->js each time i save

21:01 dnolen_: nullptr: huh? no

21:01 nullptr: chestnut uses fighwhell which just calls the cljs compiler, same as cjlsbiuld

21:01 cljsbuild

21:02 cljs->js is a helper for converting data structures - has nothing to w/ compile time generation of JS

21:02 justin_smith: are you talking about the cljs->js function, or about the compilation process?

21:02 nullptr: oh, heh

21:02 yes, compilation process -- that explains the confusion

21:02 dnolen_: nullptr: figwheel compiles only the namespaces that change, no problems here

21:02 nullptr: gotcha

21:03 dnolen_: nullptr: since a file maps to namespace, this should never be an issue even if your project is massive

21:03 * nullptr nods

21:03 nullptr: that's really nice

21:03 ed-g: ok, followup question. I have arrays working, now I would like to support an array of hstore. The error I get is "java.sql.SQLFeatureNotSupportedException Method org.postgresql.jdbc4.Jdbc4Array.getArrayImpl(long,int,Map) is not yet implemented.". How can I add this feature to jdbc?

21:04 nullptr: i am pretty familiar with gigantic js codebases, and as much as i enjoy working with cljs/om i have a vague fear of getting to a huge codebase and losing save/refresh workflow because the compilation time exceeds a certain threshold

21:04 but if it's smart about what actually gets translated that shouldn't be an issue in practice

21:04 justin_smith: ed-g: first thing I would do is see if there is a newer org.postgresql version that does implement that feature

21:05 ed-g: failing that, I guess look up the docs for extending a jdbc driver? It will explain everything in a java way of course.

21:05 dnolen_: nullptr: large code bases definitely affect first compile (I'm working on fixing that) but incremental compilation has worked for a while now

21:06 ed-g: justin_smith: maybe I should just return an array of EDN string which can be parsed into a map.

21:06 justin_smith: it seems like that would be much easier

21:06 nullptr: yeah, first compile doesn't bother me at all as long as incremental is good -- and cljsbuild has been plenty fast for everything i've done so far

21:09 justin_smith: TIL postgres has built in geometric datatypes http://jdbc.postgresql.org/documentation/head/geometric.html

21:46 verma: technomancy, sorry off topic, but what's the quicking way to turn my wired kb to a bluetooth/wireless one?

22:07 amalloy: scissors

22:17 justin_smith: stay tuned for the next episode, where we introduce the qualifying adjective "working"

22:26 technomancy: verma: I've never really looked into BT

22:31 not really keen on the idea of keeping yet another battery charged

22:48 verma: technomancy, wireless poker2 would be nice though, take it around with me, the wire always seems like a hassle

22:53 numberten: ,(clojure.string/trim-newline "test\n")

22:53 clojurebot: "test"

23:24 kenrestivo: verma: http://handheldsci.com/kb ?

23:44 amalloy: if you're carrying a whole keyboard with you, i don't imagine a wire would be a big deal

23:45 spend like two seconds wrapping it around the keyboard when you pick it up

23:46 TEttinger: amalloy, for things like android phones/tablets with BT but no plugs, maybe

23:47 amalloy: TEttinger: indeed, that sounds plausible. i was objecting to "the wire always seems like a hassle"

23:47 TEttinger: I was DCed for a while

23:55 verma: kenrestivo, nice, but it says " Currently the adapter does NOT work with Mac. Let us know if Mac support is really important to you.", could have totally pulled it out of that case though

Logging service provided by n01se.net