--- Log opened Thu Mar 27 00:00:02 2014
00:49 < rowbit1> Hourly usage stats: [developer: 0, free: 18]
01:49 < rowbit1> Hourly usage stats: [developer: 1, free: 30]
01:55 < thlorenz> defunctzombie: ah just noticed prova isn't mapping stack traces: https://i.cloudup.com/nA08e0s60b.png that looks like a bundle to me
01:55 < thlorenz> so not as nice as I thought I guess ;)
01:55 < defunctzombie> heh
01:55 < defunctzombie> well, they could add that feature just like we did
01:55 < defunctzombie> or you can make zuul look prettier if it is that important :)
01:56 < defunctzombie> for me testing in browsers via sauce is more important
01:56 < defunctzombie> which is the primary use for zuul, not just manual in browser stuff
02:08 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.57.171.229(dev-ie6-2)
02:12 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.27.70(dev-ie6-1)
02:49 < rowbit1> Hourly usage stats: [developer: 0, free: 22]
02:51 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.32.168(dev4)
03:02 < mikolalysenko> what is the current status of weakmaps in node/v8?
03:39 < mikolalysenko> I wonder if it would be useful to create a module for preprocessing json objects to support O(1) least common ancestor and upper parent queries?
03:39 < mikolalysenko> maybe it could be helpful working with esprima output
03:49 < rowbit1> Daily usage stats: [developer: 85, free: 896]
03:49 < rowbit1> Hourly usage stats: [developer: 0, free: 34]
04:49 < rowbit1> Hourly usage stats: [developer: 0, free: 42]
05:49 < rowbit1> Hourly usage stats: [developer: 0, free: 34]
06:49 < rowbit1> Hourly usage stats: [developer: 0, free: 21]
07:06 < jden> hello node scientists: offhand, does anyone knows the module that does array -> readstream
07:11 < substack> jden: var r = new Readable; r._read = function () {}; array.forEach(function (x) { r.push(x) })
07:12 < jden> this is strictly speaking more characters than `require('array-streamify')`
07:14 < jden> but ty <3
07:49 < rowbit1> Hourly usage stats: [developer: 0, free: 23]
08:09 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.57.171.229(dev-ie6-2)
08:13 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.27.70(dev-ie6-1)
08:49 < rowbit1> Hourly usage stats: [developer: 0, free: 44]
08:51 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.32.168(dev4)
09:49 < rowbit1> Hourly usage stats: [developer: 4, free: 53]
10:49 < rowbit1> Hourly usage stats: [developer: 1, free: 32]
11:49 < rowbit1> Hourly usage stats: [developer: 0, free: 73]
12:49 < rowbit1> Hourly usage stats: [developer: 0, free: 37]
13:49 < rowbit1> Hourly usage stats: [developer: 0, free: 34]
14:09 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.57.171.229(dev-ie6-2)
14:13 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.27.70(dev-ie6-1)
14:49 < rowbit1> Hourly usage stats: [developer: 6, free: 50]
14:51 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.32.168(dev4)
15:01 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
15:04 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
15:49 < rowbit1> Hourly usage stats: [developer: 24, free: 48]
16:48 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
16:49 < rowbit1> Hourly usage stats: [developer: 1, free: 55]
17:21 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
17:29 < jez0990> mikolalysenko: "least common ancestor and upper parent queries" +1, it does sound useful! I would love to know if there's an optimal way to store DAGs in json
17:37 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/7.0
17:49 < rowbit1> Hourly usage stats: [developer: 51, free: 49]
17:59 < mikolalysenko> jez0990: you can do them in O(1) time actually
18:00 < mikolalysenko> too bad we have to wait til es7
18:01 < mikolalysenko> I think if we get it though, js could be as fast as c++
18:01 < mikolalysenko> err wrong channel
18:03 < pfraze> mikolalysenko, what were you referring to?
18:03 < mikolalysenko> oh, value types for js
18:03 < mikolalysenko> was in #bbg
18:04 < mikolalysenko> this is a thing I've been kind of obsessed with for some time now
18:04 < mikolalysenko> they were originally proposed for es6, but got axed and moved up to es7
18:04 < pfraze> this is the int32 kind of stuff?
18:04 < mikolalysenko> yeah, but you could make compound objects
18:04 < mikolalysenko> and the compiler/run time would have explicit support for them
18:05 < mikolalysenko> so you could pass them around with doing allocations/dereferencing
18:05 < pfraze> interesting
18:05 < mikolalysenko> also you get the possibility to control the layout of objects in memory, allowing for algorithms that actually scale wrt cache
18:06 < mikolalysenko> which is kind of a big deal, and the main reason that js/java is slow
18:06 < pfraze> huh, can you suggest some reading on that?
18:06 < mikolalysenko> about what? value types or cache efficient algorithms?
18:07 < pfraze> either, but cache-effiency in particular
18:07 < mikolalysenko> here is a pretty accessible intro lecture on it:  http://courses.csail.mit.edu/6.851/spring14/lectures/L07.html
18:07 < mikolalysenko> and here is a kind of fuzzy historical overview: http://courses.csail.mit.edu/6.851/spring14/lectures/L22.html
18:07 < mikolalysenko> unfortunately you can't implement that stuff in js
18:08 < mikolalysenko> well, actually you kind of can
18:08 < mikolalysenko> using typedarrays
18:08 < mikolalysenko> but it is very complicated and not fun
18:08 < mikolalysenko> and also you lose a lot on unboxing/datatype conversion
18:08 < mikolalysenko> and it is hard to do in a generic way
18:09 < mikolalysenko> basically java (and js) were defined in an era when the dominant thinking about computer perf was in terms of something called a "pointer machine"
18:09 < mikolalysenko> but as computers got faster, and caches got bigger, we've had to adopt a more realistic way of thinking about programs
18:09 < mikolalysenko> which is the "hierarchical memory model"
18:10 < mikolalysenko> the hmm can implement pm algorithms, and vice-versa the pm algorithms are not efficient when translated to hmm
18:10 < mikolalysenko> since they don't scale with respect to cache parameters
18:10 < mikolalysenko> so asymptotically they are slower
18:11 < pfraze> is that related to the "cache-oblivous alg" concept?
18:11 < mikolalysenko> yeah
18:11 < mikolalysenko> cache oblivious = programming in hierarchical memory model without knowledge of specific parameters for cache
18:11 < mikolalysenko> practically though you write them in the same language
18:11 < mikolalysenko> which is actually an older model than the pointer machine!
18:12 < mikolalysenko> it is called the word ram model
18:12 < mikolalysenko> which is the model that C uses for programming
18:12 < mikolalysenko> so C programs can be interpreted in either the word ram model, the hmm model, or the cache oblivious model
18:12 < mikolalysenko> java programs only work in pointer machine
18:12 < mikolalysenko> so it is less expressive
18:12 < mikolalysenko> and ultimately less efficient
18:13 < mikolalysenko> since you can't implement some optimizations that would otherwise be possible in a word ram language, like C
18:13 < mikolalysenko> this is not a kind of constant factor optimization either
18:13 < mikolalysenko> it is an *asymptotic* speed up
18:13 < mikolalysenko> ie, the C program will scale wrt cache, while the java program can't
18:14 < mikolalysenko> this is a limtation of pointer machine languages
18:14 < mikolalysenko> *but* if we add value types, then we regain the ability to write word ram style programs!
18:14 < mikolalysenko> and life is good again
18:14 < pfraze> and that's because the pointer model doesn't factor cache into its decision making?
18:14 < mikolalysenko> it has no concept of locality
18:14 < pfraze> yeah
18:15 < mikolalysenko> every memory operation is an indirect pointer access
18:15 < mikolalysenko> ie your memory is a collection of objects, and they have references to other objects
18:15 < mikolalysenko> it does not understand or even allow you to model their distribution in physical ram
18:15 < mikolalysenko> no such concept exists in that model
18:15 < pfraze> right
18:15 < mikolalysenko> with value types though, you *can* express that concept
18:16 < rowbit1> substack, pkrumins: Encoders down: 50.57.174.105 (dev-ie8-2)
18:16 < mikolalysenko> ie, an array of value types would be arranged contiguously in memory
18:16 < mikolalysenko> this is why things like buffers/typed arrays let you get much better performance in js
18:16 < mikolalysenko> since they let you take advantage of caching and locality
18:18 < pfraze> I see. So you can avoid indirection, I'm guessing you can also consider the size of your structure relative to the cache size. Would you also be able to make good decisions about keeping most of a structure in a specific cache and updating it in-place, or is that out of reach?
18:18 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
18:18 < mikolalysenko> the specific policy for caching turns out to not matter too much
18:18 < mikolalysenko> just using lru caches comes within a constant factor of optimality
18:19 < mikolalysenko> ie an lru cache of size 2m = optimal cache of size m
18:19 < mikolalysenko> the important thing is to realize that there is a cache in the first place though
18:19 < mikolalysenko> and that memory is transfered in blocks
18:20 < pfraze> ok, thanks for explaining that for me
18:20 < mikolalysenko> sure, the more people who understand this issue hopefully the more traction we can get in the language design
18:21 < mikolalysenko> and some day, (god willing) we will get value types in js
18:21 < pfraze> well you've got my interest
18:22 < pfraze> grabbing lunch o/
18:23 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
18:25 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
18:29 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
18:38 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
18:39 < rowbit1> substack, pkrumins: A developer is waiting in the queue for explorer/8.0
18:49 < rowbit1> Hourly usage stats: [developer: 29, free: 49]
18:51 < rowbit1> substack, pkrumins: Encoders down: 50.57.174.105 (dev-ie8-2)
18:53 < defunctzombie> thlorenz: we should investigate adding code coverage to zuul, that would be cool
18:59 < thlorenz> defunctzombie: yep - shouldn't be too hard right - just inject the libcov transform and then show results nicely?
18:59 < thlorenz> or coverify or whatever ify module it is ;)
18:59 < defunctzombie> I think so (but I have no idea)
18:59 < defunctzombie> I would like to use istanbul if possible
18:59 < defunctzombie> cause it already generates reports
19:00 < defunctzombie> and has a good coverage engine from what I can tell
19:00 < thlorenz> I can convince some peeps here that we need coverage and if there is time I can work at it
19:00 < defunctzombie> haha
19:00 < thlorenz> right now I'm totally swamped outside of work - lots of stuff going on and pending PRs
19:02 < defunctzombie> https://www.npmjs.org/package/istanbulify
19:02 < defunctzombie> this looks promising
19:02 < defunctzombie> tho I have no idea how to get the data out of it haha
19:03 < thlorenz> defunctzombie: cool
19:07 < rowbit1> substack, pkrumins: Encoders down: 50.57.174.105 (dev-ie8-2)
19:34 < rowbit1> substack, pkrumins: Encoders down: 184.106.99.27 (dev-ie7-3)
19:34 < rowbit1> substack, pkrumins: seaport services down: auth, tunnels
19:49 < rowbit1> substack, pkrumins: seaport services down: web.status
19:49 < rowbit1> Hourly usage stats: [developer: 5, free: 13]
20:10 < jesusabdullah> substack: How do you make form submit buttons do ajaxy shit instead of a typical POST, without jquery?
20:10 < jesusabdullah> substack: Now I *know* you're all about ditching jquery
20:15 < pkrumins> jesusabdullah: use the raw XMLHttpRequest object
20:15 < jesusabdullah> pkrumins: I don't mean making the request, I mean disabling the default submit behavior and hooking onto the click event
20:16 < pkrumins> yes that's exactly how you do it
20:16 < pkrumins> without jquery
20:16 < jesusabdullah> yeah but how? XD
20:16 < mikolalysenko> why use a form object?  just use a div
20:16 < jesusabdullah> Like idk the APIs
20:16 < chapel> why what mikolalysenko said
20:16 < pkrumins> jesusabdullah: i've a code sample right here
20:16 < chapel> just read the input elements directly
20:16 < jesusabdullah> pkrumins: most excellent
20:16 < chapel> no form
20:16 < mikolalysenko> document.querySelectorAll(...) blah
20:16 < pkrumins> jesusabdullah: except ci.testling.com is down right now
20:16 < chapel> and hook onto a button/link to do the submit using ajax
20:17 < mikolalysenko> and get all children of the div object, cat to json object, etc.
20:17 < pkrumins> wanted to link you to source part
20:17 < pkrumins> jesusabdullah: 1 sec
20:17 < pkrumins> jesusabdullah: i'll gist it
20:18 < jesusabdullah> pkrumins: sick
20:19 < pkrumins> jesusabdullah: so you do button.addEventListener('click', function () { ... }) for the submit button: https://gist.github.com/pkrumins/ded6bf4eb105eebf2bac#file-gistfile1-txt-L81
20:19 < pkrumins> jesusabdullah: then you crate a new XMLHttpResponse object and setup all the handlers https://gist.github.com/pkrumins/ded6bf4eb105eebf2bac#file-gistfile1-txt-L160
20:19 < jesusabdullah> pkrumins: Oh, baller
20:20 < jesusabdullah> pkrumins: I'm gonna use browser-request to make the actual call ^__^
20:20 < pkrumins> jesusabdullah: and then you build and send the x-www-form-urlencoded data for the POST request https://gist.github.com/pkrumins/ded6bf4eb105eebf2bac#file-gistfile1-txt-L196
20:20 < pkrumins> jesusabdullah: yeah that works
20:34 < jesusabdullah> pkrumins: is there a corresponding html doc for that? I'm herping the derp reading y'alls querySelector
20:36 < pkrumins> jesusabdullah: yes this part https://gist.github.com/pkrumins/ff08b334da7c88c4a6c1#file-gistfile1-txt-L62-L103
20:36 < jjjohhny> githubs stupid commit graph
20:40 < jesusabdullah> cool
20:40 < jesusabdullah> thanks a ton ^__^
20:49 < rowbit1> Hourly usage stats: [developer: 0, free: 61]
20:51 < rowbit1> substack, pkrumins: These encoders are STILL down: 50.56.32.168(dev4)
20:54 < ogd> anyone know a node module for turning a string of HTML into DOM elements? similar to $('<div></div>') in jQuery
20:55 < mikolalysenko> ogd: no, but I think it should be easy to do
20:56 < jesusabdullah> substack: pkrumins: Does browserify have issues browserifying json blobs? https://gist.github.com/jesusabdullah/9818460
20:56 < mikolalysenko> just create a div, set .innerHTML = str, and return all children
20:56 < jesusabdullah> Any ideas on how to make that happier? ^^
20:56 < substack> ogd: there's also domify
20:57 < jesusabdullah> If I get this, it means I can share my schema validations between my client and server p much seamlessly, which is hella win
21:00 < ogd> cool, domify worked http://requirebin.com/?gist=maxogden/9818569
21:00 < jesusabdullah> ogd: Maybe you know. Does browserify have issues with requiring json? https://gist.github.com/jesusabdullah/9818460
21:01 < jesusabdullah> That file appears to be there
21:01 < jesusabdullah> I guess I could get around it by vendoring
21:01 < ogd> jesusabdullah: dont think i've tried that, you probably need to do fs.readFileSync and use the brfs transform I would guess
21:03 < ogd> jesusabdullah: it appears browserify supports json though https://github.com/substack/node-browserify/tree/da64917a77d1e46d2c2061647508bcc4256b9acf/test/json
21:04 < mikolalysenko> I think require("blah.json") works
21:04 < mikolalysenko> I've used it multiple times anyway
21:09 < jesusabdullah> ogd: ugh I dun' wanna have to vendor joi
21:09 < jesusabdullah> ehhhhh I'll take a look around I guess
21:11 < jesusabdullah> Oh, no, I bet adding brfs will fix this
21:11 < jesusabdullah> looks like he's not requiring. Noob.
21:13 < jesusabdullah> WHAT THE CRAP uuuugh
21:13 < mikolalysenko> more fun with weakmaps!  https://github.com/mikolalysenko/least-common-ancestor
21:16 < jesusabdullah> cool
21:18 < mikolalysenko> I am beginning to really like that feature for building augmentations to data structures
21:18 < ogd> tree.baz.zardoz.potato ftw
21:21 < mikolalysenko> I think I will try level ancestors next
21:23 < pfraze> nice
21:46 < jjjohhny> wi(zard(of)oz)
21:47 < mikolalysenko> https://www.youtube.com/watch?v=kbGVIdA3dx0
21:49 < rowbit1> Hourly usage stats: [developer: 1, free: 47]
21:50 < mikolalysenko> and this:  http://godzillu.daa.jp/zardoz/
21:56 < mikolalysenko> that movie is both profoundly terrible and bizarre
21:57 < mikolalysenko> having seen it, I think it has damaged some part of my subconscious brain permanently, and no amount of alcohol can undo the damage
22:00 < jjjohhny> i think it is neutured, conceptually.  should have been MST3K'd.  But as an oddity in cinema, it stands on its own.
22:06 < mikolalysenko> are there any random json generator modules/projects on npm?
22:06 < mikolalysenko> would be useful for testing things I think
22:07 < mikolalysenko> or maybe a standard json object test suite
22:16 < navaru> I have a video stream coming from a service, content type: video/x-mpegts, I want to buffer it (like ~1mb) convert the buffer to mp4 and multi-cast it
22:16 < navaru> any suggestions what to use or how to approach this?
22:21 < jjjohhny> avconv for convert
22:21 < substack> mikolalysenko: faker
22:21 < jjjohhny> faker is awesome
22:22 < mikolalysenko> interesting
22:22 < mikolalysenko> can it generate random dynamic trees though?
22:22 < mikolalysenko> I am looking for something which generates totally arbitrary json objects
22:43 < mikolalysenko> lca/ancestor queries could be useful in parsing I think
22:44 < mikolalysenko> for example, maybe you visit a node and you want to get its great-great-great-grand parent
22:44 < mikolalysenko> which should be possible in O(1) using level ancestors
22:45 < mikolalysenko> or maybe you have two different nodes and you want to find the common block which contains them
22:49 < rowbit1> Hourly usage stats: [developer: 0, free: 31]
22:56 < pfraze> maybe something to do with search or classification. Compare a set of objects, treat each difference as a tree, treat each attribute as a leaf. Create an index of the leafs. Search against the leafs, then use lce to trace up the tree and discover... some commonality
23:48 < jjjohhny> gonna hash the fuck out of this work computer digging dogecoins
23:49 < rowbit1> Hourly usage stats: [developer: 1, free: 26]
--- Log closed Fri Mar 28 00:00:08 2014