Da
Show enters and exits. Hide enters and exits.
| 00:00:46 | antares | rue, I am jumping across the codebase because sometimes I need the idea of what this structure is like or what this function really does... |
| 00:01:35 | ezmobius_ enters the room. | |
| 00:02:41 | antares | rue, what IP_TYPE *data in rubinius_cpu for? usages are not clear unfortunately |
| 00:03:09 | _ADS_ leaves the room. | |
| 00:03:18 | ttmrichter_ enters the room. | |
| 00:03:36 | GMFlash leaves the room. | |
| 00:03:45 | GMFlash enters the room. | |
| 00:03:45 | rue | That is the bytecode being executed, I think |
| 00:04:14 | headius | what do you guys use to edit the C code, btw |
| 00:04:35 | antares | headius, Emacs with etags for navigation |
| 00:04:55 | antares | rue, look: #define IP_TYPE uint32_t — hm, I am still confused |
| 00:05:07 | antares | rue, bytecode itself or reference somewhere or what? :) |
| 00:05:23 | dctanner enters the room. | |
| 00:06:21 | rue | headius: Veem |
| 00:06:25 | headius | etags lets you jump to, for example, the definition of a function being called? |
| 00:06:53 | antares | headius, right. It is precise in most of cases |
| 00:07:27 | jessop enters the room. | |
| 00:07:30 | rue | I am fixing the hash issue in case brixen went out |
| 00:07:33 | _ezmobius_ enters the room. | |
| 00:08:10 | headius | antares: one of these days I'll have to try using etags |
| 00:08:27 | headius | I've been using netbeans C/C++ support, which allows jumping around as well |
| 00:08:38 | antares | headius, but of course it uses string matching without any type inference. Though if you know the codebase it is fine |
| 00:08:38 | headius | macro defs, method defs, structures, typedefs, headers |
| 00:08:43 | headius | ahh |
| 00:09:07 | headius | netbeans actually parses all the files and makes judgments about what to jump to based on the whole codebase |
| 00:09:25 | antares | headius, netbeans is better for navigation but this is typical IDE vs editor dilemma |
| 00:09:26 | headius | it will jump out into standard C headers as well |
| 00:09:46 | antares | headius, when you switch to another branch in git netbeans sometimes goes wild |
| 00:09:54 | dctanner leaves the room. | |
| 00:10:17 | antares | headius, I used netbeans for C. I just love Emacs but do not stick to a single tool. |
| 00:10:50 | headius | I have not used netbeans much with git, but I know it tends to freak out if I pull in a lot of changes |
| 00:10:56 | headius | usually recovers ok though |
| 00:11:44 | antares | headius, yeah it works fine most of the times. |
| 00:12:46 | antares | rue, ah is ip for instruction pointer? |
| 00:13:28 | rue | Yep |
| 00:13:34 | boyscout | 1 commit by Eero Saynatkari |
| 00:13:35 | boyscout | * Removed hash_rehash() that was causing the linker to fail. Build OK.; 1c77481 |
| 00:19:20 | rue | I guess I should say I /hope/ the build is OK :P |
| 00:22:10 | ttmrichter leaves the room. | |
| 00:22:45 | dctanner enters the room. | |
| 00:24:16 | rubuildius | Eero Saynatkari: 1c7748110; 4824 examples, 17537 expectations, 0 failures, 0 errors |
| 00:24:59 | headius_ enters the room. | |
| 00:28:39 | brixen | thanks rue |
| 00:29:03 | brixen | dynamic linking on osx |
| 00:31:21 | headius leaves the room. | |
| 00:33:36 | crossblaim enters the room. | |
| 00:36:08 | antares | rue, by the way what are initial plans for mod_rubinius? |
| 00:42:21 | rue | antares: I am trying to decide between two models still :) |
| 00:42:31 | dctanner leaves the room. | |
| 00:42:36 | rue | Might just plug in both |
| 00:42:40 | antares | rue, what are these? |
| 00:43:17 | _ezmobius | embedding the rubinius vm in the apache proces or having rubinius vm run in a separate managed process |
| 00:43:39 | rue | Might offer both, too, I suppose |
| 00:43:42 | oleganza enters the room. | |
| 00:43:45 | antares | ezmobius, I mostly meant feature set |
| 00:43:45 | oleganza | hi all |
| 00:43:56 | rue | Heya |
| 00:44:00 | antares | ezmobius, like things you discussed on ruby on rails podcast |
| 00:44:27 | oleganza | sorry, guys. mod_rubinius is for nginx or apache? |
| 00:44:35 | antares | ezmobius, separation of say merb and merb applications to reduce memory footprint |
| 00:44:36 | _ezmobius | hopefully both |
| 00:44:46 | oleganza | got it. thnx |
| 00:45:32 | Yurik enters the room. | |
| 00:46:40 | oleganza | ezra: i've recently came up with "nginxr": nice wrapper for nginx.conf in ruby DSL (on gitorious.org). I'm gonna try to create "yet-another" framework with controllers running in separate processes (tcp ports/unixsockets) |
| 00:47:00 | oleganza | i wonder, is it a smart idea? |
| 00:47:03 | rue | _ezmobius_: Not sure why wsgi dude thought putting everything in one 10k-line C file was a good idea |
| 00:47:52 | tarcieri | hahahaha |
| 00:48:06 | tarcieri | hates 10,000 line C files :( |
| 00:49:14 | headius | I'd be interested in wiring up a mod_ruby sort of thing with JRuby |
| 00:49:58 | headius | it basically works already for the most part...seems to me the biggest issue with such a model is that most frameworks don't play well with each other (or with themselves, as in Rails) |
| 00:50:17 | headius | rue: what models are you considering |
| 00:51:19 | crossblaim leaves the room. | |
| 00:52:11 | _ezmobius | rue: is that the apache mod_qsgi? |
| 00:52:16 | _ezmobius | wsgi? |
| 00:52:47 | _ezmobius | oleganza: what do you mean by yet aother framework? |
| 00:53:09 | rue | _ezmobius_: Yep |
| 00:53:22 | _ezmobius | lame |
| 00:54:13 | ragnard | hehe... i just had to check it out: http://modwsgi.googlecode.com/svn/trunk/mod_wsgi/mod_wsgi.c |
| 00:55:55 | oleganza | _ezmobius_: i mean i need some framework for controlling heterogenous webservers in a single "routes.rb" which is nginx.conf for me. I use rails, mongrel handlers and some homebrewed eventmachine webservers. I need some tool to manage them all and update nginx.conf appropriately. |
| 00:55:58 | rue | headius: In general, whether to run a daemon or embed into Apache. Think offering both is feasible. Then more specifically how to split up the threads and processes etc. Still trying to make up my mind, there are essentially two competing interests involved |
| 00:57:01 | oleganza | * well, mongrel handlers start up with rails mongrel, so it's not the case. |
| 00:57:09 | headius | yes, it's not as simple as with php and such that don't have as large a footprint to prime an application and so many conflicts between apps |
| 00:57:10 | oleganza | * but i might use merb for some activities |
| 00:57:36 | ragnard leaves the room. | |
| 00:57:43 | headius | in general I don't see how the embedded plugin gains you anything |
| 00:58:06 | _ezmobius | yeah i think separate pool pf managed processes is going to work much better then embedded |
| 00:58:15 | headius | not to mention making apache less stable in the process |
| 00:58:18 | _ezmobius | for rails/merb apps at least |
| 00:58:21 | headius | yeah |
| 00:58:53 | langenberg enters the room. | |
| 00:59:06 | headius | of course once you have that I don't know why existing proxy plugins for apache aren't enough |
| 00:59:26 | _ezmobius | with proxies you have to parse the http headers twice |
| 00:59:27 | headius | so you mostly would want to work on the "front daemon" that spins off requests into rubinius mvms |
| 00:59:49 | headius | _ezmobius_: I'd be absolutely stunned if that had any impact on a merb or rails app's overall performance |
| 00:59:57 | headius | but you're right |
| 00:59:59 | _ezmobius | basically my thoughts were when apache starts, it spanws a mod_rubinius master process, this master process can fork or spawn new vms form there |
| 01:00:16 | _ezmobius | headius: it does have a measurable impact |
| 01:00:50 | _ezmobius | nginx with fcgi runing rails/merb is 8-20% faster |
| 01:00:58 | _ezmobius | then nginx + proxy to mongrels |
| 01:01:03 | headius | that's largely how most large Java EE apps are fronted, with apache proxying to Java app servers |
| 01:01:10 | headius | and they can handle tens of thousands of requests |
| 01:02:02 | _ezmobius | i think one of the main goals for mod_rubinius is to make it dead easy to run ruby apps |
| 01:02:39 | _ezmobius | a one stop shop where you set it up in the apache or nginx conf and then it handles spawning/forking/monitoring of subprocesses |
| 01:02:43 | headius | well, let me play devils advocate for a moment...if we get the glassfish gem working perfectly, what more is needed |
| 01:02:52 | oleganza | what's the drawback of fastcgi? |
| 01:02:58 | _ezmobius | fcgi is not stable |
| 01:03:14 | oleganza | random 500? |
| 01:03:49 | headius | something similar to the glassfish gem ought to be mostly possible with rubinius |
| 01:04:00 | oleganza | i've experienced random 500 errors on a cheap dreamhost account. What about highload production server? |
| 01:04:10 | headius | more footwork to implement the pieces, but the same model would work |
| 01:04:19 | _ezmobius | yup |
| 01:05:03 | ezmobius leaves the room. | |
| 01:05:07 | headius | so I think part of what I'm getting at is whether this couldn't be a common piece of code that depends only on a standard MVM API |
| 01:05:31 | headius | a master VM in the process that manages a pool, is aware of the apps and their contexts and how to scale up/down |
| 01:06:00 | headius | if it were built on the MVM APIs, the same code would work as well for JRuby fronted by glassfish |
| 01:07:22 | antares | headius, what is standard MVM API, is there any or you just stress the need for standartization? |
| 01:07:31 | ezmobius | yeah how do you spawn a new vm in jruby? |
| 01:08:01 | headius | ruby = org.jruby.Ruby.new_instance |
| 01:08:06 | gnufied leaves the room. | |
| 01:08:18 | headius | it's all available through Java integration currently...anything implemented in Java you can call from Ruby, so it's just a matter of constructing the same Java objects |
| 01:08:31 | headius | but it would take a few minutes to implement the rubinius MVM APIs |
| 01:08:52 | ezmobius | or an abstracted api that delegated to the underlying vm impl |
| 01:09:23 | headius | yes, that's my hope...we're (JRuby) working with ko1 on MVM stuff but I had ko1 add evan to that list |
| 01:09:52 | headius | so I think we'll be able to come up with a standard API...and it seems like any hosting frontend should be based on that model |
| 01:10:31 | rue | headius: While there is a performance aspect, I would say the main reason for the module is easy access to Apache itself |
| 01:11:45 | headius | access for what purpose |
| 01:12:06 | headius | any access to apache would limit usefulness in nginx, no? |
| 01:13:58 | rue | Well, it would BE OF limitED usefulness in nginx :P |
| 01:14:54 | wycats_ enters the room. | |
| 01:14:54 | headius | so what does it gain you |
| 01:15:07 | headius | I'm curious |
| 01:15:15 | headius | putting any performance gain aside for now |
| 01:15:34 | rue | The daemon should pretty much be constant, any direct server access would come separately (but it needs a backend) |
| 01:15:51 | rue | Hooks, handlers etc. |
| 01:17:34 | headius | can you elaborate on that a bit? :) |
| 01:19:31 | ezmobius | if we use rack as the interface to rubyland then we get middleware, hooks, cascades and all that stuff for 'free' |
| 01:19:42 | wycats__ enters the room. | |
| 01:20:01 | headius | yeah, I could get behind something rack-like...we've already got a rack connector for Java app servers |
| 01:20:12 | rue | Yep, I think Rack is a fine modus outputti |
| 01:20:19 | ezmobius | that's the tentative plan, to base it on rack |
| 01:20:23 | ezmobius | which is based on wsgi |
| 01:22:25 | menator leaves the room. | |
| 01:22:52 | headius | well, I'd love to help with it |
| 01:23:06 | rue | headius: mod_perl, for example, can be used to write Apache modules |
| 01:23:13 | rue | Instead of doing it in C |
| 01:23:14 | headius | we're trying out several of these models for JRoR stuff and it would be nice to work on a common system |
| 01:23:24 | headius | rue: seems like a niche at best |
| 01:23:52 | bitbckt enters the room. | |
| 01:24:42 | rue | headius: Oh, I dunno.. we are sort of fans of using high-level languages for what is considered to be a low-level domain ;) |
| 01:24:52 | wycats___ enters the room. | |
| 01:25:12 | ezmobius | rue: i think for the first pass at least that deep apache integration is not a priority, especially if we just go rack all the way. you can pretty much do most things you woudl ever want to do in a ruby web app in a rack handler/adapter |
| 01:25:45 | rue | ezmobius: Oh, sure. It is just another reason to integrate through a module |
| 01:26:01 | rue | It does allow simplifying the config quite a bit too, obviously |
| 01:26:25 | ezmobius | true |
| 01:27:01 | headius | another reason to loosely tie apache and the daemon is that you might want only one or two apache machines handling static content and dispersal of requests across several backend daemon machines (or slices) that don't even run apache |
| 01:27:44 | headius | they can be brought up and down at will without affecting anything else |
| 01:28:01 | d2dchat leaves the room. | |
| 01:29:26 | langenberg leaves the room. | |
| 01:30:17 | wycats leaves the room. | |
| 01:30:44 | antares leaves the room. | |
| 01:31:30 | rue | Well, the daemon should not really care what (if anything) is driving it |
| 01:31:43 | headius | absolutely |
| 01:32:10 | headius | so any tight binding to any specific front-end should be out of scope, a separate project |
| 01:33:44 | rue | It is |
| 01:33:49 | rue | 'S called mod_rubinius ;) |
| 01:34:29 | headius | well, if that's what mod_rubinius is supposed to be, I guess I'm not as interested :) |
| 01:34:34 | rue | I have not come up with a daemon nomenclature yet |
| 01:35:00 | defunkt enters the room. | |
| 01:35:16 | headius | I'm interested in the webapp-serving side |
| 01:36:47 | rue | That is kinda vague |
| 01:37:36 | jeremydurham enters the room. | |
| 01:38:16 | headius | as is your current definition of mod_rubinius :) so I think we've got some wiggle room |
| 01:39:06 | rue | No, mod_rubinius is well-defined in that sense. It is a module built into a webserver. It will either allow dispatching to a daemon, process as part of the webserver or both |
| 01:39:15 | ezmobius | headius: i think the main thing jruby would need to do to be compatible with all the apps that want to run on mod_rubinius woudl be to have a good rack interface for glasfish |
| 01:39:37 | ezmobius | since rack is pretty much how any ruby code will interact with this system |
| 01:39:52 | rue | The second component is the daemon, as yet unnamed |
| 01:40:30 | headius | ezmobius: that much exists now, along with a couple rails-specific adapters |
| 01:41:56 | wycats__ leaves the room. | |
| 01:41:57 | ezmobius | cool, i know dudley is working to get the latest merb-core all running on glassfish |
| 01:42:26 | rue | I think I will call the daemon Rubicante |
| 01:42:27 | headius | yeah, I want to scrap goldspike entirely |
| 01:42:46 | headius | I just haven't gotten around to it hoping someone else would do it...but nicksieger has a rack version for glassfish that's probably the way to go |
| 01:43:08 | rue | (Working name only, to be changed should any abyssals complain.) |
| 01:43:36 | headius | my situation is that we can basically do everything we need to have a single daemon host as many apps as needed but this is an entirely new model for deploying ruby web apps, so we're treading new ground |
| 01:43:49 | headius | whatever the "second component" works out to be, it's the area we're trying to solidify too |
| 01:44:31 | wycats_ leaves the room. | |
| 01:46:07 | b3b0p leaves the room. | |
| 01:46:26 | b3b0p enters the room. | |
| 01:48:02 | b3b0p leaves the room. | |
| 01:48:14 | b3b0p enters the room. | |
| 01:48:14 | b3b0p leaves the room. | |
| 01:48:22 | wycats___ leaves the room. | |
| 01:48:50 | b3b0p enters the room. | |
| 01:52:50 | bburcham leaves the room. | |
| 01:54:35 | ttmrichter_ leaves the room. | |
| 01:58:20 | wvl_ enters the room. | |
| 02:02:47 | headius | ezmobius: I'm going to be in the bay for a few days next week, probably not up near you but I'd be up for getting dinner somewhere halfway |
| 02:02:50 | RyanTM leaves the room. | |
| 02:03:49 | ezmobius | cool, this week is a bit crazy for me but we could probably work something out |
| 02:05:08 | headius | ok...we're staying at the W in newark, across the dumbarton from Menlo Park...Monday night is probably out, but Sunday or Tuesday might work |
| 02:05:52 | headius | just seems like we're solving the same problems, and I hate to see duplicated effort |
| 02:07:05 | ezmobius | i think tuesday night we're having a little hackfest here in south park if you wanteds to come |
| 02:09:54 | headius | where's that? |
| 02:10:02 | headius | that might be ideal |
| 02:10:29 | headius | as in South Park, in SF? |
| 02:10:35 | ezmobius | yes |
| 02:10:40 | ezmobius | thats where the EY office is |
| 02:10:44 | ezmobius | 82 South Park st |
| 02:11:00 | ezmobius | in the soma |
| 02:11:33 | headius | that, that could work...few blocks from caltrain |
| 02:11:57 | ezmobius | yup |
| 02:12:19 | ezmobius | it;s at 8:30pm on tuesday, but if you wanted to come early to chat i'll be here |
| 02:12:39 | headius | ok...I have a few things to keep me and tom busy at Sun but it might be possible |
| 02:12:42 | headius | I'll keep you posted |
| 02:12:47 | ezmobius | congrats on the rc2 release btw |
| 02:12:54 | ezmobius | looks like you got a lot done |
| 02:13:04 | headius | thanks...yeah, that bug list astounds me |
| 02:13:38 | headius | we're rapidly approaching the point where the "Ruby" part of "JRuby" is mostly done |
| 02:14:08 | wvl leaves the room. | |
| 02:27:52 | zenspider | YAY!!!!!! |
| 02:28:14 | zenspider | lexer has gotten the beatdown! |
| 02:31:17 | headius | do tell! |
| 02:31:24 | _ADS enters the room. | |
| 02:31:30 | headius | your pure-ruby lexer? |
| 02:35:34 | tarcieri | A Ruby lexer that isn't completely insane? |
| 02:35:54 | kofno enters the room. | |
| 02:36:05 | headius | zenspider: you ever get a chance to look at tom's refactored JRuby lexer? |
| 02:36:20 | headius | it's a considerable improvement over the version you used for your port |
| 02:37:29 | zenspider | headius: no, I never looked at it |
| 02:37:51 | zenspider | here is my improvement: |
| 02:37:57 | zenspider | 1.0.0: loc = 2751 flog = 3801.8 |
| 02:38:03 | zenspider | dev : loc = 2108 flog = 3140.5 |
| 02:38:10 | zenspider | delta: loc = -643 flog = -661.3 |
| 02:39:08 | zenspider | I switched it over to stringscanner and cleaned it up a lot |
| 02:39:19 | headius | nice...run any new numbers with it? |
| 02:39:56 | headius | I've got some preliminary jvm bytecode generation utilities hanging around I'd like to eventually wire into a pure-ruby parser |
| 02:40:11 | zenspider | numbers? |
| 02:40:46 | headius | well, simplified code ought to have some performance improvement, no? |
| 02:40:50 | headius | especially leaning on stringscanner |
| 02:41:11 | zenspider | no clue |
| 02:41:32 | zenspider | I don't really care about the performance at this point... it is MUCH more readable and debuggable |
| 02:41:59 | headius | did you get line numbers, newlines, etc in? |
| 02:42:18 | headius | tom mentioned that ruby 1.9 doesn't have newline nodes, and instead has a newline flag on all nodes |
| 02:43:52 | zenspider | no, I'm still not doing positional yet |
| 02:43:58 | zenspider | that's probably next |
| 02:44:28 | zenspider | I've been thinking I'd go the position on every node rather than the newline node stuff... newline nodes piss me off |
| 02:45:13 | headius | I don't really care either way, though newline nodes are a bit of a hassle to represent cleanly |
| 02:45:37 | VVSiz_ enters the room. | |
| 02:45:40 | headius | but positioning is a must for tooling and backtraces |
| 02:47:49 | zenspider | yup. |
| 02:49:17 | wycats enters the room. | |
| 02:49:33 | wifelette_ enters the room. | |
| 02:49:43 | _mutle enters the room. | |
| 02:52:27 | VVSiz leaves the room. | |
| 03:00:41 | wifelette leaves the room. | |
| 03:01:22 | ttmrichter enters the room. | |
| 03:05:24 | mutle leaves the room. | |
| 03:06:07 | headius leaves the room. | |
| 03:08:40 | edward_g leaves the room. | |
| 03:11:55 | eventualbuddha enters the room. | |
| 03:22:39 | headius enters the room. | |
| 03:25:03 | twbray enters the room. | |
| 03:27:09 | dodecaphonic leaves the room. | |
| 03:43:48 | Yurik leaves the room. | |
| 03:48:17 | femtowin enters the room. | |
| 03:55:14 | macournoyer enters the room. | |
| 03:59:42 | twbray leaves the room. | |
| 04:11:33 | technomancy enters the room. | |
| 04:14:10 | therealadam enters the room. | |
| 04:14:14 | therealadam leaves the room. | |
| 04:17:10 | ezmobius leaves the room. | |
| 04:17:11 | ezmobius_ leaves the room. | |
| 04:23:43 | menator enters the room. | |
| 04:30:21 | twbray enters the room. | |
| 04:31:44 | menator leaves the room. | |
| 04:35:45 | gdagley leaves the room. | |
| 04:41:41 | mernen leaves the room. | |
| 04:51:44 | rue | Ew, I think I have to use Linux :P |
| 04:52:02 | rue | What do the cool kids use now? |
| 04:55:33 | headius | the cool kids or the kids who have grown to old to futz about with configuring linux? |
| 04:55:59 | headius | I'm in the latter camp, and I wouldn't choose anything but ubuntu right now |
| 04:58:23 | tmm1 | ubuntu, gentoo or archlinux |
| 04:58:58 | headius | gentoo might fit you better, ports and all |
| 04:59:10 | tyler_ enters the room. | |
| 04:59:21 | tmm1 | rue: why do you have to use linux? |
| 05:03:23 | headius | rue: use solaris :) |
| 05:04:10 | technomancy | hehe |
| 05:05:01 | technomancy | rue: the only thing icky about ubuntu is the way apt-get breaks up the ruby packages; if you install from source you won't notice much |
| 05:05:13 | technomancy | wonky gem locations for the lose |
| 05:05:45 | headius | yeah, that's a debian lunacy |
| 05:05:56 | headius | I still think they're nuts |
| 05:06:23 | technomancy | it's better than it used to be, but i still like going from source |
| 05:06:32 | macournoyer leaves the room. | |
| 05:06:35 | headius | I get tired of waiting |
| 05:06:47 | headius | but for a server I generally choose gentoo |
| 05:07:02 | headius | much more static environment |
| 05:07:05 | technomancy | ruby, emacs, and mozilla are where i spend 95% of my time; each is built from source except mozilla (it's too scary to build) |
| 05:09:07 | rubymaverick_ enters the room. | |
| 05:20:21 | defunkt leaves the room. | |
| 05:25:34 | headius leaves the room. | |
| 05:27:36 | binary42 leaves the room. | |
| 05:27:57 | mass | I would agree that debian shouldn't be trying to do what they are doing with ruby and gem without having core developers of both as package maintainers |
| 05:30:51 | eventualbuddha leaves the room. | |
| 05:30:52 | eventualbuddha_ enters the room. | |
| 05:31:01 | eventualbuddha_ leaves the room. | |
| 05:38:58 | eventualbuddha enters the room. | |
| 05:48:27 | Ingmar leaves the room. | |
| 05:48:59 | Ingmar enters the room. | |
| 05:50:30 | evan | is in Santa Barbara |
| 05:52:00 | aasmith | how are you liking it? |
| 05:52:37 | gnufied enters the room. | |
| 05:52:47 | lstoll enters the room. | |
| 05:53:32 | technomancy | that's a nice town. If i had stayed in CA i would have wanted to live there. |
| 05:53:51 | technomancy | doesn't wycats live there? |
| 05:54:06 | aasmith | yea, its funny that it's so nice, being to so close to LA. :) |
| 05:54:16 | jeremydurham | I thought wycats lived in like NY or something :) |
| 05:54:28 | wycats | evan: we should do dinner! |
| 05:54:35 | wycats | I happen to be in SB atm |
| 06:02:26 | rue | You hippies |
| 06:04:06 | tbmcmullen | Wow. JRuby RC2 was released today, and meets or exceeds MRI 1.9 in most benchmarks. |
| 06:05:17 | wycats | wow... MatzRuby 1.9 is pretty fast |
| 06:10:07 | jeremydurham | tbmcmullen: link? |
| 06:10:21 | tbmcmullen | http://headius.blogspot.com/2008/02/jruby-rc2-released-whats-next.html |
| 06:10:49 | jeremydurham | I assume he is referring to compiled, not interpreted |
| 06:14:04 | tizianobis enters the room. | |
| 06:21:45 | tbmcmullen leaves the room. | |
| 06:30:37 | TheVoice enters the room. | |
| 06:35:50 | technomancy leaves the room. | |
| 06:41:34 | tizianobis_ enters the room. | |
| 06:44:02 | jeremydurham leaves the room. | |
| 06:45:27 | _ADS leaves the room. | |
| 06:48:05 | felipec | what's MRI? |
| 06:48:15 | tbmcmullen | Matz Ruby Interpreter |
| 06:48:28 | tbmcmullen | i.e. normal Ruby |
| 06:50:23 | twbray leaves the room. | |
| 06:50:34 | srbaker leaves the room. | |
| 06:53:58 | tizianobis leaves the room. | |
| 07:00:11 | evan | wycats: you going to be in SB tomorrow? |
| 07:01:15 | wycats | evan: yeah |
| 07:01:21 | wycats | the movers are coming tomorrow |
| 07:01:37 | evan | lets have a meal tomorrow |
| 07:02:36 | zf enters the room. | |
| 07:09:46 | wycats | where are you staying? |
| 07:10:01 | rue | Oh, here and there |
| 07:11:03 | tizianobis_ leaves the room. | |
| 07:11:36 | wycats | evan: pm |
| 07:14:27 | evan | got it |
| 07:14:35 | evan | wycats: The Inn on East Beach |
| 07:15:22 | ko1 | evan: how about RubyKaigi? |
| 07:15:40 | evan | ko1_: just got your email, was about to reply |
| 07:15:43 | evan | ko1_: i'd love to speak |
| 07:15:54 | evan | ko1_: is there anything you'd like me to speak on? |
| 07:16:10 | ko1 | evan: thank you! we love to :) |
| 07:16:14 | ko1 | too |
| 07:16:44 | ko1 | evan: introduction of rubinius? |
| 07:16:54 | brixen | rue: were you planning on migrating the rest of the benchmarks to the toplevel dir? |
| 07:17:16 | wycats | evan: you gotta use heckle on your intro to rubinius talk |
| 07:17:25 | wycats | mutate it for all these different events |
| 07:17:59 | rue | brixen: Tomorrow |
| 07:20:27 | brixen | rue: cool |
| 07:20:29 | ezmobius enters the room. | |
| 07:39:28 | twbray enters the room. | |
| 07:40:54 | lstoll leaves the room. | |
| 07:42:41 | lstoll enters the room. | |
| 07:42:44 | evan | ko1_: is it ok if I get you a proposal on my monday? |
| 07:44:12 | srbaker enters the room. | |
| 07:44:26 | ko1 | evan: sure |
| 07:55:27 | lstoll leaves the room. | |
| 07:55:42 | lstoll enters the room. | |
| 07:56:39 | oleganza leaves the room. | |
| 08:00:20 | twbray leaves the room. | |
| 08:01:51 | eventualbuddha leaves the room. | |
| 08:06:57 | smtlaissezfaire enters the room. | |
| 08:07:07 | fork_ leaves the room. | |
| 08:10:59 | fork enters the room. | |
| 08:11:06 | srbaker leaves the room. | |
| 08:12:14 | rby enters the room. | |
| 08:12:20 | rby_ enters the room. | |
| 08:12:35 | twbray enters the room. | |
| 08:18:34 | yipstar leaves the room. | |
| 08:27:04 | rby leaves the room. | |
| 08:27:04 | rby_ leaves the room. | |
| 08:31:13 | TheVoice leaves the room. | |
| 08:41:22 | ezmobius leaves the room. | |
| 08:46:50 | tbmcmullen | so in looking at the CGI.unescapeHTML method in MRI... It appears to be missing a huge amount of the html entities that it could escape. |
| 08:47:07 | tbmcmullen | So my question is... in porting it over, should I go above and beyond what MRI does? Or just leave it as is? |
| 08:58:56 | smtlaissezfaire | tbmcmullen: Would you consider them bugs? |
| 08:59:46 | tbmcmullen | smtlaissezfaire: Not really sure. Perhaps it was intentional, though I can't imagine why. |
| 09:00:03 | tbmcmullen | smtlaissezfaire: perhaps due to the lack of unicode support there was no point to it... |
| 09:03:42 | wifelette_ enters the room. | |
| 09:04:27 | smtlaissezfaire | tbcmullen: Isn't 2.0 supposed to support unicode? |
| 09:05:01 | tbmcmullen | smtlaissezfaire: Right. And I was wrong anyway... the html entities its missing are simple ASCII ones. Which makes it more strange. |
| 09:07:20 | lstoll leaves the room. | |
| 09:11:14 | nathansobo enters the room. | |
| 09:16:34 | wycats leaves the room. | |
| 09:16:42 | wifelette leaves the room. | |
| 09:21:40 | ragnard enters the room. | |
| 09:31:45 | bitbckt leaves the room. | |
| 09:36:26 | smtlaissezfaire leaves the room. | |
| 09:38:09 | nathansobo leaves the room. | |
| 09:40:33 | twbray leaves the room. | |
| 09:42:39 | thehcdreamer enters the room. | |
| 09:43:01 | thehcdreamer leaves the room. | |
| 09:44:49 | dewd leaves the room. | |
| 09:58:58 | hassox enters the room. | |
| 09:59:16 | tbmcmullen leaves the room. | |
| 10:19:05 | crossblaim enters the room. | |
| 10:20:16 | crossblaim | hello |
| 10:25:46 | comboy enters the room. | |
| 10:33:42 | olabini enters the room. | |
| 10:53:28 | thehcdreamer enters the room. | |
| 10:54:17 | jp_tix | http://pastie.textmate.org/153345 ← any ideas? |
| 10:54:43 | lstoll enters the room. | |
| 10:56:25 | antares enters the room. | |
| 10:59:53 | antares leaves the room. | |
| 11:00:06 | antares enters the room. | |
| 11:08:35 | ragnard | jp_tix: can you pastie your platform.conf? |
| 11:11:25 | jp_tix | ragnard: http://pastie.caboo.se/153348 |
| 11:12:41 | ragnard | jp_tix: huh. seems like the typegenerator didn't do its job... |
| 11:12:44 | context | rubinius will run on my Amiga right ? |
| 11:13:46 | context | 9223372036854775807 is the new infinity |
| 11:13:48 | context | sexy |
| 11:15:05 | thehcdreamer leaves the room. | |
| 11:15:23 | ragnard | jp_tix: what do you get when doing this: http://pastie.caboo.se/153349 |
| 11:16:18 | jp_tix | ragnard: "rbx.platform.typedef.qaddr_t = pointer\nrbx.platform.typedef.caddr_t = pointer\n" |
| 11:17:37 | ragnard | jp_tix: this is on ppc, right? |
| 11:17:42 | jp_tix | yes |
| 11:18:01 | ragnard | i though dbussink got it running yesterday... |
| 11:18:45 | dbussink | yeah, i got it running on a ppc system yesterday |
| 11:18:58 | jp_tix | well, the build completes for me as well |
| 11:19:14 | ragnard | maybe a missing -D...something... |
| 11:19:18 | dbussink | what version are you on? |
| 11:19:26 | dbussink | 10.5 or older? |
| 11:19:31 | jp_tix | 10.5.2 |
| 11:19:36 | dbussink | hmm, strange |
| 11:19:46 | ragnard | jp_tix: try this |
| 11:19:49 | dbussink | are you sure you have the latest version? |
| 11:20:02 | jp_tix | dbussink: latest version of rubinius? |
| 11:20:08 | dbussink | yeah |
| 11:20:10 | ragnard | http://pastie.org/153350 |
| 11:20:25 | dbussink | i've made some fixes to the types generator these last few days |
| 11:20:47 | jp_tix | `git pull` says I'm already up-to-date |
| 11:21:09 | jp_tix | and working directory is clean |
| 11:21:15 | dbussink | what does it say when you run that pastie? |
| 11:21:35 | jp_tix | typedef __int64_t __darwin_off_t; |
| 11:21:35 | jp_tix | typedef __darwin_off_t off_t; |
| 11:22:48 | dbussink | is that a copy past? |
| 11:22:57 | dbussink | because it looks like there are spaces missing |
| 11:23:10 | jp_tix | yes, tabs gets messed up |
| 11:23:23 | jp_tix | http://pastie.caboo.se/private/wturj2hv5bwzj7lwezurkg |
| 11:23:33 | jp_tix | err |
| 11:23:36 | dbussink | hmm, there too :) |
| 11:23:45 | jp_tix | now you got my grep coloring |
| 11:23:50 | jp_tix | :P |
| 11:23:58 | ragnard | maybe that's it? grep coloring? |
| 11:24:09 | dbussink | hmm, might be a problem |
| 11:24:23 | jp_tix | http://pastie.caboo.se/153352 |
| 11:24:47 | dbussink | probably should convert this to something not using grep |
| 11:25:10 | dbussink | do the grepping in ruby, will be a bit slower but probably more often correct |
| 11:25:20 | dbussink | jp_tix: does it work if you disable your grep coloring? |
| 11:25:43 | dbussink | i'm off for a shower, be back later |
| 11:25:46 | ragnard | dbussink: btw, i tracked down the openbsd sigsegv... really strange though |
| 11:25:51 | ragnard | dbussink: oh, later. |
| 11:27:33 | jp_tix | what do I need to do after disabling grep coloring? |
| 11:28:06 | jp_tix | dbussink: ↑ |
| 11:29:23 | ragnard | jp_tix: you could just rm runtime/platform.conf |
| 11:29:28 | jp_tix | ah, yes |
| 11:29:44 | jp_tix | that's it, the TypesGenerator.generate has a lot more output now |
| 11:29:59 | dctanner enters the room. | |
| 11:37:07 | dbussink | jp_tix: thnx for the debugging :) |
| 11:38:57 | dbussink | jp_tix: i'll fix this right away :) |
| 11:39:03 | jp_tix | yay, it works now |
| 11:39:09 | jp_tix | great, thanks to you too :) |
| 11:39:28 | dbussink | jp_tix: you also get some failures on File#truncate? |
| 11:40:05 | jp_tix | running tests now |
| 11:40:17 | jp_tix | `rake spec`, right? |
| 11:42:16 | dbussink | ./bin/ci |
| 11:42:18 | boyscout | 1 commit by Dirkjan Bussink |
| 11:42:19 | boyscout | * Remove grep dependency in TypesGenerator; c7450c3 |
| 11:42:25 | dbussink | jp_tix: the fix is in |
| 11:42:35 | jp_tix | cool |
| 11:44:16 | jp_tix | yeah, I have 10 errors on File#truncate |
| 11:45:46 | cjheath leaves the room. | |
| 11:48:05 | _mutle leaves the room. | |
| 11:48:09 | mutle enters the room. | |
| 11:50:26 | dbussink | jp_tix: yeah, i found those too, no idea what's wrong though |
| 11:50:40 | dctanner_ enters the room. | |
| 11:53:32 | thehcdreamer enters the room. | |
| 11:54:10 | graza enters the room. | |
| 11:54:16 | rubuildius | Dirkjan Bussink: c7450c379; 4824 examples, 17537 expectations, 0 failures, 0 errors |
| 11:55:54 | jp_tix | dbussink: it works on Intel OS X? |
| 12:02:12 | antares leaves the room. | |
| 12:03:38 | GMFlash leaves the room. | |
| 12:03:45 | GMFlash enters the room. | |
| 12:05:20 | context | jp_tix: try it ? |
| 12:05:35 | jp_tix | context: I don't have an intel machine :( |
| 12:05:42 | context | oh :x |
| 12:08:48 | dctanner leaves the room. | |
| 12:13:31 | boyscout | 1 commit by Vladimir Sizikov |
| 12:13:32 | boyscout | * New specs for RangeExceptions out of Fixnum and Array methods.; f43383a |
| 12:13:37 | lstoll leaves the room. | |
| 12:15:22 | lstoll enters the room. | |
| 12:15:36 | lstoll leaves the room. | |
| 12:16:02 | lstoll enters the room. | |
| 12:17:03 | lstoll leaves the room. | |
| 12:24:11 | rubuildius | Vladimir Sizikov: f43383a15; 4824 examples, 17537 expectations, 0 failures, 0 errors |
| 12:25:03 | dbussink | jp_tix: yeah, no problems on intel and also works ok on linux |
| 12:29:05 | Fullmoon leaves the room. | |
| 12:29:24 | jp_tix | dbussink: perhaps an endianness issue? |
| 12:29:40 | jp_tix | or perhaps not |
| 12:31:06 | dbussink | jp_tix: well, try FFI::TypeDefs[:off_t] in rubinius, that gives a sigbus for me |
| 12:31:15 | dbussink | jp_tix: which isn't good :) |
| 12:31:54 | dctanner_ leaves the room. | |
| 12:32:06 | jp_tix | dbussink: no sigbus here |
| 12:32:53 | dbussink | jp_tix: and on other types? |
| 12:34:31 | jp_tix | `shotgun/rubinius -e "p FFI::TypeDefs"` works fine |
| 12:35:15 | thehcdreamer leaves the room. | |
| 12:36:44 | octopod enters the room. | |
| 12:41:08 | jp_tix | I didn't rebuild after you removed the grep thing though, let me try that |
| 12:49:47 | graza_ enters the room. | |
| 12:49:48 | graza leaves the room. | |
| 12:54:14 | jp_tix | still works fine |
| 12:57:38 | geekounet leaves the room. | |
| 13:08:55 | crossblaim leaves the room. | |
| 13:21:42 | graza enters the room. | |
| 13:21:43 | graza_ leaves the room. | |
| 13:29:36 | rubymonk enters the room. | |
| 13:34:21 | graza_ enters the room. | |
| 13:34:22 | graza leaves the room. | |
| 13:37:22 | rby enters the room. | |
| 13:41:13 | Arjen_ enters the room. | |
| 13:45:45 | antares enters the room. | |
| 13:46:56 | graza enters the room. | |
| 13:46:57 | graza_ leaves the room. | |
| 13:53:28 | dbussink | jp_tix: hmm, strange, if i access some fields in FFI::TypeDefs or the whole hash, i get a sigbus at some point |
| 13:59:06 | graza_ enters the room. | |
| 13:59:07 | graza leaves the room. | |
| 14:08:30 | edward_g enters the room. | |
| 14:12:18 | dbussink | jp_tix: looks like the gcc options break stuff |
| 14:30:04 | graza_ leaves the room. | |
| 14:43:03 | imajes enters the room. | |
| 14:45:42 | langenberg enters the room. | |
| 15:02:21 | enebo enters the room. | |
| 15:03:37 | langenberg leaves the room. | |
| 15:09:06 | jp_tix | dbussink: do you think this is related to the File#truncate errors? |
| 15:09:20 | dbussink | jp_tix: probably not |
| 15:10:46 | rby_ enters the room. | |
| 15:10:55 | rby leaves the room. | |
| 15:11:46 | jp_tix | dbussink: so in rubinius, the kernel/core stuff is compiled, right? how do I recompile a single file? |
| 15:14:07 | dbussink | just do rake build after you changed something |
| 15:20:30 | chris2 enters the room. | |
| 15:21:25 | rubyconsumer leaves the room. | |
| 15:22:56 | binary42 enters the room. | |
| 15:23:28 | rubyconsumer enters the room. | |
| 15:24:16 | RyanTM enters the room. | |
| 15:30:11 | srbaker enters the room. | |
| 15:31:17 | geekounet leaves the room. | |
| 15:35:03 | gnufied leaves the room. | |
| 15:35:28 | geekounet enters the room. | |
| 15:36:38 | jp_tix | dbussink: this goes way over my head, no idea why ftruncate() would behave differently on intel vs ppc |
| 15:36:52 | jp_tix | I'll leave it to you guys :) |
| 15:38:18 | kofno leaves the room. | |
| 15:59:55 | twbray enters the room. | |
| 16:01:01 | headius enters the room. | |
| 16:01:50 | be9 enters the room. | |
| 16:07:53 | menator enters the room. | |
| 16:09:24 | nemerle enters the room. | |
| 16:09:32 | twbray leaves the room. | |
| 16:15:10 | mutle leaves the room. | |
| 16:16:40 | mutle enters the room. | |
| 16:24:22 | Arjen | Is rake build expected to fail at the moment? |
| 16:24:49 | Arjen | configure: error: C++ preprocessor "/lib/cpp" fails sanity check |
| 16:30:52 | Arjen | Hmm, probably installed the wrong gcc package... |
| 16:38:35 | mernen enters the room. | |
| 16:39:54 | crossblaim enters the room. | |
| 16:40:39 | crossblaim | hi |
| 16:47:31 | manveru | strange... |
| 16:47:48 | manveru | one can't set nor alias $1 |
| 16:48:07 | manveru | is that for speed reasons in 1.8? |
| 17:01:31 | dewd enters the room. | |
| 17:11:08 | wycats enters the room. | |
| 17:20:28 | boyscout | 1 commit by oleg dashevskii |
| 17:20:29 | boyscout | * Cross-breed and update for, while & until language specs. Little fix for hash spec.; 02225da |
| 17:22:35 | headius_ enters the room. | |
| 17:28:38 | headius | afternoon |
| 17:29:16 | djwhitt | headius: has anyone published benchmarks yet using the new JRuby? |
| 17:29:27 | headius | published, not that I know of |
| 17:29:47 | headius | the benchmarks game/shootout will probably get updated pretty quickly |
| 17:30:01 | djwhitt | ah, cool, I suppose I could just run some myself too... |
| 17:30:55 | VVSiz | new specs, yay :) |
| 17:30:58 | headius | yeah, go right ahead |
| 17:31:08 | VVSiz | btw, who is Oleg Dashevskii here? |
| 17:33:15 | be9 | VVSiz: I am |
| 17:33:21 | eventualbuddha enters the room. | |
| 17:33:48 | twbray enters the room. | |
| 17:34:01 | rubuildius | oleg dashevskii: 02225daa5; 4827 examples, 17543 expectations, 0 failures, 0 errors |
| 17:34:32 | VVSiz | be9: cool. :) I see that you've commented one spec out and was going to suggest ot use fails_on :rubinius to wrap it instead.... but it doesn't work in this case and still breaks rubinius |
| 17:35:04 | therealadam enters the room. | |
| 17:35:36 | be9 | VVSiz: yeah, that's a compiler error |
| 17:36:13 | wifelette enters the room. | |
| 17:36:26 | nathansobo enters the room. | |
| 17:36:31 | VVSiz | understood. it's just we use the specs over in #JRuby as well, so I keep en eye on them, just in case :) |
| 17:38:05 | twbray leaves the room. | |
| 17:38:23 | headius leaves the room. | |
| 17:38:30 | VVSiz | be9: do you think the compiler bug would be fixed soon? |
| 17:38:49 | be9 | VVSiz: dunno |
| 17:40:00 | menator leaves the room. | |
| 17:40:11 | VVSiz | hmmm, I can't come up with any way to enable this test for JRuby and MRI but disable for rbx... |
| 17:40:33 | headius | because it's a compile problem? |
| 17:41:14 | dbussink | VVSiz: it's not possible until we fix the compile problem |
| 17:41:34 | djwhitt | could you put it in another file that you require conditionally? |
| 17:41:46 | headius | that seems pretty cumbersome |
| 17:42:00 | headius | I think it should just stay broken until the compiler is fixed |
| 17:42:22 | VVSiz | yeah, it can wait, especially since the compiler bug is known and is going to be fixed |
| 17:50:44 | headius | http://blog.erichsen.net/2008/02/17/benchmarking-fun-with-jruby-11-rc2-glassfish-and-rails-202/ |
| 17:51:43 | wifelette_afk leaves the room. | |
| 17:52:23 | boyscout | 2 commits by Dirkjan Bussink |
| 17:52:24 | boyscout | * Replace unsafe sprintf and strcat with safe snprintf calls; 81c223d |
| 17:52:25 | boyscout | * Remove GCC inline optimizations; 2a8cf76 |
| 17:52:52 | wycats_ enters the room. | |
| 17:53:16 | wycats leaves the room. | |
| 17:56:22 | pietia enters the room. | |
| 17:59:49 | wycats_ leaves the room. | |
| 17:59:52 | wycats enters the room. | |
| 18:00:24 | wycats leaves the room. | |
| 18:00:52 | wycats enters the room. | |
| 18:03:59 | rubuildius | Dirkjan Bussink: 81c223dc6; 4827 examples, 17543 expectations, 0 failures, 0 errors |
| 18:07:20 | dbussink | headius: nice |
| 18:07:27 | boyscout | 1 commit by Dirkjan Bussink |
| 18:07:28 | boyscout | * Replace more unsafe sprintf calls with snprintf; 3857386 |
| 18:07:30 | dbussink | headius: any ideas that the cause of the difference is? |
| 18:07:38 | headius | well, the benchmark has a lot of problems |
| 18:07:49 | headius | but it's possible his ruby instance is using the native mysql extension for one |
| 18:08:17 | dbussink | probably, i don't think anyone uses the pure ruby version in a production environment |
| 18:08:51 | headius | as far as I can tell he's also running it in development mode, which would mean jruby's probably recompiling a lot of stuff |
| 18:09:07 | headius | and btw, sorry for posting in here, I didn't realize I was pasting to #rubinius |
| 18:09:39 | langenberg enters the room. | |
| 18:15:56 | kofno enters the room. | |
| 18:16:01 | dbussink | hehe, no problems, keeps us on our toes :) |
| 18:17:46 | nathansobo leaves the room. | |
| 18:18:51 | rubuildius | Dirkjan Bussink: 3857386b6; 4827 examples, 17543 expectations, 0 failures, 0 errors |
| 18:19:26 | benburkert enters the room. | |
| 18:22:07 | wycats leaves the room. | |
| 18:22:36 | wmoxam enters the room. | |
| 18:23:36 | wycats enters the room. | |
| 18:36:11 | bitbckt enters the room. | |
| 18:42:49 | geekounet leaves the room. | |
| 18:42:59 | geekounet enters the room. | |
| 18:43:29 | rue | Morning |
| 18:45:15 | be9 | damn, these syntax errors narrow the abilities of speccing. I've just found out that rbx doesn't give a syntax error where it should: http://pastie.caboo.se/153436 |
| 18:45:19 | be9 | morning, rue |
| 18:48:55 | wycats | be9: that's a spec failure |
| 18:48:56 | wycats | ;) |
| 18:57:21 | lopex enters the room. | |
| 18:58:01 | headius leaves the room. | |
| 19:11:42 | dctanner enters the room. | |
| 19:17:51 | boyscout | 1 commit by oleg dashevskii |
| 19:17:52 | boyscout | * Update specs for calling methods.; b1d3ba9 |
| 19:24:43 | antares | rue, hey there :) |
| 19:26:04 | pietia leaves the room. | |
| 19:26:38 | pietia enters the room. | |
| 19:28:51 | rubuildius | oleg dashevskii: b1d3ba9d1; 4842 examples, 17569 expectations, 0 failures, 0 errors |
| 19:29:48 | geekounet leaves the room. | |
| 19:36:16 | antares leaves the room. | |
| 19:44:04 | rby_ leaves the room. | |
| 19:55:47 | geekounet leaves the room. | |
| 19:57:27 | wifelette_ enters the room. | |
| 19:59:35 | wycats_ enters the room. | |
| 20:03:12 | radarek enters the room. | |
| 20:07:50 | wycats leaves the room. | |
| 20:09:19 | twbray enters the room. | |
| 20:10:37 | wifelette leaves the room. | |
| 20:12:30 | wifelette_ leaves the room. | |
| 20:13:48 | evan | dbussink: you around? |
| 20:13:50 | evan | please add those inline options back. |
| 20:13:54 | evan | they DO make a difference. |
| 20:14:02 | dbussink | evan: hmm, where do they make the difference? |
| 20:14:15 | dbussink | evan: because the cause segbus errors on ppc |
| 20:14:42 | evan | thats very strange. |
| 20:14:47 | evan | hm |
| 20:14:54 | evan | i guess they might not be necessary anymore |
| 20:15:01 | evan | -Winline should be left in though |
| 20:15:14 | evan | those other options were required to get gcc to actually do the inlining |
| 20:15:21 | dbussink | i've tried it over and over on ppc and each it failed when adding those options |
| 20:15:22 | evan | without them, it decided that there was too much code to inline and would stop |
| 20:15:45 | evan | thats really odd |
| 20:16:00 | dbussink | well, i've run ./bin/bm and ./bin/ci a few times and saw no significant difference |
| 20:16:04 | evan | because they don't tell gcc to generate code any differently |
| 20:16:11 | evan | only to work harder to inline |
| 20:16:17 | evan | leave -Winline on |
| 20:16:25 | dbussink | well, that's still in there |
| 20:16:35 | evan | not in your diff it's not |
| 20:16:43 | dbussink | ah no, wait i removed it because it did trigger those warnings indeed |
| 20:16:47 | cjheath enters the room. | |
| 20:16:50 | evan | right see |
| 20:16:54 | dbussink | that it stopped inlining becasue it would be too much |
| 20:17:00 | evan | right |
| 20:17:01 | dbussink | but where do these options make a difference? |
| 20:17:13 | evan | where we really do want gcc to inline |
| 20:17:19 | evan | but it stops because it thinks it's too much |
| 20:17:31 | evan | it's fine for now i guess |
| 20:17:31 | dbussink | but what about the breakage? |
| 20:17:36 | evan | i'll look at it when i get back home |
| 20:17:45 | evan | the breakage is completely bizarre |
| 20:17:49 | evan | if you ask me |
| 20:17:56 | evan | if you could |
| 20:18:01 | evan | try them one option at a time. |
| 20:18:17 | rue | How was the skydiving? |
| 20:18:20 | evan | i used those options on PPC before |
| 20:18:22 | evan | and it worked fine |
| 20:18:41 | evan | rue: fun! |
| 20:18:50 | evan | managed to tweak my back a little though |
| 20:18:51 | evan | :/ |
| 20:19:03 | dbussink | well, i could trigger them by simply doing FFI::TypeDefs[:off_t] and then FFI::TypeDefs in irb |
| 20:19:14 | evan | huh? |
| 20:19:22 | dbussink | then i tested with DEV=1, and it didn't fail |
| 20:19:32 | evan | the inline options were causing causing problems with Hash lookup?! |
| 20:19:42 | evan | whats 'no significant difference'? |
| 20:19:58 | evan | 1 second? .1 second? .01 second? |
| 20:20:30 | dbussink | evan: well, i ran ./bin/bm and ./bin/ci a few times and the difference were within the margins of difference that each run has |
| 20:21:07 | dbussink | a 1 second difference can easily occur within two ./bin/ci runs that are exactly the same |
| 20:21:10 | evan | ok, well, i guess we can leave it for now |
| 20:21:19 | evan | those options used to make a significant difference |
| 20:21:27 | dbussink | i get 25 secs on a ./bin/ci right now |
| 20:21:38 | evan | perhaps the VM has been reworked enough now that that is not the case |
| 20:21:39 | dbussink | and thats in the same ballpark as it was before |
| 20:21:46 | dbussink | i guess so |
| 20:21:48 | headius enters the room. | |
| 20:22:11 | rue | evan: Chute opening or landing? |
| 20:22:11 | dbussink | but i'm off, i could setup some proper profiling / testing somewhere in the next few days |
| 20:22:20 | evan | rue: not sure |
| 20:22:22 | dbussink | to do a really good comparison |
| 20:22:24 | dbussink | later |
| 20:22:27 | rue | *wave |
| 20:22:43 | evan | ok, i gotta get some food. |
| 20:22:51 | headius | howdy |
| 20:23:08 | rue | Hola |
| 20:24:59 | eventualbuddha leaves the room. | |
| 20:32:45 | antares enters the room. | |
| 20:33:29 | djwhitt enters the room. | |
| 20:35:08 | lvmc enters the room. | |
| 20:38:59 | wycats_ leaves the room. | |
| 20:39:45 | be9 leaves the room. | |
| 20:45:41 | rubyconsumer leaves the room. | |
| 20:56:51 | rubyconsumer enters the room. | |
| 21:00:49 | twbray leaves the room. | |
| 21:07:05 | felipec leaves the room. | |
| 21:07:25 | dodecaphonic enters the room. | |
| 21:10:22 | rby enters the room. | |
| 21:11:09 | headius leaves the room. | |
| 21:11:42 | meisdave enters the room. | |
| 21:13:00 | radarek leaves the room. | |
| 21:16:38 | twbray enters the room. | |
| 21:17:55 | jeremydurham enters the room. | |
| 21:18:48 | radarek enters the room. | |
| 21:20:47 | lopex leaves the room. | |
| 21:21:49 | smtlaissezfaire enters the room. | |
| 21:30:38 | graza enters the room. | |
| 21:37:50 | smtlaissezfaire leaves the room. | |
| 21:38:36 | enebo leaves the room. | |
| 21:39:21 | twshelton leaves the room. | |
| 21:41:47 | _ADS enters the room. | |
| 21:42:00 | wycats enters the room. | |
| 21:42:01 | agardiner enters the room. | |
| 21:44:55 | crossblaim leaves the room. | |
| 21:45:48 | crossblaim enters the room. | |
| 21:57:46 | graza_ enters the room. | |
| 21:57:47 | graza leaves the room. | |
| 21:59:07 | crossblaim leaves the room. | |
| 22:03:39 | lstoll enters the room. | |
| 22:03:55 | octopod leaves the room. | |
| 22:12:05 | lstoll leaves the room. | |
| 22:14:52 | ezmobius enters the room. | |
| 22:15:14 | crossblaim enters the room. | |
| 22:19:44 | RobJ enters the room. | |
| 22:19:56 | graza_ leaves the room. | |
| 22:24:33 | rby leaves the room. | |
| 22:25:24 | dc_ leaves the room. | |
| 22:25:59 | dodecaphonic leaves the room. | |
| 22:25:59 | pietia leaves the room. | |
| 22:27:59 | crossblaim leaves the room. | |
| 22:31:20 | twbray leaves the room. | |
| 22:32:30 | twbray enters the room. | |
| 22:34:15 | twbray leaves the room. | |
| 22:42:31 | Arjen_ leaves the room. | |
| 22:43:39 | RobJ leaves the room. | |
| 22:48:25 | Arjen_ enters the room. | |
| 22:49:14 | wycats_ enters the room. | |
| 22:50:32 | langenberg leaves the room. | |
| 22:51:46 | boyscout | 17 commits by Eero Saynatkari |
| 22:51:47 | boyscout | * Better randomization of possible large reverse-sorted, another 10x drop.; a586ac1 |
| 22:51:48 | boyscout | * Small stupid heuristic to slash large reverse-sorted Array#sort times.; 38ab306 |
| 22:51:49 | boyscout | * Added blurb about benchmark/ to README-DEVELOPERS.; 52812f0 |
| 22:51:50 | boyscout | * Added array generation benchmark.; 70e3427 |
| 22:51:51 | boyscout | * Renamed strategies to benchmark convention.; 381ce01 |
| 22:51:52 | boyscout | ... |
| 22:52:11 | djwhitt | egad! |
| 22:52:14 | jeremydurham | rue: you are sick |
| 22:52:16 | wycats__ enters the room. | |
| 22:52:17 | wycats leaves the room. | |
| 22:52:18 | djwhitt | nice work |
| 22:53:37 | rue | Heh, I think it looks more work than it is. I was mainly just moving stuff around. |
| 22:54:13 | rue | Although I am happy with the sort algo change. It does not regress on non-reverse-sorted either |
| 22:54:21 | djwhitt | nice |
| 22:55:34 | wycats__ leaves the room. | |
| 22:55:38 | crossblaim enters the room. | |
| 22:55:44 | wycats enters the room. | |
| 22:59:48 | AndrewO enters the room. | |
| 23:03:55 | rubuildius | Eero Saynatkari: a586ac18b; 4842 examples, 17569 expectations, 0 failures, 0 errors |
| 23:04:09 | djwhitt | interesting, seems like I actually see a slight speed improvement |
| 23:04:21 | djwhitt | I wouldn't have expected it from that sort change alone |
| 23:07:27 | antares | rue, are you around? |
| 23:07:36 | rue | AndrewO: Yeah, hey |
| 23:07:41 | rue | antares: Erm ^ |
| 23:07:43 | eventualbuddha enters the room. | |
| 23:07:55 | rue | djwhitt: Improvement where? |
| 23:08:02 | djwhitt | oh, just running ./bin/ci |
| 23:08:06 | antares | rue, which args does machine_save_args save (it is in machine.c) ? |
| 23:08:31 | antares | rue, just launch arguments it seems? |
| 23:09:36 | antares | rue, got it, nevermind |
| 23:09:58 | rue | antares: Yep, minus of course the program name itself |
| 23:10:12 | antares | rue, I see it is set to ARG0 Ruby constant |
| 23:10:55 | rue | djwhitt: Hm, yeah, maybe a 1/30 here |
| 23:11:16 | djwhitt | rue: yeah, that sounds about right |
| 23:11:24 | rue | antares: Yep |
| 23:11:38 | rue | djwhitt: Could be directory sorting |
| 23:12:05 | djwhitt | rue: yeah, I was wondering if it was something like that. doesn't seem like the specs themselves are that sort heavy |
| 23:15:02 | wycats_ leaves the room. | |
| 23:23:06 | lvmc | rue, mod_rubinius will be writer first for nginx or apache? |
| 23:23:33 | rue | lvmc: Apache |
| 23:23:38 | lvmc | ;) |
| 23:25:22 | hornbeck leaves the room. | |
| 23:25:33 | hornbeck enters the room. | |
| 23:27:58 | antares | rue, is there an overview of external libs shotgun uses? |
| 23:30:14 | wycats | JRuby: "Nobody can question that JRuby is more compatible with Ruby 1.8.6 than any other Ruby implementation available." |
| 23:30:15 | wycats | yay |
| 23:32:09 | lstoll enters the room. | |
| 23:32:32 | rue | antares: Not really. shotgun/external_libs/ has all of them |
| 23:32:33 | lstoll leaves the room. | |
| 23:33:03 | lstoll enters the room. | |
| 23:34:19 | rubuildius_ppc enters the room. | |
| 23:36:30 | aotearoa enters the room. | |
| 23:39:10 | crafterm enters the room. | |
| 23:45:06 | jeremydurham | I am still just surprised by the performance comments. And hoping that regex is significantly faster (in JRuby) |
| 23:45:43 | rue | Whu? |
| 23:46:30 | graza enters the room. | |
| 23:47:05 | rue | graza: Hey, was it you having issues with git? |
| 23:47:22 | graza | rue: yeah |
| 23:47:37 | graza | you've seen my post on the forum? |
| 23:49:14 | rue | The ML only |
| 23:50:15 | rue | I will be right back--did you try recompiling/reinstalling git and/or a different version? |
| 23:50:32 | rue | The repo seemed to be fine when I tried on 1.5.3.7 |
| 23:52:16 | kofno leaves the room. | |
| 23:52:50 | crossblaim leaves the room. | |
| 23:53:26 | djwhitt | benchmarks: http://pastie.org/153523 |
| 23:53:31 | djwhitt | just in case anyone is interested |
| 23:55:40 | chris2 leaves the room. | |
| 23:55:53 | crafterm enters the room. | |
| 23:55:57 | crafterm | hi all |
| 23:56:14 | jeremydurham | the ruby in this context is yarv or 1.8? |
| 23:56:20 | jeremydurham | or sorry, 1.9 or 1.8? ;) |
| 23:56:20 | graza | rue: just checked git relnotes. there's a segfault fix in 1.5.4.2 that might help me. |
| 23:56:57 | djwhitt | jeremydurham: 1.8 |
| 23:57:01 | wycats_ enters the room. | |
| 23:57:06 | djwhitt | I'll add that to the pastie |