Da
Show enters and exits. Hide enters and exits.
| 00:12:07 | evan | brixen: sup? |
| 00:12:14 | evan | had to apply some updates |
| 00:16:05 | brixen | np |
| 00:16:19 | brixen | so, spec/ruby will exist for some people |
| 00:16:36 | brixen | and when they pull this commit it won't apply |
| 00:16:45 | evan | ack |
| 00:16:45 | brixen | until they delete that dir |
| 00:16:55 | brixen | I can't think of a good way around that |
| 00:16:57 | evan | well |
| 00:17:06 | evan | maybe we make a hard break from spec/ruby and spec/frozen |
| 00:17:12 | evan | to something else |
| 00:17:14 | brixen | could |
| 00:17:20 | brixen | but I dislike that |
| 00:17:36 | brixen | better to have some pain now for a better organization I think |
| 00:17:41 | evan | ok |
| 00:17:51 | evan | well, those people who ask wtf is up will have to rm -rf the dir first |
| 00:17:53 | evan | then git pull. |
| 00:17:58 | brixen | I don't think it will be too many people |
| 00:18:04 | evan | do it then. |
| 00:18:14 | evan | we live in an uncertain world. |
| 00:18:22 | brixen | but yes, a bit of a hiccup |
| 00:18:24 | brixen | sure |
| 00:18:34 | brixen | and the solution is just rm spec/ruby |
| 00:18:39 | evan | right. |
| 00:18:42 | brixen | k |
| 00:18:47 | evan | no biggy. |
| 00:18:50 | brixen | cool |
| 00:30:01 | brixen | evan: btw, it's the Mutex#lock and Mutex#try_lock specs that where hanging last time I synced rubyspec |
| 00:30:11 | evan | oh ok. |
| 00:30:12 | evan | 10-4 |
| 00:30:15 | evan | need to check those out. |
| 00:30:19 | brixen | k |
| 01:13:50 | evan | mmm |
| 01:13:55 | evan | i think i've been over-inlining |
| 01:14:12 | brixen | where? |
| 01:14:26 | evan | well, i got to thinking about certain code shapes |
| 01:14:40 | evan | and the current code Inlines if there is only one class in an IC |
| 01:14:44 | evan | which is fine |
| 01:15:03 | evan | but it does that irrespective of how many times that one class was actually used |
| 01:15:12 | evan | in otherwords, for |
| 01:15:19 | evan | if whatever; foo; else; bar; end |
| 01:15:28 | evan | if whatever was true 1999 |
| 01:15:37 | evan | and the JIT kicks in |
| 01:15:45 | evan | it would inline both foo and bar |
| 01:15:51 | evan | but bar should probably not be inlined |
| 01:15:54 | evan | it's only been used once |
| 01:16:02 | evan | it's basically uncommon. |
| 01:16:10 | brixen | I see |
| 01:16:21 | evan | so i'm going to add a threshold |
| 01:16:30 | evan | so that even ICs with a dominator are ignored |
| 01:16:37 | evan | if that dominator isn't actually used much. |
| 01:16:53 | brixen | makes sense |
| 01:18:11 | evan | oh btw |
| 01:18:17 | evan | I added -Xjit.log last week |
| 01:18:22 | evan | -Xjit.log=filename |
| 01:18:34 | evan | and using -Xjit.inline.debug goes to that file |
| 01:18:36 | evan | rather than stdout |
| 01:18:40 | brixen | nice! |
| 01:18:42 | brixen | I missed that |
| 01:18:47 | evan | making it A LOT easier to sort through. |
| 01:18:51 | brixen | indeed |
| 01:19:09 | evan | you can do 'tail -f filename' then to follow things |
| 01:19:18 | brixen | yeah |
| 01:27:00 | brixen | docs are hard |
| 01:28:58 | evan | lets go shopping! |
| 01:29:18 | brixen | heh |
| 01:29:30 | brixen | noooo, I can't take any more xmas |
| 01:29:43 | brixen | I'll suffer with docs |
| 01:29:50 | evan | hah |
| 01:30:05 | evan | hello nari! |
| 01:30:14 | evan | i've been following your tweets! |
| 01:30:34 | nari | hi! |
| 01:40:15 | nari | thanks :) |
| 02:45:40 | rue | Reproducing from earlier |
| 02:45:54 | rue | 1. Everyone should be committing to rubyspec directly. Single point of entry. Also avoids duplication of work (rare as it may be) |
| 02:46:04 | rue | 2. Pull from rubyspec to get the new stuff to rbx, with $ rake spec:update or whatever. Run specs. |
| 02:46:27 | rue | 3. Solve issues caused by other commits to rubyspec head (unless already did before your commit). brixen: Which issues are likely to be encountered, aside from missing guards? |
| 02:46:35 | rue | 4. Automation, including CI, runs off head as well. Avoids sync lag. |
| 02:46:53 | rue | Maybe a boat-sailed, but I would like to understand the problems with phase 3 |
| 02:49:47 | boyscout | Moved spec/frozen to spec/ruby. See below. - cc04522 - Brian Ford |
| 02:49:48 | boyscout | Updated CI specs to RubySpec 9f5238ef. - 2ea8b9b - Brian Ford |
| 02:50:36 | brixen | rue: we can't just run from rubyspec head |
| 02:50:50 | brixen | the issues are tags and specs that cause segfaults and hangs |
| 02:51:35 | rue | Rephrasing 4., ci runs off the spec/ruby retrieved in 2. |
| 02:52:42 | rue | So, provided item 3., the segfaults and such will get fixed/tagged |
| 02:53:14 | brixen | 3rd time a charm? |
| 02:54:19 | boyscout | CI: 2ea8b9b success. 3019 files, 11617 examples, 35755 expectations, 0 failures, 0 errors |
| 02:54:48 | rue | So the idea is that instead of you having to always sync, whoever is committing the spec can do it (or someone can pull it if they know there are changes in there) |
| 02:54:54 | brixen | commits that combine specs and code will be flagged by boyscout |
| 02:55:09 | brixen | pushing spec changes from spec/ruby is easy otherwise |
| 02:55:19 | brixen | syncing spec/ruby with rubyspec is a manual process |
| 02:55:32 | brixen | if you want to run rubyspec head, you're free to do so |
| 02:56:59 | rue | That still leaves us playing with two sets of specs, and sync lag |
| 02:57:19 | brixen | nothing is perfect |
| 02:57:33 | brixen | this is a lot simpler |
| 02:57:38 | brixen | I need fooooood |
| 02:57:40 | brixen | bbl... |
| 02:57:59 | rue | I do not see how it is simpler, it has more moving parts and bottlenecks |
| 03:17:10 | evan | rue: we can't run off rubyspec head |
| 03:17:15 | evan | we've tried that |
| 03:17:19 | evan | it's too disruptive. |
| 03:17:29 | evan | so thats not an option. |
| 05:09:31 | rue | OK. I do not really see how it would be different excepting more people sorting through problems in head |
| 05:10:24 | rue | Not going to press it further, I will see what the flow ends up like. |
| 14:05:40 | bjorn_de_meyer | Say, is nobody here interested in contributing to the Ruby ISO standard? I think it will be important for Rubinius too. |
| 16:36:10 | rue | Heh |
| 16:36:27 | rue | No, we want nothing to do with any standardisation or specification efforts. |
| 17:10:43 | evan | hrm |
| 17:10:47 | evan | my mom needs holiday gift ideas. |
| 17:10:51 | evan | for me. |
| 17:21:11 | scott | thinkgeek.com/*.* |
| 17:24:30 | Defiler | evan: gearjunkie.com |
| 17:38:10 | evan | hm |
| 17:38:14 | evan | a new leatherman would be nice |
| 17:50:06 | evan | ok, gotta help a friend this morning. |
| 17:50:07 | evan | bbiab. |