| *** tristan has quit IRC | 03:57 | |
| *** tristan has joined #buildstream | 04:51 | |
| *** ChanServ sets mode: +o tristan | 04:51 | |
| tristan | ironfoot, So I realized something interesting :) | 04:51 |
|---|---|---|
| tristan | http://stackoverflow.com/questions/28890754/eoferror-eof-read-where-object-expected | 04:51 |
| tristan | That is the bug that's happening (apparently fixed in 3.3) | 04:52 |
| tristan | which is causing me to require notparallel | 04:52 |
| tristan | in some of the builds, notably always when it involves running concurrent python | 04:52 |
| tristan | And, the reason why it doesnt happen with YBD, is because I have not _yet_ sealed up the read-onlyness of the rootfs while build commands run | 04:53 |
| tristan | Which, prevent pyc files from being generated | 04:53 |
| tristan | Thinking on that, adding an integration command which does generate the pyc files may (or may not) have a side effect of speeding up builds | 04:54 |
| *** tiagogomes has quit IRC | 09:37 | |
| *** tiagogomes has joined #buildstream | 09:43 | |
| *** ssam2 has joined #buildstream | 09:58 | |
| ironfoot | hmf... so they say it's fixed in 3.3 but we are using 3.5? | 10:14 |
| tristan | ironfoot, doesnt mean it's fixed in anything that uses python2 | 10:17 |
| ironfoot | rrright | 10:17 |
| ironfoot | and in this case it's definitely using python2 | 10:18 |
| ironfoot | gah! | 10:18 |
| ironfoot | I've never seen this error before :/ | 10:19 |
| ironfoot | I don't really understand that bug tracking system, but I believe there are patches also for py2.7 | 10:22 |
| tristan | In any case, we needed to close the gap on read-only-ness at build time | 10:23 |
| ironfoot | at the moment, with BuildStream the build-dependencies rootfs is read-only? | 10:25 |
| tristan | not in master, thats why we got the race conditions | 10:26 |
| tristan | I have been holding off because, I have some polishing I want to do on the sandbox, and, I need to figure out the integration commands problem (which I'm leaning towards solving with fusepy and copy-on-write hardlinks) | 10:27 |
| tristan | even though it doesnt happen really much in practice, and if it does; it wont *really* make the ultimate build result inconsistent, still, if it happens that we write to linked files, we taint extracted artifacts | 10:28 |
| tristan | With copy-on-write hardlinks, we can probably optimize by skipping the artifact extract stage completely, too | 10:29 |
| *** csoriano has joined #buildstream | 10:35 | |
| *** csoriano has quit IRC | 10:38 | |
| *** csoriano has joined #buildstream | 10:48 | |
| tristan | OK ! | 11:25 |
| tristan | https://gitlab.com/tristanvb/defs2bst updated | 11:25 |
| tristan | https://people.gnome.org/~tvb/gnome-build-1.log and https://people.gnome.org/~tvb/gnome-build-2.log | 11:25 |
| tristan | paulsher1ood, now, you can be excited. | 11:26 |
| tristan | ironfoot, you are also allowed to be excited :) | 11:26 |
| tristan | Couldnt get it into one log, and it's (my mistake) a conversion of a little bit older definitions (week or two old) | 11:27 |
| tristan | But I'll wipe the cache and rebuild, bang my SSDs on a fresh version of definitions, which I'm sure, will also build. | 11:27 |
| ironfoot | woOT! | 11:28 |
| tristan | Please no comments on the UI :) | 11:28 |
| * tristan has to get to that | 11:28 | |
| ironfoot | great news, of course :D | 11:29 |
| tristan | ironfoot, so that is just building, now we need to convert / handle the integration commands, perhaps automatically convert the install-files extension with a local source and import element, and then a bootable image deployment | 11:30 |
| ironfoot | yeah.. these bits are going to be fun | 11:34 |
| tristan | install-files is relatively easy, but that manifest needs to be parsed | 11:35 |
| tristan | which means, defs2bst will have to also have access to the definitions repo | 11:35 |
| tristan | (not just the YBD dump) | 11:36 |
| tristan | The deployment will certainly be fun :) | 11:36 |
| *** csoriano has quit IRC | 11:41 | |
| * paulsher1ood yells a muffled w00t! because he's having to bite his tongue about ui | 12:00 | |
| tristan | paulsher1ood, make comments on the UI after I get a chance to improve it :) | 12:01 |
| paulsher1ood | ack | 12:02 |
| * ironfoot tries to bite his tongue and say wOOt at the same time | 12:02 | |
| ironfoot | yep, definitely muffled :D | 12:03 |
| paulsher1ood | :-) | 12:03 |
| * tristan closes shop for the night, heads off to take a shower and go out... will kickoff a build-from-scratch off of a fresher gnome.yml dump | 12:05 | |
| ironfoot | have fun o/ | 12:08 |
| *** tristan has quit IRC | 12:08 | |
| *** tristan has joined #buildstream | 12:17 | |
| *** ChanServ sets mode: +o tristan | 12:48 | |
| *** csoriano has joined #buildstream | 12:52 | |
| *** xjuan has joined #buildstream | 13:34 | |
| *** tristan has quit IRC | 15:10 | |
| *** ghishadow_ has joined #buildstream | 15:15 | |
| *** csoriano has quit IRC | 16:52 | |
| *** xjuan has quit IRC | 17:53 | |
| *** ghishadow_ has quit IRC | 17:58 | |
| *** xjuan has joined #buildstream | 18:05 | |
| *** xjuan has quit IRC | 18:08 | |
| *** ssam2 has quit IRC | 18:08 | |
| *** tiagogomes has quit IRC | 18:23 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!