| *** persia has quit IRC | 05:51 | |
| *** persia has joined #buildstream | 05:52 | |
| *** tristan has quit IRC | 06:01 | |
| *** tristan has joined #buildstream | 06:23 | |
| gitlab-br-bot | tristanvb opened MR !1389 (tristan/exit-on-nonblock-terminal-1.2->bst-1.2: _frontend/cli.py: Exit with error if output streams are set to nonblocking) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1389 | 07:29 |
|---|---|---|
| *** bochecha has joined #buildstream | 07:43 | |
| benschubert | Hey tristan , would you have time to look at https://gitlab.com/BuildStream/buildstream/merge_requests/1388 ? | 08:11 |
| *** ChanServ sets mode: +o tristan | 08:44 | |
| tristan | Ah thanks | 08:44 |
| *** jonathanmaw has joined #buildstream | 09:06 | |
| *** raoul has joined #buildstream | 09:21 | |
| *** tristan has quit IRC | 09:26 | |
| jjardon | hi jonathanmaw ; would it be possble to tag another release of bst-external when you have a second, please? | 10:03 |
| *** tristan has joined #buildstream | 10:22 | |
| *** bochecha_ has joined #buildstream | 10:56 | |
| *** bochecha has quit IRC | 10:57 | |
| *** bochecha has joined #buildstream | 10:57 | |
| *** bochecha_ has quit IRC | 10:59 | |
| jonathanmaw | okie doke | 11:17 |
| jonathanmaw | jjardon: tag pushed https://gitlab.com/BuildStream/bst-external/tree/0.14.0 | 11:47 |
| laurence | is the website https://docs.buildstream.build/ relevant to 1.2 stable? or master? | 11:49 |
| *** ChanServ sets mode: +o tristan | 11:52 | |
| tristan | laurence, 1.2 | 11:52 |
| tristan | laurence, except we noticed that we never backported some of the adjustments we made to the docs at the time of creating the website | 11:53 |
| tristan | so oddly, the install guide now appears twice (once on the website where we moved it to, and again in the buildstream docs themselves) | 11:54 |
| tristan | laurence, we were discussing that it might be time to revisit https://gitlab.com/BuildStream/buildstream/issues/178 | 11:54 |
| laurence | i was thinking of a quick patch to master to explicitly mention the docs reference an unstable version - but actually i've not seen them, they might mention it | 11:57 |
| laurence | what's the url for master docs? | 11:57 |
| tpollard | I don't like that the install from source will default to bst master on https://docs.buildstream.build/ | 11:59 |
| tristan | laurence, master docs are still at https://buildstream.gitlab.io/buildstream/ | 12:08 |
| tristan | tpollard, Do they ? I think they clearly indicate that you should choose the right tag | 12:08 |
| tristan | at least I recall making sure they mentioned that | 12:08 |
| tristan | we even have the badges which indicate the latest release tag | 12:08 |
| tristan | tpollard, https://buildstream.build/source_install.html#install_git "Before installing..." | 12:09 |
| tpollard | I was referring to https://docs.buildstream.build/ | 12:10 |
| tristan | tpollard, that's not supposed to have any install instructions | 12:11 |
| tristan | <tristan> so oddly, the install guide now appears twice (once on the website where we moved it to, and again in the buildstream docs themselves) | 12:11 |
| laurence | right, so we need to fix that, remove it from 1.2 | 12:12 |
| tristan | tpollard, the changes we made to master docs at the time of creating the website need to be backported to the branch so that the website displays what we intended | 12:12 |
| laurence | i may be a little ignorant here, but: 1.2 is now frozen in time, and should just generate the stable docs that the website shows, so we just backport to the 1.2 branch | 12:12 |
| tristan | laurence, That is more or less correct yes: The docs published on docs.buildstream.build are published from the bst-1.2 branch, and yes as I said, need to backport those specific changes to the website | 12:15 |
| tristan | laurence, I recall quite specifically that we did not care about backporting website/docs related things after releasing 1.2 | 12:15 |
| tristan | because we published all of that from master, and since we were never going to break API, it was going to be more or less fine (we'd just see new APIs appear with "Since: 1.4" and know that that is still not available) | 12:16 |
| laurence | i see what you're saying....i think, ha | 12:18 |
| laurence | do we have a firm list of the backports ? seems like a nice low hanging fruit if we do | 12:19 |
| laurence | and i think that a note at the top of the master docs to say they are master / unstable and anyone looking for stable 1.2 should go to <insert url> | 12:20 |
| tpollard | I think it'd be nice to have everything under one url, with master under something like 'latest' and then specific versions under their own etc | 12:21 |
| tristan | laurence, found it | 12:24 |
| tristan | https://gitlab.com/BuildStream/buildstream/merge_requests/872 | 12:24 |
| tristan | 3738dd06eeb59b8d275f59890dc812e00d274ff0 | 12:24 |
| tristan | That's precisely the point in time where we finished migrating the install guide to the website and out of the repo | 12:25 |
| tristan | tpollard, that is https://gitlab.com/BuildStream/buildstream/issues/178, indeed | 12:25 |
| tristan | tpollard, for the documentation itself, it should have a link to every stable version + master ideally, and you should be able to choose | 12:25 |
| laurence | why do we not want master to have an install guide though? it'd be different to stable (no ostree, etc) but master still needs to have an install guide, no? | 12:25 |
| tristan | sigh | 12:26 |
| tristan | second round of same conversation for me today | 12:26 |
| laurence | tristan, sorry, just point me to it | 12:26 |
| tristan | laurence, there are two sides of this... one is: The install guide is not the installation documentation, although they are currently the same | 12:26 |
| tristan | laurence, on the phone | 12:26 |
| tpollard | tristan: thanks for that link | 12:26 |
| tristan | laurence, I mean.. the conversation happened on the phone, I'm not on the phone right now :) | 12:27 |
| laurence | tristan, ha, got that :) | 12:27 |
| tristan | So... the install guide is intended for the innocent newcommer who needs to get off the ground, ideally it should just say "use apt-get install buildstream" | 12:27 |
| tristan | And, at the time, we didnt expect so much churn, so we figured it would be fine to keep the install from source at that location too | 12:28 |
| tristan | laurence, Point being, the instructions to install 1.4 were not really going to be different from 1.2 or 1.0 | 12:28 |
| tristan | Now at this stage, since things are diverging, it might make sense to have advanced instructions about installing in the repo, separated by version | 12:29 |
| tristan | i.e. with master you now need cython and compilers depending on your install scenario | 12:29 |
| tristan | laurence, However, the principle remains that: The website should give you an easy guide, for people who just want to use it right now - and deep(er) in the source code you should be able to find out how to hack your way through getting the thing built and installed from upstream source code | 12:30 |
| tristan | that is mostly for hackers like us, not for end users | 12:30 |
| laurence | alright, yes this is clear | 12:31 |
| tristan | traditionally that goes into a text file called INSTALL, but meh, it's 2019 and we don't mind having it in rst and rendered in pretty ways I guess | 12:31 |
| laurence | ok, seems time for a solution to https://gitlab.com/BuildStream/buildstream/issues/178 | 12:33 |
| tristan | laurence, at this point, actually solving it seems like less effort than re-discussing it every few months :) | 12:39 |
| laurence | tristan, surely it's solved problem already, somewhere | 12:41 |
| laurence | (and don't call me Shirley!) | 12:41 |
| tristan | We can work around it in some way | 12:43 |
| tristan | laurence, https://gitlab.com/BuildStream/docs-website | 12:43 |
| tristan | That is a repo which we use to publish the docs on docs.buildstream.build | 12:44 |
| tristan | laurence, it currently yanks the built documentation from the latest built artifacts of bst-1.2 and shoves them into docs.buildstream.build | 12:44 |
| tristan | perhaps it could yank *both* and shove them *both* there, while generating a toplevel selection page | 12:44 |
| tristan | a choose your poison page which lets you choose which docs to view | 12:45 |
| SotK | is there a reason we're self-hosting the docs rather than just using readthedocs? | 12:46 |
| SotK | (readthedocs handles versioning for you) | 12:46 |
| tristan | Another conversation I think we had many months ago | 12:51 |
| tpollard | I'm looking at the subproccesing of stream entry points into the scheduler, has anyone used the python deco package before? | 12:51 |
| tristan | SotK, I think readthedocs wants to build it for us, and/or impose constraints on the docs you publish | 12:52 |
| tristan | I don't recall exactly what it was | 12:52 |
| tristan | I was never fond of the idea of publishing our docs there though | 12:52 |
| tristan | SotK, https://gitlab.com/BuildStream/buildstream/issues/178#note_58726525 ... looks like it doesnt let you run custom build commands | 12:54 |
| tristan | like running BuildStream to collect the output and render them in the docs as we do | 12:54 |
| tristan | it wants to be in control | 12:54 |
| SotK | ah, I didn't realise that you need to be able to run buildstream to build the docs, ignore that suggestion then :) | 12:59 |
| laurence | irc monthly team meeting starting in 3 minutes | 13:58 |
| laurence | over on #buildstream-meetings | 13:58 |
| gitlab-br-bot | marge-bot123 closed issue #1045 (Overnigth tests are failing: "No matching distribution found for setuptools>=36.6.0") on buildstream https://gitlab.com/BuildStream/buildstream/issues/1045 | 14:13 |
| gitlab-br-bot | marge-bot123 merged MR !1388 (bschubert/fix-overnight->master: tests: Build wheel before installing BuildStream in overnight tests) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1388 | 14:13 |
| jjardon | jonathanmaw: thanks!! | 14:52 |
| *** tristan has quit IRC | 14:58 | |
| laurence | Ah, i notice the upstream bug in marge bot has been closed earlier today - https://github.com/smarkets/marge-bot/issues/188 | 15:02 |
| *** bochecha has quit IRC | 15:04 | |
| tpollard | laurence: nice! | 15:09 |
| *** shash has joined #buildstream | 15:20 | |
| *** tristan has joined #buildstream | 15:28 | |
| jonathanmaw | hmm, I think I've painted myself into a corner while trying to define what state the frontend should have access to. I've ended up creating a `State` class that holds all this state, and can have callbacks assigned that will report changes to this State, but I've come to the conclusion that the function calls to State to change its data would end up sending the exact same data to the frontend by calling the callbacks, too | 16:02 |
| jonathanmaw | so now I'm having trouble seeing any benefit to this plan that I originally made | 16:02 |
| jonathanmaw | i.e. at the moment I could manage just as well by either not actually holding any information in State, or giving the frontend direct access to the State object | 16:03 |
| jonathanmaw | (with process separation, the equivalent would be giving State shared memory and locks | 16:04 |
| jonathanmaw | ) | 16:04 |
| WSalmon | juergbi, not buildbox related but enbeded realated, if i want the sand box to alwasy run as spesific uuid, can i set it in my element? | 16:20 |
| juergbi | WSalmon: you can specify the uid, yes: https://buildstream.gitlab.io/buildstream/format_declaring.html#sandbox | 16:22 |
| WSalmon | thanks | 16:22 |
| juergbi | you can set the default in a project and you can override it per element | 16:22 |
| WSalmon | thanks | 16:22 |
| gitlab-br-bot | tpollard opened MR !1392 (tpollard/shellbuildtree->master: _frontend/cli.py: Tweak non-interactive bst shell buildtree handling) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1392 | 16:40 |
| *** jonathanmaw has quit IRC | 17:07 | |
| *** raoul has quit IRC | 17:37 | |
| *** bochecha has joined #buildstream | 18:54 | |
| *** slaf_ has joined #buildstream | 19:01 | |
| *** slaf_ has joined #buildstream | 19:01 | |
| *** slaf has quit IRC | 19:03 | |
| *** slaf_ is now known as slaf | 19:03 | |
| *** shash has quit IRC | 20:47 | |
| *** bochecha has quit IRC | 22:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!