IRC logs for #baserock for Monday, 2017-03-13

*** Guest45613 has joined #baserock00:19
*** Guest45613 has quit IRC00:32
*** gtristan has joined #baserock03:09
*** ctbruce has joined #baserock07:53
*** paulwaters_ has joined #baserock08:19
*** rdale has joined #baserock08:37
*** jonathanmaw has joined #baserock08:55
paulsher1oodgtristan: i approved
paulsher1oodcould you take a look at (and particularly, advise how buildstream will handle this usecase)08:59
*** toscalix has joined #baserock09:00
gtristanohhh, ok so thats the thing... now I'll have to support migrations of it :-S09:21
gtristanpaulsher1ood, so binary inputs are supported already by buildstream from a long time, basically we support it by simply stating that you use an 'import' element, which can use any 'source' implementation to "put files into the artifact"09:22
gtristanwe only support currently ostree, git and 'local' source types (currently relying on ostree for the bases we've used so far)09:22
gtristaniiuc, this requires some kind of artifactory source implementation ?09:23
gtristan"YBD can now handle binary inputs from urls, as well as git inputs" this sounds very abstract09:23
gtristanbinary inputs can be tarballs, can be anything, most anything is retrieved by a url, including gits09:24
gtristanusernames and passwords are better suited in user configuration, rather than project data (just a randomly related comment), hopefully they can be avoided, too09:25
*** rdale has quit IRC09:39
*** rdale_ct has joined #baserock09:39
paulsher1oodi agree about the usernames and passwords... but rjek pointed out that there may be situations where multiple auths are required, per component09:49
paulsher1ood(seems a bit daft to me, tbh)09:49
paulsher1oodthe implementation i've done is *supposed* to work with jrandom (un+)authorised url09:50
paulsher1oodso not just artifactory09:50
rjekOr multiple different artifactories09:50
gtristanWhat we want from a source implementation (which is pretty much the same for any build input you'd want for YBD, plus one added feature) is basically:09:52
gtristan  o An alias-able url09:52
gtristan  o An exact identifier, usually implemented as a checksum09:52
paulsher1oodwe discussed checksums earlier, and concluded that we need to support up to two per component09:53
gtristan  o A symbolic tracking reference, allowing us to get the latest exact identifier for that tracking reference (usually a symbolic branch or tag name)09:53
gtristanthat last is more than what ybd normally supported, but I suppose is similar to what ybd now does with the 'sha'/'ref' distinction09:54
gtristanA source should be able to fetch remotely the latest exact identifier on demand and return it (from Source.track()) and it should be able to set it onto a core provided dict with Source.set_ref()09:55
gtristancore uses this to rewrite projects inline09:55
gtristanSo anyway... I think method of obtaining and mirroring sources should be implementation specific09:56
gtristanmeaning, we'd probably want one for artifactory and another for other things09:56
paulsher1oodgtristan: no, this is for binaries - if we're downloading a blob, rjek argued that md5 is insufficient, so this new implementation supports md5 + sha25609:56
gtristanSo this is download a data blob, much like what we'd do for the tarball source09:57
gtristanexcept with no tar extract09:57
paulsher1oodi expect that instructions on what to *do* with the blob are in *-commands09:58
gtristanThat I think is same for everything09:58
paulsher1oodunclear how to checksum for case of url =
gtristanI mean, when you get that blob, the way you currently use it... is I suppose it appears in the build dir09:58
gtristanwith a name09:59
*** CTtpollard has joined #baserock09:59
* paulsher1ood is just commenting here, in case anything in this use-case is surprising for buildstream09:59
gtristanSo with a Source, we either use it as a Source for a BuildElement derivative (where we'll use commands to manipulate it)09:59
gtristanor we use an import (like a literal "put the content of this here" statement)09:59
paulsher1ood(and checksumming?)10:00
gtristanfwiw, in semi-related news, today the conversion gained support for the integration commands10:00
paulsher1oodw00t :)10:00
gtristanand a build of gnome passes, without requiring root at all, even with the system integration commands10:00
* gtristan thought there might have been chowns in there, but there are only groupadd/useradd)10:01
paulsher1oodwhole of gnome, including build essential and device nodes?10:01
gtristanpaulsher1ood, right for the checksumming what we want for tarballs, is basically automatically update the exact identifier with an sha256 sum on demand (with bst track)10:02
gtristanSo the user doesnt have to memorize shas10:02
gtristanand of course, no symbolic refs10:02
gtristanpaulsher1ood, device nodes are omitted in the base, that's a part of the manual port I needed to do for build-essential10:03
gtristanso I still have to try to boot it, but it should not be a problem, udev or systemd will create the nodes10:03
* gtristan out for dinner, back in a bit...10:09
* paulsher1ood looks forward to confirmation that it boots :)10:09
* gtristan has to write the deployment for that... coming this week10:09
gtristancurrently I've whipped something up for the splits10:09
gtristanwhich, were a sort of a trainwreck imho10:10
gtristanbut still pertinent feature to have, wielded properly10:10
*** gtristan has quit IRC10:15
*** tiagogomes has joined #baserock10:16
*** tiagogomes has quit IRC10:18
*** ssam2 has joined #baserock10:24
*** ChanServ sets mode: +v ssam210:24
*** tiagogomes has joined #baserock10:24
*** jude__ has joined #baserock10:29
*** jude_ has quit IRC10:30
*** gtristan has joined #baserock10:48
* ironfoot ponders the deprecation of,, and openid.baserock.org11:01
paulsher1oodyou're right to raise this11:09
ironfootWe are in the process of renovating the SSL certs, and I went through the services we currently have11:11
*** ctbruce has quit IRC11:14
ssam2fine by me11:14
ssam2would be nice to dump the storyboard database somewhere for reference though11:14
ironfootYeah, I was pondering only swiching them off, but I can try to extract that information first, yes11:15
ironfootAlso, it looks like we won't need the database and mail VM's either11:16
ironfoot(we might if we decide to selfhost gitlab in the future)11:16
ssam2they're fully represented in the Ansible scripts in baserock-infrastructure.git so nothing to lose from destroying them now if we don't need them11:17
ssam2can redeploy within an hour or so from the Ansible scripts11:17
ironfoothehe yes, unless django package upgrades, and breaks openid11:18
ssam2if we need an OpenID provider, Ipsilon is probably better to use now11:19
ssam2although the docs are truly awful11:19
ironfootright, will start announcing changes soon via baserock-dev11:19
ssam2but I know the maintainer who says it is actually possible to link in a simpler password storage system than Kerberos or LDAP11:19
*** jude__ has quit IRC11:21
*** ctbruce has joined #baserock11:26
*** jude__ has joined #baserock11:35
*** noisecell has joined #baserock11:47
*** ssam2 has quit IRC13:22
*** ssam2 has joined #baserock13:38
*** ChanServ sets mode: +v ssam213:38
*** ctbruce has quit IRC13:54
*** ctbruce has joined #baserock13:55
*** noisecell has quit IRC13:57
*** noisecell has joined #baserock14:01
*** gtristan has quit IRC14:59
*** noisecell has quit IRC15:07
*** gtristan has joined #baserock15:14
*** noisecell has joined #baserock15:32
*** tiagogomes has quit IRC15:48
*** leeming has quit IRC15:48
*** noisecell has quit IRC15:49
*** ssam2 has quit IRC15:49
*** CTtpollard has quit IRC15:49
*** jude__ has quit IRC15:49
*** jonathanmaw has quit IRC15:49
*** ctbruce has quit IRC16:01
*** jude__ has joined #baserock16:01
*** CTtpollard has joined #baserock16:01
*** ctbruce has joined #baserock16:01
*** leeming has joined #baserock16:02
*** ssam2 has joined #baserock16:02
*** ChanServ sets mode: +v ssam216:02
*** noisecell has joined #baserock16:02
*** tiagogomes has joined #baserock16:02
*** jonathanmaw has joined #baserock16:04
*** jude__ has quit IRC16:12
*** leeming has quit IRC16:26
*** jude_ has joined #baserock16:32
*** leeming has joined #baserock16:40
*** jude_ has quit IRC16:41
*** jude_ has joined #baserock16:46
*** noisecell has quit IRC17:05
*** toscalix_ has joined #baserock17:21
*** toscalix has quit IRC17:23
*** toscalix_ has quit IRC17:28
*** jude_ has quit IRC17:43
*** jonathanmaw has quit IRC18:01
*** ctbruce has quit IRC18:38
*** ssam2 has quit IRC18:57
*** jude_ has joined #baserock19:04
*** paulwaters_ has joined #baserock19:45
*** gtristan has quit IRC21:19
*** rdale_ct has quit IRC22:47

Generated by 2.15.3 by Marius Gedminas - find it at!