IRC logs for #baserock for Monday, 2015-10-26

*** gtristan has quit IRC02:23
*** gtristan has joined #baserock03:00
*** gtristan has quit IRC03:53
*** gtristan has joined #baserock04:14
*** petefoth has quit IRC05:11
*** flatmush1 has quit IRC05:11
*** rdale_ has quit IRC05:11
*** gary_perkins has quit IRC05:11
*** rdale has joined #baserock05:11
*** petefoth has joined #baserock05:12
*** gary_perkins has joined #baserock05:12
*** flatmush has joined #baserock05:13
gtristanok, lunch time...06:57
gtristanin the meanwhile... Have we sorted out http support for svn on b.g.o for the WebKit lorry ... yet ?06:57
* gtristan has a handful of other lorry patches uploaded, and refreshed the adding-webkit branch against master, only the WebKitGtk chunk is unsure06:58
* gtristan has a queue of patches that are ready to land *after* WebKit is sorted out06:58
gtristangnome-initial-setup is running nicely, some config issue remains with geoclue and tls (invalid or missing cert)... but works well06:59
* gtristan did see 439e6430538d955b5f9f3c2661d7d0b297a5052b in lorries - is it not safe to say that: An installed svn with http[s] URI scheme support is indeed a hard requirement for a machine running the lorry controller ?07:01
gtristananyway... I suppose people will show up soon while I eat... good morning all ! :D07:01
*** gtristan has quit IRC07:02
*** gtristan has joined #baserock07:53
gtristanjjardon, good morning !07:55
gtristanjjardon, can you help me with a gitmodules situation ?07:55
* gtristan has 2 packages in the queue which use submodules, and needs at least a short term solution for that07:56
gtristanthose are gnutls (should have it's submodule pointing at what is openssl-new on b.g.o)07:57
gtristanand gnome-online-accounts07:57
lostduckgtristan, is it not sufficent to lorry each submodule and modify the .gitmodules file?08:02
gtristanlostduck, that would be the short-term pre-automatic-handling-of-gitmodules approach I think yes08:03
gtristanlostduck, I think that approach works for people with the power to modify those git repositories on b.g.o08:04
gtristanwhich excludes me :)08:04
lostduckyou can always modify a local repo and submit your delta to the mailing list08:04
gtristanI could, yes08:05
gtristanof course there is a little bit of chicken-and-egg situation08:05
lostduckyes, it assumes you've got the lorries approved beforehand, if that's also a problem then i'd suggest deploying yourself a trove08:05
gtristanbut that would be the case anyway08:05
* gtristan looks into that08:07
gtristanI guess I can get a trove that is downstream of b.g.o right ?08:08
* lostduck nods08:08
* gtristan wonders how many hours of cloning that will involve :)08:08
lostduckdepends on your network, but probably "a few" :)08:08
lostduckmaybe we can just give you push access?08:08
lostducksince you're clearly doing lots of very useful stuff08:09
petefothThere's a clear way to do that on, though it will take at least 3 days08:10
* gtristan is looking at wondering if I have to run a whole secure service and such, or if I can get it in a flat directory08:11
gtristanpetefoth, that's not very long really, but I'm not sure what my part is in that :)08:13
gtristan"If two existing committers say that someone should have commit access, and none speaks up against it within 3 days, then that person gets commit access"08:13
* gtristan thinks last week's irc logs indicate jjardon and paulsher1ood "saying" that... 08:14
petefothgtristan: I don't think you have a part :) Yoiu need one of the existing committers to send an email (or ask in here) suggesting you should have access. I'm not a committer myself and I'm not sure who is08:14
petefothCommitters are defined on,members08:17
petefothSo lostducj could propose you if he were so inclined08:17
lostducki think that's a good idea08:19
jjardongtristan: the fact that you need access to the trove is clearly a bug in the tooling; in the meantime, send a request to the ml and i Will +1 it08:21
lostduckit's more of a side effect of our convention that all definitions must be in trove, and i think that's a good thing08:22
lostducknothing stops you from writing definitions that pull straight from github08:23
gtristanright, of course I need to upstream the definitions08:23
lostducki agree that it can be inconvenient though08:23
gtristanregarding CI, we'll also clearly need a better strategy for submodules and identifying downstream patches against upstream modules08:24
lostduckjjardon, are you happy with ?08:25
gtristanlostduck, which also raises an interesting point, even if I were using baserock locally - I would *still* need to use a local trove of some sort08:25
gtristanlostduck, otherwise I would have nowhere to store my downstream deltas against upstream modules (which are _always_ required if you need to really build a fully integrated/working system)08:26
jjardongtristan: merged :)08:26
lostduckgtristan, you could use local repos for any local modifications, if you wanted08:26
jjardonthen you lost the possibility to collaborate, which is the point of all this08:28
gtristanlostduck, true - but it *would* be more pleasant if I could automatically "pull my upstream stable branches and try-rebase my local deltas" all in one convenient step, with the help of the trove :D08:28
gtristanbut true :)08:29
jjardongtristan: Ive just filled!/story/6108:32
lostduckgtristan, in why do we take this from github and not ?08:32
lostduckthere is room for improvement, certainly08:33
jjardonIts a bug; lostduck good catch!08:33
gtristanOh look, I was able to actually leave a comment on story 61 !08:36
gtristanthat doesnt work for story 508:36
jjardongtristan: mind to take a look to when you have a moment ?08:36
gtristanlostduck, I dont believe it exists on git.gnome.org08:36
* gtristan casts another check08:36
jjardonh :)08:37
gtristanlostduck, no good reason :)08:37
gtristansorry my bad08:37
gtristanfew gnome things do only exist on github, looks like that was eventually moved to gnome.org08:38
* lostduck nods08:38
jjardongtristan: im still not sure about splitting webkitgtk and deps out of the gnome stratum08:38
gtristanjjardon, and I'm not sure about ever putting more than one chunk into a given stratum08:38
gtristanbut hey08:38
gtristanso, I could make yet another branch, but really currently what is blocking this is the lorry itself08:40
gtristanrichard_maw, got a long way debugging that situation, we spent energy together also figuring out how the thing should be lorried08:41
gtristannow it seems all we need to do is consider that svn with http support is a hard requirement for lorry and end it at that08:41
gtristanlooks like we came a long way to just fall short of fixing that detail :-S08:42
jjardonwell, that will defeat one of the original baserock objectives, which was avoid the combinatorial explosion of using packages. But its ok if we reach to the conclusion that its actually not possible08:43
jjardonbut in this case not sure if the splitting makes sense: a GNOME system will always have webkitgtk, and its not being used by any other system at the moment08:44
jjardonremember that every time you split a stratum, more difficult is to keep all the stratum in your system definitions08:45
*** tiagogomes_ has joined #baserock08:46
jjardonMaybe this is what we should fix really (automatic runtime dependencies)08:46
gtristanjjardon, I fully expect that that is something that needs fixing (re: defining every single required thing in the system, instead of just a few toplevel dependencies)08:46
gtristanactually, I fully expect that I will be spending lots of time untangling all of these deps into nicely seperated inter-dependand chunk .morph files, but let's see what the future brings08:47
*** tiagogomes_ has quit IRC08:48
jjardongtristan: actually, my top bugs in baserock at the moment is: not conditional in definitions and no automatic runtime deps08:49
gtristanjjardon, in terms of today, in this pre-morphology-refactor day and age; we have 2 blockers: WebKit lorry from svn, and a clear policy to dictate to me about what people expect08:49
*** 7GHABH4Z0 has joined #baserock08:51
gtristanIf we can acknowledge that grouping chunks in stratum the way we do, loses valuable dependency information, makes things rigid to work with, bloats end result builds, etc, and that that needs to be changed post-refactor anyway; then we can move on and not care too much about how the current chunks are organized08:51
gtristanIf we cannot agree on that; then I would prefer a clear message from the baserock maintainership: "Tristan: do it like this"08:51
lostducki agree it's a problem, i know ssam2 is aware of it as well08:52
*** petefoth has left #baserock08:52
*** petefoth has joined #baserock08:53
gtristanI can shove it all into gnome.morph right now; but that means another work load, so it would be nice if at least the baserock maintainership is in clear agreement of how it should be done08:54
lostduckgtristan, which change are you referring to, sorry?08:56
jjardongtristan: probably nobody have a clear idea: I can give you my opinion, witch is (with the _current_ model) putting everything in GNOME as all the packages makes sense together. Im happy to help split everything in small chunks if we agree we have to change this later08:57
lostduckoh i guess you mean
jjardongtristan: Im looking as well to the possibility of using xdg-app, I think it can for some use cases
gtristanlostduck, right08:58
gtristanjjardon, I was just looking at that... I wonder why it messes with ostree though08:59
gtristanisnt that another beast... gnome-continuous uses that right ?08:59
jjardonostree? yes08:59
jjardonthere were/are plans to use ostree in morph as well09:00
gtristanIs "sandboxed applications" really a thing yet ?09:00
gtristanor still just something to give fancy talks about at conferences09:01
jjardongtristan: I make it work in my laptop09:01
gtristanjjardon, ok practical question; this only concerns me for GNOME system: Will this quietly live beside everything else without disruption ?09:02
jjardonand alexl and hertbert are working on it, so I expect only good things09:03
gtristanjjardon, and also: When I am installing applications; which is coming very soon; will I have added system integration headache to do things "specially" for this ?09:03
gtristanjjardon, fwiw; gnome-initial-setup is working very well right now; but I'm not pushing the branch, because gerrit just cant handle it09:04
gtristanand of course; webkit09:04
jjardongtristan: no disruption, I added it there to play with it09:04
jjardongtristan: cant handle it?09:05
gtristan+1 :)09:05
gtristanjjardon, I have adding-webkit branched from master locally, and I have integrating-gnome-initial-setup branched off of that09:06
gtristanI rebase the other branch off of 'adding-webkit'09:06
lostduckgtristan, no opinion on that change, probably best to wait for others09:06
gtristanI can push 'adding-webkit' and gerrit eats it up, of course we can't push the tip of that09:06
*** bashrc has joined #baserock09:06
lostduckwas worth a try :309:07
* perryl needs to check which terminal has focus...09:07
*** mariaderidder has joined #baserock09:08
gtristanjjardon, anyway, should I push my other topic branch, there will just be more unmergable patches in the queue, and it will also mistakenly move the webkit branch to the new topic branch09:08
jjardongtristan: for my experience in gerrit; its often useful to group patches together to avoid the merge conflict mess; push the braches, at least they can get reviews09:08
jjardonI will try to help with the merge conflict later ;)09:09
jjardongtristan: I was thinking on create a gnome-apps stratum, to separate from the core, as we do in GNOME upstream: are you ok with that?09:10
gtristanOk; How can I push _only 3 patches_ as a separate topic branch; which depends on the merge of 'adding-webkit', so that the new 3 patches or so show up as a separate topic branch ?09:10
*** franred has joined #baserock09:10
gtristanjjardon, that's what I was going to do; as soon as we get webkit/gnome-initial-setup behind us09:11
gtristanjjardon, before that, I want to add nautilus and gnome-terminal and gnome-control-center to the same core gnome group09:11
gtristanactually I want to do that *before* default enabling gdm09:11
lostduckgtristan, if your branch is already based on the HEAD of adding-webkit then an ordinary git-review should suffice, assuming your current branch isn't the same branch you originally submitted from09:11
jjardongtristan: Yeah, makes sense09:12
jjardongtristan: do we agree on putting webkitgtk in gnome stratum then? we can open the topic in the mailing list about  the problems/advantages of the strata model09:13
*** paulwaters_ has joined #baserock09:16
gtristanjjardon, it will probably be a single patch at this point09:17
gtristanjjardon, I've reorganized them twice already... this time we might get this done _today_ if I just add all of the WebKit related changes directly into gnome/ as one patch09:18
gtristanand I'm sorry but; it looks like we'll need a WebKitGtk-tarball lorry in interim09:19
jjardongtristan: im ok with that09:19
gtristanuntil someone fixes the svn installation on b.g.o09:19
*** nowster has joined #baserock09:20
gtristanok, will do09:20
gtristanjjardon, can you fix the submodules issues for gnutls and gnome-online-accounts for me ? (or did you already ?)09:20
*** jonathanmaw has joined #baserock09:20
jjardongtristan: yes, give me 10 min09:20
gtristanthat way I will just update my second branch to refer to the baserock delta09:21
gtristansure, no worry, I'll anyway do the webkit change first09:21
*** toscalix__ has joined #baserock09:21
jjardonall depends of how good the internet connection in this train is :)09:21
*** ssam2 has joined #baserock09:22
*** ChanServ sets mode: +v ssam209:22
gtristanjjardon, gotta love wifi at 300 kmph :)09:25
jjardongtristan: this is uk, we are far away from that :)09:26
* gtristan gets decent free wifi on the highspeed train in korea09:27
rjek300,000 mph?09:27
*** 7GHABH4Z0 is now known as tiagogomes__09:27
rjekThe "Fast, Reliable WiFi from T-Mobile" on the UK's 200km/ph trains is neither fast nor reliable, and more expensive and less reliable than just using your phone's 3G.09:28
*** ssam2 has quit IRC09:35
gtristanugh... this is more tricky09:38
gtristaneh, should be ok09:39
*** CTtpollard has joined #baserock09:51
*** mariaderidder has quit IRC09:58
*** ssam2 has joined #baserock09:59
*** ChanServ sets mode: +v ssam209:59
gtristanIn the meantime... I've posted
gtristanjjardon, that will let us get past this until we get g.b.o updated with an svn which handles http schemes properly10:10
gtristanwhich... we should figure out too :)10:10
gtristanat this point, makes no sense to block on it anymore, though10:10
jjardongtristan: nice, thanks10:11
jjardonnot sure I can help you with the submodules thing though; not able to clone the repos with this internet connection10:11
richard_mawwow, the webkit tarball is pretty big10:12
*** mariaderidder has joined #baserock10:12
richard_mawlooks well formed though10:12
* richard_maw votes10:12
richard_mawssam2 beat me to it10:13
* gtristan has a huge rebuild to do10:16
gtristanI will submit the webkit patch, and the few other patches for gnome-initial-setup to work, after I re-verify that they build10:16
gtristanprobably a few hours10:16
jjardongtristan: maybe makes sense to put gnutls in the strata/network-security.morph stratum10:17
* gtristan senses another round of arbitrary stratum choosing10:22
gtristanjjardon, honestly right now I had it in libsoup, but agree it has no business there; as libsoup's dependency is only runtime; and having a TLS enabled glib-networking is enough10:22
gtristanjjardon, but at the moment I feel that we will move more quickly by moving this directly into gnome stratum10:23
jjardongtristan: sure10:23
gtristancracklib should also sit below linux-pam and shadow10:24
pedroalvarezour definitions are growing really fast, I feel like we need to think a bit about this10:25
gtristanbut all of this (again) will be much more dynamic and various setups supportable by untangling all of this10:25
jjardonin core? that will make you rebuild some things :)10:25
gtristanpedroalvarez, ranging freely, my organic chicken ?10:25
pedroalvarezgtristan: happily laying eggs!10:26
gtristanpedroalvarez, jjardon ok I promise a full proposal by the end of this week, I only have a writeup now and I want to put it though some theoretical tests first10:28
jjardonpedroalvarez: I still think they are growing very slow, if you compare with what it should be (lots of people creating systems)10:29
gtristanI will A.) Identify the problems I believe the refactor aims to solve... B.) present the model I have in mind, and C.) present the use cases I've been able to think about and show how the approach stands up to those10:29
gtristanC is basically what needs work10:30
pedroalvarezjjardon: yes, but that means that we might not be ready for a situation in where we have loads of contributions10:31
gtristanonce we provide stable branches as actually reliable reference builds, I think we dont have to worry about loads of contributions10:33
gtristanerr, s/once/until10:33
gtristananyway, we can consider that as a lucky thing, gives us time to work out kinks :)10:34
*** gtristan has quit IRC10:40
tiagogomes__oh the partition stuff was merged10:58
*** gtristan has joined #baserock10:58
*** locallycompact has joined #baserock10:59
pedroalvarezif anything is broken, let me know11:00
jjardonTo support a new board, I need patches for a _very_ active branch of linuc that it's not in upstream repos. Is it OK to lorry that repo?11:02
richard_mawjjardon: which repo is this?11:04
pedroalvarezFor raspberry-pi I was planning to just push a branch to the current linux.git repo11:05
ssam2jjardon: i think we have the same problem with u-boot already ...11:10
ssam2so, i don't really like how messy is getting, but we should lorry it if you will get blocked otherwise11:10
ssam2we have 3 u-boots already:
richard_mawssam2: is it the big list of duplicated repositories that bugs you, or the fact that we don't get to share subtrees?11:22
richard_mawssam2: since I've been looking at git namespaces recently, and I think we could work out a way to share the git objects there, but it would probably require some changes to the git server11:26
ssam2richard_maw: both11:26
gtristanjust a thought: afaics we generally need to lorry that linux separately... OR... lorry needs to be taught how to streamline branches from various repositories into the same git repo on git.baserock.org11:27
ssam2yeah, Trove is kind of designed for the "one repo per project", but that doesn't really hold water for Linux11:27
richard_mawgtristan: aye, that would fix needing to make new copies, though it wouldn't help to remove our existing ones in a compatible manner11:28
ssam2one big problem we have with Trove is that, to avoid breaking historical builds, all the existing repo URLs need to keep working forever11:29
ssam2s/historical/old/ :-)11:29
gtristannod, just a thought, I suppose it's rather rare, but what with github trends, forking seems to be more of a thing than it used to be11:29
ssam2yeah. We did try once look at bunging Gitlab into Trove as the git server, but it didn't prove to be practical11:30
richard_mawssam2: yep, so we need a way to deprecate the repository and either redirect to the new version (which could be in a namespace of another repository) or we could hide the old repositories from listings, somehow mark them read-only and use git-relink to deduplicate any shared objects they had11:31
ssam2i think redirects makes more sense11:31
richard_mawhm, needs support in the git server then11:31
ssam2and show them in cgit as sort of greyed out repos, or some such thing11:31
richard_mawwith some changes to the git server we could deduplicate the objects from the u-boot and linux forks by migrating their content into namespaces of the canonical upstream mirror, and when you try to fetch from the old URL, the git server translates that into exec'ing the recieve-pack etc. command11:34
richard_mawwith GIT_NAMESPACE=u-boot-socfpga (or whatever) set11:34
ssam2'the git server' being Gitano, or git-daemon ?11:35
ssam2or the lighttpd-git systemd service?11:35
richard_mawthem, also cgit and whatever serves the http URLs11:35
ssam2all of it ! right, makes sense11:36
* richard_maw will write this up properly in the storyboard and send to the list11:36
*** rdale has quit IRC12:00
*** paulwaters_ has quit IRC12:13
*** paulwaters_ has joined #baserock12:19
*** paulwaters_ has quit IRC12:31
gtristanjjardon, after a couple hours of rebuilding everything... webkit passes configure phase and is building:
jjardongtristan: great! :)12:54
*** toscalix__ is now known as toscalix12:54
tiagogomes__can anyone else have a look at ?13:04
pedroalvareztiagogomes__: I take that the tests are currently broken?13:06
richard_mawtiagogomes__: reviewing13:06
tiagogomes__pedroalvarez at least on my machine they are, but I am not seeing anyone else complaining13:07
pedroalvarezhat's why I was wondering :/13:08
richard_mawtiagogomes__: reviewed13:08
richard_mawtl;dr: Use sorted(), it's presumably a non-determinism from filesystems returning entries in arbitrary order13:08
richard_mawor convert the lists into sets13:08
tiagogomes__okey dokey13:10
*** paulwaters_ has joined #baserock13:35
pedroalvarezradiofree, jonathanmaw: do you remember why this was needed?
jonathanmawpedroalvarez: I do not.13:36
pedroalvarezIt's hard to answer review comments, or send patches upstream without knowing what is the fix about13:36
* radiofree can't remember13:43
radiofreei think without it, it will create a lib, so.113:43
radiofreesome other project that uses it fails to run because it links against .so13:43
radiofreei won't be able to tell you what the other project is though13:44
pedroalvarezhm.. I think that library is created by pers_common, and with this `sed` we fix pers_client_library13:45
pedroalvarezI'll try to make them revert this change:
gtristanjjardon, cool !14:40
gtristanjjardon, can we get one for ?14:41
gtristanjjardon, in the trove, it's openssl-new14:41
gtristanI wont push the patches tonight, it's closing on midnight - but we'll have a working gnome-initial-setup tomorrow14:42
gtristanhopefully... maybe by wednesday, I can get nautilus, gnome-terminal and g-c-c built and functional, enable gdm by default and announce a second milestone14:44
gtristanthen it's apps and beyond :)14:44
rjek2/win 2014:47
jjardongtristan: nice :) I've sent a patch to add gtk2 and ruby to the GNOME system; we missed that when adding webkitgtk to the GNOME stratum14:47
richard_mawwhat does GNOME want ruby for?14:48
gtristanjjardon, ah, I missed that when collapsing the WebKitGtk stratum14:48
gtristanrichard_maw, WebKitGtk wants it14:48
richard_mawat runtime, or does it just have build tooling written in ruby?14:48
gtristanwell, GNOME wants webkitgtk to have it for some reason at least:
ssam2i think some of the webkit build stuff is Ruby14:50
ssam2rdale went through similar pain for QtWebkit I think14:50
lostducktiagogomes__, wrt you need the next change in the series which fixes the tests, or are you saying it fails with that?14:58
tiagogomes__lostduck no, I missed that it needed the other change15:01
lostduckahh ok15:02
tiagogomes__jjardon, you'll have to rebase the patches on the xdg-app topic15:29
*** paulwaters_ has quit IRC16:04
lostducktiagogomes__, nice spot on the other change, if your -1 of 1213 is no longer current can it be removed?16:18
tiagogomes__lostduck I am reviewing it at the moment16:19
*** locallycompact has quit IRC16:24
franredpedroalvarez, I think we shouldn't do this - we don't apply patches in baserock and try to modify sources a minimum as we can. This changes made in this chunk are massive for me and I think we should put them in a branch (no only the git apply, the sed is also very intrusive codewise)16:35
franreds/for me/ from my point of view/16:35
pedroalvarezI know, did you see my comment?16:37
pedroalvarezapologies for that patch, it wasn't my work, and I didn't realise about all of these problems before sending for review16:38
pedroalvarezI'm getting there though16:38
pedroalvarezI'm now trying to figure out if we need any gtk in the GDP16:38
jjardonpedroalvarez: if you need it, Im sure gtk3 is the one you need; gtk2 doesnt support wayland16:41
franredpedroalvarez, yes, I saw the comment :) - in any case, I think we should change it and try to have the repo as clean as we can, because if we add it as it is, someone could take it as an example16:43
pedroalvarezI totally agree, I missed that hack :)16:43
pedroalvarezjjardon: my guess is that I don't need it at all16:43
pedroalvarezI wonder if radiofree or jonathanmaw knows anything about gtk being present in the GDP system16:44
jjardonpedroalvarez: yeah, I think so too16:44
jonathanmawI don't think the GDP has gtk16:45
jonathanmawi.e. I grepped the manifest for the GDP system and the only match was "navit-graphics-gtk", which isn't going to be much use without gtk16:45
pedroalvarezgood, ta!16:46
pedroalvarezjjardon: regarding your comment here, I still don't know what to do with it (
jjardonpedroalvarez: do we really need weston-genivi and weston-genivi-gdp ?16:48
pedroalvarezthey are the same (I believe) except for the "WESTON_NATIVE_BACKEND" hack16:49
jonathanmawpedroalvarez: I'd be cautious about changing the version of weston in the GDP. there's a bit of version hell between weston, wayland-ivi-extension and the GDP16:49
pedroalvarezwhich if we remove, i don't know how to make it work out of the box in VM's16:49
pedroalvarezjonathanmaw: yes, I'm aware of that. I'm not changing that version, but the version of the baseline to match16:50
jjardonstill, why do we need 2 "genivi" weston ? can we use only one and remove the other?16:50
pedroalvarezjjardon: I can do that, but that doesn't solve my main concern (WESTON_NATIVE_BACKEND)16:52
jjardonpedroalvarez: keep the version with the hack?16:53
pedroalvarezI wonder if it's possible at runtime to determine if you are on (e.g.) a real machine or a VM16:54
pedroalvarezto execute weston with fbdev or drm16:54
pedroalvarezso it won't be that hacky16:55
jjardonnot sure ,but probably systemd can provide you that info in some way?16:55
pedroalvarezhehe, this patch series will never end :)16:56
pedroalvarezok, I'm going to keep the "hack" for now, and try to address other remaining bits that I know how to solve16:56
*** paulwaters_ has joined #baserock16:56
jjardonpedroalvarez: yeah, Id file file a story in storyboard and keep going16:57
pedroalvarezright, they are a bit different:
pedroalvarezmain difference is that one creates and enables a weston.service17:08
pedroalvarezthis could go to a configure extension I guess17:08
franredpedroalvarez, and 2 install files?17:09
*** franred has quit IRC17:22
*** jonathanmaw has quit IRC17:23
pedroalvarezfranred: hmm.. no, everything in the configure extension17:36
*** ssam2 has quit IRC17:36
*** nowster has quit IRC17:40
richard_mawhm, I didn't expect it to take all day to write up how I think we can deprecate lorries, but I dove into details, and it looks workable17:54
*** bashrc has quit IRC17:56
pedroalvarezI would say that we can remove the working area for the disabled repositories17:56
richard_mawwe can probably already hide a repository by running `ssh config delta/u-boot-socfpga.git set project.archived true`17:56
pedroalvarezhiding them from cgit is just remove a file from the repo I believe17:57
pedroalvarezbut `gitano ls` will show it17:58
richard_mawpedroalvarez: I understood that gitano is the one that decides whether to generate that file17:58
richard_mawproject.archived will hide it from ls unless you do ls --all17:58
richard_mawah, hrm, the git-daemon-export-ok doesn't look at project.archived18:00
richard_mawwhich is because that file is also a flag to git-daemon to decide whether clients can clone the repo18:01
richard_mawcgit has other ways of determining which repositories to show18:01
richard_mawaha! you could do this by having gitano set hide=1 in check_local_git_files if project.archived is set18:07
* richard_maw wonders if he has mis-configured his e-mail18:11
*** mariaderidder has quit IRC18:12
*** mariaderidder has joined #baserock18:12
richard_mawhmm, repo.hide may be in a newer version of cgit than we have18:13
* richard_maw calls it a day18:13
*** gtristan has quit IRC18:14
*** gtristan has joined #baserock18:15
pedroalvarezthis is baserock, upgrade everything! :)18:16
*** mariaderidder has quit IRC18:19
*** paulwaters_ has quit IRC18:21
* Kinnison notes that project.archived is fairly new stuff and so far only hides it from ls18:21
Kinnisonnot also from cgit18:21
*** toscalix has quit IRC18:36
richard_mawKinnison: yep, seems like it would be reasonable to hook it up to write hide=1 when generating cgitrc files21:42
* richard_maw is a fan of project.archived already21:42
pedroalvarezI keep wondering how to get u-boot in the raspberry pi work with a extlinux.conf file21:57
pedroalvarezWell, I can add support for config.txt (rpi bootloader config file) too, and fix system-version-manager to not depend on extlinux.conf being present22:08
pedroalvarezI feel like we need to think about BOOT_DEVICE flag, our new partitioning feature, Jetson flasing scripts, and Upgrades22:09
richard_mawyeah, it's a big lump of functionality that has grown over time without being rationalised22:11
richard_mawplus we invented that wheel because we didn't think we needed the complexity of other tools, so we should have a look around at what the alternatives are, since potentially it would be a useful subproject to split out22:12

Generated by 2.14.0 by Marius Gedminas - find it at!