IRC logs for #baserock for Wednesday, 2015-03-11

*** zoli__ has joined #baserock04:16
*** zoli__ has joined #baserock04:19
*** zoli__ has quit IRC04:20
*** Zara has quit IRC04:22
*** Zara has joined #baserock04:22
*** zoli__ has joined #baserock04:23
*** zoli__ has quit IRC06:35
*** mike has joined #baserock07:32
*** mike is now known as Guest5636307:32
*** gfinney has joined #baserock08:01
*** a1exhughe5 has joined #baserock08:10
*** gfinney has quit IRC08:30
*** gfinney has joined #baserock08:30
*** rdale has quit IRC08:51
*** mdizzle has joined #baserock08:53
*** bashrc_ has joined #baserock09:01
*** ssam2 has joined #baserock09:05
*** ChanServ sets mode: +v ssam209:05
*** rdale has joined #baserock09:05
*** tiagogomes_ has joined #baserock09:18
*** gary_perkins has joined #baserock09:44
*** jonathanmaw has joined #baserock09:46
ssam2i thought it might be an easy job to add an 'include' parameter to lorry-controller, to avoid it importing the whole of baserock/* into Gerrit (since some of those repos are obsolete)09:47
ssam2but then I looked at the code and realised it uses an SQLite database for storage and doesn't have any concept of schema versioning or database migrations09:48
ssam2i think it should probably be illegal to use SQL without having some kind of mechanism for schema migrations09:48
*** edcragg has joined #baserock09:51
ssam2i guess I could just let it import all the repos, then hide the obsolete ones manually.09:53
*** CTtpollard has joined #baserock10:05
*** petefoth has quit IRC10:15
*** petefoth has joined #baserock10:19
* ssam2 discovers that a bunch of settings that we tell people to set in lorry-controller.conf actually have no effect10:29
pedroalvarezoh! like?10:30
ssam2stagger, for example10:30
ssam2'tarball'10:30
Kinnison;(10:31
ssam2also seems like create, destroy, serial, and uuid are all ignored10:32
bashrc_I have a firehose question: why are debian-style version numbers referenced? Does that have something to do with "unpetrify-ref" ?10:49
KinnisonBack when I specified firehose, we needed a way to generate a strict ordering on ref names (particularly tags)10:49
Kinnisonthe decision I came to was that we should use a stable version ordering system (and I selected debian-style) and then transformation regexps to convert tag names into something which sorted well10:49
* Kinnison selected debian-style ordering because it's well understood, well documented, widely implemented, and I happen to be a DD :)10:50
bashrc_I notice that within strata refs are usually sha1's, so is the version number coming from unpetrify-ref?10:52
Kinnisonfirehose isn't retrieving version numbers from the definitions10:52
Kinnisonthe version numbering stuff in firehose is purely related to ordering of *upstream* ref names such as tags10:52
KinnisonIt then determines what there is to do, based on the SHA1s in the history of the selected ref10:52
bashrc_ok I see10:53
* ssam2 realises he didn't need the lorry-controller 'include' feature anyway11:04
wdutchwhen morph can't autodetect a buildsystem the MorphologyNotFoundError is very confusing when first encountered, could I introduce a new error or reword MorphologyNotFoundError to take this into account?11:17
* Kinnison would rather you removed the buildsystem autodetection11:17
Kinnisonbut to do that requires that we first make it possible to specify buildsystem in the stratum (for when there's no chunk morphology) or make chunk morphologies mandatory11:17
wdutchthis would require patching a lot of existing strata11:18
KinnisonAye11:18
SotKwdutch: what is the error message that's confusing ooi?11:19
* Kinnison wants build systems moved out into definitions (instead of in morph's codebase) and so on, but that I think is already being considered by the version-1 team11:19
radiofreei wouldn't like to make chunk morphologies mandatory, for a lot of stuff the autodetection works great11:19
*** tiagogomes_ has quit IRC11:19
wdutchSotK: I have a strata containing a chunk called luci, there is no chunk morphology for luci specified but when I try and build the strata I get: ERROR: Couldn't find morphology: luci.morph11:20
ssam2there's a version-1 team ??11:20
Kinnisonssam2: well there's people discussing it on the ML11:20
*** tiagogomes_ has joined #baserock11:23
SotKwdutch: I'd reword it to mention that it couldn't find the morphology it wanted, and also it couldn't detect the build system and see what people say then11:23
ssam2SotK: that seems like a good solution to me11:23
mauricemoss_I was running into this Morph error as well and found it quite confusing at the beginning. The autodetection works quite well for me.11:23
mauricemoss_SotK, +111:24
petefothDumb question alert: I want to work with my Baserock VM while I'm offline. Usually I ssh in using the ip address of the bridged network adaptor (eth0). But when my host is offline, that adapter doesn't have an ip address.11:26
ssam2petefoth: you added a 'Download' link to the sidebar on the wiki, but 'Get baserock' already links to that page11:26
petefothI also have a 'host-only' adapter (eth1). Do I have to work directly in thew VM console or is there some way I can still ssh in using eth1? (Output from `ipaddr` is at http://paste.baserock.org/apemoperur)11:26
ssam2petefoth: 'Download' is definitely a better name, but we don't need both :)11:26
petefothssam2: oops= failed to ßpot that11:27
Kinnisonpetefoth: providing you can persuade eth1 to DHCP (and you have a DHCP server on your host-only network) then it should be possible to ssh in on that interface11:27
*** wschaller has joined #baserock11:28
Kinnisonpetefoth: If you're using virtualbox then it puts such a server on host-only adapters by default IIRC11:28
petefothKinnison: ta! Any idea what the ip address I need to ssh to would be11:29
Kinnisonpetefoth: it'd depend entirely on what was given to it at DHCP time :(11:30
petefothKinnison: How can I find that out11:30
Kinnisonip addr show dev eth111:31
Kinnisonor simliar11:31
petefothHmm - eth1 stil dpesn't have an ip address :(11:34
bashrc_I may have asked this before but didn't get to the bottom of it. Within the firehose examples tracking section how is the location of the upstream repo defined? I can see the landing repo, but no upstream one11:40
persiabashrc_: In definitions.11:40
persiaAs I understand it, the goal is to read definitions, go find a new foo, ad submit it as a potential change.11:41
persias/ad/and/11:41
persiaSo any information in the format which can be derived from definitions is redundant and confusing.11:41
straycatcode would be much simpler if we had a build-system attribute for each chunk and that had to be in definitions11:43
*** zoli__ has joined #baserock11:43
straycatwdutch, that error was actually reworded from a more detailed err msg, because it's confusing when you mistype a path and morph tells you it can't autodetect a build system11:45
persiaWe should definitely have that for v111:45
straycati think we should have different exceptions if we can11:45
straycati definitely don't want morph to go back to telling me it can't autodetect a build system just because i made a typo11:46
KinnisonPete had an old baserock which had been upgraded repeatedly.  As such he needed a stanza in /etc/network/interfaces for his eth111:46
wdutchstraycat: I agree, a distinct error for each case seems sensible11:47
straycatwdutch, i have a feeling that might be trickier than it sounds, since it might currently be difficult to tell one case from another11:49
straycathopefully i'm wrong about that :)11:50
*** petefoth has quit IRC11:51
*** petefoth has joined #baserock11:52
petefothKinnison: thanks - I'm in :)11:55
radiofreessam2: rsync is gpl312:05
radiofreelooks like the strip-gpl3 script won't remove it?12:06
kejiahuhi, we are trying to cross build a baserock system for armv8b64 on HP Moonshot. I created a wiki page to record what we have done so far, and some sections are still WIP. anybody interested and suggestion/feedbacks are welcome. http://wiki.baserock.org/x86_64_to_armv8b64/12:09
persiaI don't really understand how this is different from any other cross-bootstrap.12:10
kejiahupersia, it is under example of cross-build section. so maybe duplicate in some of the sections. Is the example worth to be recorded on baserock wiki? or should I move it to private logs?12:14
ssam2radiofree: ah, thanks for pointing that out12:15
ssam2i'll update the strip-gpl3 script12:15
persiakejiahu: I tend to think of the wiki as a source of docs, for searching for how to do things, so I'd recommend improving the existing guides if they are confusing/misleading, and publishing reference definitions for anything particularly interesting.12:17
ssam2might be interesting to have a 'success stories' section on the wiki :)12:17
ssam2if they are identified as 'stories' rather than 'documentation' it places less load on the people who take time to ensure the documentation on the wiki is kept up to date in line with changes to morph.git and definitions.git12:19
kejiahusounds a good idea12:23
persiaThat sounds like an interesting idea.  It also provides a way to show off some of the ways people are actually using baserock, rather than just the capabilities.12:23
SotK+1 for success stories12:24
radiofreei want to add new instructions for flashing a jetson12:41
radiofreethey've pretty much entirely changed, should i create a new page of just replace everything in the current one?12:42
ssam2aren't the old ones broken ?12:44
ssam2or at least, worse than the new ones?12:44
petefothradiofree: I'd say replace.12:44
radiofreeit's not broken, this is just an entirely new method that we should encourage people to do12:44
radiofreei.e /boot is now on an ext4 partition12:45
ssam2i think replace12:45
ssam2wiki has history anyway12:45
radiofreethe instructs rely on people cloning a script from my gitlab repo, is that ok?12:45
ssam2as long as you don't suggest they do `curl xxx | bash`12:45
persiaIs the script something that could be a chunk so that it could be put in a devel system?12:46
radiofreeyeah, the instructions would actually be a hell of a lot easier if it was in a baserock system + we added the nvidia tools to a devel system12:46
persiaCan that just be done then?12:47
radiofreewell i'll put these up now, submit patches for baserock, then modify them once thats done?12:47
radiofreethere's nothing to stop you following these instructions on a baserock system12:47
persiaThat sounds reasonable: it lets people critique the new approach early.12:47
radiofreethese also have the added advantage of allowing to to flash a jetson from a jetson12:48
persiaThis is a good thing.  Everything should be self-hosting.12:54
radiofreemason is red12:56
petefothradiofree: at least it's stopped being yellow :)12:57
rjekviolets are blue?12:57
ssam2wow, I love that rsync built in my chroot but failed to build on Mason12:58
radiofree+  morph: strata/tools/rsync.morph, it should be strata/foundation/rsync.morph ?12:58
ssam2oh, hmm12:59
ssam2maybe it built for me because I had that locally12:59
radiofreeprobably, the patch you sent moved it from tools to foundation12:59
ssam2thanks for spotting it, I've pushed a fix13:00
radiofreehmm.. should releases be announced on the front page of http://wiki.baserock.org/ ?13:10
petefothI think that w.b.o shoudl have 'news' pages, and a summary of the most recent should appear on the front page13:15
petefothbut I think we've beenround this loop before :)13:15
KinnisonIf we have news as a set of pages then we could do an inline to make that happen13:15
KinnisonI could set that up if you want13:15
KinnisonOr anyone else with good ikiwiki skills could13:15
petefoth*I* think it would be a really good idea, but I think there should be some form of consensus beofre we do it13:15
SotKpetefoth: Mason was green before the bout of yellows which ended on Monday.. Would it perhaps make sense to indicate the most recent result that wasn't a failure to check for new changes rather than it looking broken?13:15
SotKI also think news would be great13:15
petefothOr there's always the 'better to seek forgiveness' route :)13:15
ssam2SotK: once we upgrade to the Zuul-mason the whole status page will disappear13:16
ssam2I'd like it if we could to add support for continous building of 'master' and generating that status page to the new Mason13:17
ssam2also, seems my fix for rsync didn't actually work!13:17
SotKssam2: that should be trivial, Zuul has support for a timer trigger as well as a Gerrit trigger13:18
SotKhopefully this upgrade can be soon and people can stop worrying every time there is old NONETs from checking for changes at the top of the results list13:19
Kinnisonpetefoth: if you can come up with one or two news items you'd like present, then I'd be happy to help you set up the inlines etc.13:19
petefothKinnison: the only recent news I know of is the release of 15.1013:22
KinnisonOkay, I'll set things up13:22
ssam2ok, rsync built from the same commit on my machine13:22
ssam2using build-morphology, so no local changes at work13:22
KinnisonUnless at least one person here -1s the idea within the next 5 minutes13:22
ssam2oh! cache.baserock.org is probably full13:22
ssam2no..13:23
radiofreehmmm13:23
radiofreecan someone test extracting http://download.baserock.org/baserock/baserock-current-build-system-armv7lhf-jetson.img.gz ?13:24
radiofreei've tried twice now, file-roller tells me "An error occurred while extracting files"13:24
radiofreesha256sum of the download matches13:24
petefothradiofree: I'll give it a try but it'll be in Mac OS13:24
ssam2cache.baserock.org does seem to be giving Mason a 500 error, it's not a problem with rsync13:25
radiofreepetefoth: thanks13:25
radiofree-rw-r-----. 1 james james 1.9G Mar 11 13:23 build-system-armv7lhf-jetson.img13:25
radiofreeaccording to file-roller it should be 2.1G13:25
straycatradiofree, extracts fine here and is 2G13:28
radiofreehmm13:29
petefothradiofree: worked for me - -rw-r-----@ 1 pete  staff  2147483648 11 Mar 13:28 baserock-current-build-system-armv7lhf-jetson.img13:29
Kinnisonpetefoth: Okay, so where on the front page do you want the news inline?13:29
* petefoth thinks13:30
radiofreeok thanks, must be me then13:30
radiofreeusing gzip worked...13:32
petefothI think immediately after the 2nd paragraph (beginning 'Our objectives...'). Everything under that is (relaivley old) news IMHO13:32
Kinnisonpetefoth: I concur13:32
petefothKinnison: I'm sure others will have different opinions however13:33
Kinnisonpetefoth: I'll insert the inline there, and we can see how it looks13:33
Kinnison:)13:33
*** wschaller_ has joined #baserock13:34
* ssam2 tweaks firewall rules and hopes Masons turn green now13:34
*** wschaller___ has joined #baserock13:35
radiofreei'm an idiot, i was trying to extract to /tmp -  tmpfs           1.9G   12M  1.9G   1% /tmp13:36
*** wschaller_ has quit IRC13:39
Kinnisonpetefoth: how does that look to you?13:42
petefothKinnison: looks good to me. I'd also be tempted to add a link in the Sidebard to a 'News' page with *all* the news13:42
Kinnisonpetefoth: please hold...13:42
*** wschaller has quit IRC13:42
Kinnisonpetefoth: where in the sidebar?13:42
petefothKinnison: hmmmm13:42
petefothKinnison: In the 'community' section?13:42
Kinnisonokay13:42
Kinnisontop, bottom, middle?13:42
Kinnison(the inline on the front page is limited to a max of 5 most recent news items)13:42
Kinnisonbtw13:42
petefothKinnison: you decide! :)13:42
KinnisonArgh, the agony of choice13:42
Kinnisondone13:42
petefothKinnison: ta!13:42
Kinnisonpetefoth: and the "news" page also has a postform for creating new news items on the web13:42
* petefoth sits back and waits for any reaction13:42
*** zoli__ has quit IRC13:45
*** paulw has quit IRC13:48
radiofreelooks good13:50
kejiahuhi I got an error when doing native build for armv8b64 within the bootstrap rootfs on moonshot node, can anybody help me to figure out what could be happening? http://paste.baserock.org/rohiniluko13:52
bashrc_the last I saw in emails was that BE on moonshot mostly works but that there are a couple of remaining issues13:54
*** bjdooks has joined #baserock13:56
petefothwhat lives in the /cache/ccache directories - they seem to be empty in my VM?13:56
bashrc_PCI bridge. I think it's being patched13:56
* bjdooks perks up at pci bridge13:57
Kinnisonpetefoth: it's a native-bootstrap phase13:57
bjdooksquestion, are we differentiating between armv7hf, and armv7hf with vfp4-d16?13:58
* petefoth translates that as 'nothing you need to worry about' :)13:58
radiofreei don't think so13:58
Kinnisonbjdooks: armv7lhf is VFPv3-d16 for us IIRC13:59
radiofreebjdooks: armv7hf = --with-fpu=vfpv3-d1613:59
bjdooksKinnison: ok, so it shouldn't be an issue for use on Cortex-A5 based devices13:59
KinnisonThat's why we picked it13:59
Kinnison(general likelihood of being usable)13:59
*** paulw has joined #baserock14:01
bjdooksthanks14:03
radiofreeok, fresh baserock install (downloaded from website), morph init, morph checkout...14:06
radiofreemorph build systems/devel-system...... - error ERROR: Stratum build-essential has no build dependencies for chunk stage1-binutils in string14:06
petefoth'morph edit --no-git-update bash' tries to talk to g.b.o http://paste.baserock.org/otesuwejug (which is odd) and fails (which is not because I'm 'offline')14:06
radiofreeNoBuildDependenciesError: Stratum build-essential has no build dependencies for chunk stage1-binutils in string14:07
radiofreei suppose i need to use latest morph?14:08
SotKI think so14:08
petefothDi I have to do the morph edit while I'm online, and only go offline after that?14:08
petefoths/Di/Do/14:08
Kinnisonpetefoth: unless you already have bash cached, I'd presume so14:08
petefothKinnison: i beleive I do have bash cached, in cache/gits/git___git_baserock_org_delta_bash14:10
petefothbut that doesn't have any source in it :(14:10
radiofreeinstructions for using latest morph are broken, i'll fix it14:10
SotKpetefoth: looks like morph edit doesn't honour --no-git-update14:11
SotKit always runs `git remote update --prune`14:11
petefothSotK: looks like it. IS that a bug or a feature? If it's a bug, I'll raise a story in StoryBoard14:12
SotKI'd imagine its a bug14:12
petefothSotK: ta!14:12
SotKcertainly it will never work if you're offline14:12
radiofreeAppException: In staging area /src/tmp/failed/tmphjN2ag: Command failed: sh -c 'export STAGE2_SYSROOT="$(dirname $(pwd))"14:13
radiofree:\14:13
jjardonradiofree: you beat me editing the wiki by 10 seconds!14:13
petefothSotK: which is a bit of a hole in the 'Use Baserock when offline' story :)14:13
radiofreethat's stage2-busybox14:14
SotKpetefoth: it sure is14:14
SotKshould be fairly simple to fix14:14
* SotK will write a patch tonight if he has time14:14
* petefoth is almost tempted to try and fix it himself14:15
straycatWas there not a suggestion to split edit in something like fetch, and set-ref commands?14:15
straycat*into14:15
SotKI think so14:15
radiofreelog - http://paste.fedoraproject.org/196654/83324142/14:16
radiofreethats a downloaded build-system trying to build a devel-system from master (jetson)14:16
straycatIt does seem a bit nicer, and clearly you wouldn't attempt to do a morph fetch foochunk if you offline, but you might want to morph set-ref foochunk master... or something14:17
straycat*you were14:17
straycats/attempt/expect to be able/14:17
* radiofree nukes his cache14:19
*** CTtpollard has quit IRC14:27
*** CTtpollard has joined #baserock14:36
petefothWould http://paste.baserock.org/afeciyeyan be an acceptable patch to fix the 'morph edit doesn't honour --no-git-update' issue?14:48
ssam2the logic seems to be inverted14:51
ssam2other than that, I think it'd be OK, although I don't know all the code paths that use that function14:51
petefothssam2: you spotted the deliberate mistake :)14:52
* petefoth tries to workout why the patch seemed to work 14:52
SotKpetefoth: I'd be tempted to put the logic in the edit plugin, since the git fat stuff will break too (if there is a .gitfat file in the repo)14:53
petefothSotK: ta! I'll look there14:54
SotKalternatively, duplicate the logic in GitDirectory.fat_pull14:55
petefothbut update_remotes is called from a number of places as well as the edit plugin I think14:55
bashrc_so how do you reference an upstream tarball? tarballs don't contain commits15:02
rdaleyou need to get the tarball lorried so that it becomes a git repo, that you can reference as 'upstream:dialog-tarball' in a stratum15:03
bashrc_sounds complicated15:04
franredbashrc_, there are a lot of examples in the lorry repository15:07
franredbashrc_, http://git.baserock.org/cgi-bin/cgit.cgi/baserock/local-config/lorries.git/tree/open-source-lorries/autoconf-tarball.lorry15:07
franredfor example15:07
straycatpetefoth, did we document conf extensions anywhere? like ATTACH_DISKS, FSTAB_* etc ?15:14
jmacsATTACH_DISKS is documented in the extension code, I think15:15
straycatjust thought it might have been something for the wiki, but probably i'm being stupid15:16
straycat*shrug*15:16
jmacsDon't have a baserock image to check on15:17
straycatATTACH_DISKS isn't even a conf ext anyway15:18
jmacsHmm, it's not actually well documented in writeexts either15:21
straycatheh yeah, i was wondering how to attach multiple disks, i mean it's obvious from the code but still15:23
petefothstraycat: all extension documnetation is in the source code. I mostly worked on write extensions. If you're working with the extension, you can improve the documentationas you go :)15:24
straycatoh okay15:24
kejiahutiagogomes_: have you got this error when you native build armv8l64 under ubuntu before? http://paste.baserock.org/kuwemopize15:29
tiagogomes_kejiahu, you need to mount the Linux API filesystems before entering the chroot15:30
kejiahutiagogomes_: this? http://paste.baserock.org/ololuxufoj15:31
tiagogomes_I used to do more stuff15:32
tiagogomes_PATH=/bin:/usr/bin:/sbin:/usr/sbin15:33
tiagogomes_LD_LIBRARY_PATH=/lib:/lib6415:33
tiagogomes_ldconfig15:33
radiofreeok, build-system downloaded from website, using morph master to build definitions master, on a jetson, is failing15:34
radiofreeclean cache as well15:35
radiofreeconfigure: error: cannot compute suffix of object files: cannot compile15:35
radiofreestage2-glibc15:35
* radiofree tries building 15.1015:36
paulsherwoodinteresting. i had that with ybd...15:36
rdalei think it just means that the compiler is broken15:37
radiofreethat sounds like a problem...15:37
radiofreethis was with a clean cache, built from step 115:37
* petefoth doesn't unsersatnd the line 'PYTHONPATH="$morphpath/morph" "$morphpath/morph" "$@" in the morph script in http://wiki.baserock.org/using-latest-morph/ Why is "$morphpath/morph" there twice?15:37
radiofreepetefoth: because the second one is saying "run $morphpath/morph" the script15:38
SotKshould that not be 'PYTHONPATH="$morphpath" "$morphpath/morph" "$@"'?15:38
jjardonI think so15:39
radiofreeyep15:39
rdalethat page dosn't mention pylru in the latest morph, i'm sure i've seen a very similar page which did15:39
radiofreechanged15:40
petefothprevious version (wht pylru) was PYTHONPATH="$morphpath:$pylrupath" "$morphpath/morph" "$@"15:40
petefothrdale: not needed with the latest morph which has pylru added15:40
* radiofree hopes changing it on his system makes glibc build15:40
jjardonrdale: new release doesnt need pylru, as its already builded in the release15:40
rdalei just installed pylru - i don't see the advantage of keeping a local copy15:40
rdaleah ok15:40
straycatpetefoth, don't see the point in commenting it, because if you can see the code then it's obvious, and it's not like that part can be accessed by morph help <ext>15:40
radiofreestill no luck :\15:41
radiofreeanyone know what the problem might be?15:41
petefothstraycat: fair enough. If the code comments are good enough that's grand15:41
rdaleradiofree: there was a sym link for the dynamic loader that was changed recently - maybe that change doesn't work on arm15:42
radiofreei don't think it's that?15:42
radiofreei'm building everything from stage 1, my host is glibc/gcc 4.915:42
* petefoth now understands why his change to that script broke it :(15:43
radiofreehmmm http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/tree/strata/build-essential.morph#n7315:43
radiofreestage2-linux-api-headers probably shouldn't be in there?15:43
rdaleok, so the sym link won't be there at stage1 glibc15:43
radiofreeoh right, maybe it should15:43
radiofreeso i'm pretty stumped, looks like stage1-gcc is broken15:44
radiofreerdale: you might be right15:46
paulsherwoodradiofree: did you do the 'using latest morph' thing?15:48
radiofreepaulsherwood: yep15:48
paulsherwoodalso are you sure what you have checked out is the branch you think it is?15:48
paulsherwoodsorry to ask dumb qursitons, but these things have bitten me15:48
radiofreei'm on master15:49
radiofreeand master of morph15:49
radiofreehe's the config.log http://paste.fedoraproject.org/196720/2608908715:51
* pedroalvarez can't open that link15:51
radiofreepedroalvarez: should work now15:52
radiofreearmv7lhf-bootstrap-linux-gnueabi-gcc seems to be segfaulting15:52
pedroalvarezI'm still getting 503 trying to open that url15:53
radiofreehttp://paste.baserock.org/fequcegiga15:53
pedroalvarezradiofree: I found an oddity in stage1-gcc some weeks ago, it may be related15:54
radiofreewhat was that and what was the fix?15:56
pedroalvarezradiofree: http://paste.baserock.org/tozequyeme.diff15:57
pedroalvarezbut it was only my workaround15:57
pedroalvarezand I don't think that a patch like this should be merged16:01
*** wschaller_ has joined #baserock16:03
*** wschaller has joined #baserock16:04
radiofreedoesn't work for me anyway16:04
radiofreeERROR: In staging area /src/tmp/failed/tmpW2Cndw: Command failed: sh -c '# GCC is not passing the correct host/target flags to GMP'"'"'s configure16:05
pedroalvarezmeh, my usecase was bootstraping from x86 to arm16:05
pedroalvarezcross-bootsraping*16:05
pedroalvarezbut maybe changing host=armv7a for host=armv7lhf?16:06
pedroalvarezI don't know, but I assume the error is when building stage1-gcc16:06
radiofreei thought armv7lhf was a baserock creation16:07
*** jonathanmaw has quit IRC16:07
*** wschaller___ has quit IRC16:07
radiofreehas anyone tried latest jetson build image to build a devel image?16:07
*** wschaller_ has quit IRC16:07
radiofreeit might just be something wrong with my setup?16:08
radiofreei have built master very recently, however that was on an older devel image that didn't have the /lib/ld-linux.so symlink on it...16:10
pedroalvarezdoes removing the link change anything?16:13
*** jjardon has quit IRC16:14
radiofreethat's what i'm trying now16:14
pedroalvarezwhenever I had ld.so problems, the binary wasn't executable at all, and not seg faulting :/16:16
*** jjardon has joined #baserock16:18
*** Guest56363 has quit IRC16:28
radiofreethis results in an entirely new error16:30
straycatjjardon, wow I didn't realise we could just drop all that busybox delta16:31
radiofreewhat's the "log to stdout" option again?16:33
radiofree'--build-log-on-stdout'16:33
straycatyeah16:34
jjardonstraycat: our delta were only a small patch to fix some includes (and the morph files but they are are now in definitions)16:34
straycatoh i thought there was more16:34
straycatcool16:35
jjardonyep16:39
radiofreeso if i download a baserock image from the website, it should be able to build the same version out of the box right?16:39
radiofreeso i checkout 15.10, the version of morph on their will be able to build 15.10?16:39
* jjardon happy when our delta is 0 and we only have changes in our morph files. This remembers him we should have a solution for .gitmodules16:40
pedroalvarezradiofree: yes...16:40
jjardonradiofree: I think so, yes16:40
radiofreeok, i'll get another jetson and try everything from scratch16:40
* pedroalvarez waits the "that doesn't work"16:40
* radiofree is an optimist and knows it absolutely will work with a clean install16:41
pedroalvarezI need release testers..16:42
radiofreedoes anyone have a jetson build system around from a previous verison?16:42
radiofreei.e the release before http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/commit/?id=a51a8ee9fe2c691a4b2d73fe06c1fde1d18f115316:42
rdalethat was the change i was thinking of earlier16:44
paulsherwoodradiofree: mine is somewhere, i believe, and may pre-date that16:44
paulsherwoodmwilliams_ct: ^^ ?16:44
mwilliams_ctradiofree: Please please please come and take the one off my desk16:45
mwilliams_ctit's been sitting looking sad for ages16:45
mwilliams_ctpaulsherwood: I also have yours, boxed up and ready for your return16:45
mwilliams_cts/your return/return to you16:46
perrylhi all; i submitted a patch fixing my morph protocol version error message which has +2, could someone with push access merge it for me? :)16:51
pedroalvarezyup!16:55
*** a1exhughe5 has quit IRC16:55
straycatjjardon, is cool cause it happens to solve a problem i had with buggy grep -x16:58
straycatso thanks16:58
radiofreesooo16:59
radiofreehttp://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/commit/?id=a51a8ee9fe2c691a4b2d73fe06c1fde1d18f115316:59
jjardonstraycat: np :) . Id like to have the option to use coreutils instead busybox command, but we need to finish the "usrmove" first16:59
radiofreethis breaks the build16:59
radiofreeif i rm /lib/ld-linux.so.3 and rebuild from binutils again it appears to work17:00
jjardonssam2: ^^17:00
radiofreeat least i know how a functional stage1-gcc17:00
radiofrees/know/now17:00
radiofrees/how/have17:00
radiofree:\17:00
radiofreeso i guess the current jetson arm build-system on the website can't actually build anything?17:01
ssam2that breaks the build ??? wow, ok17:03
ssam2any idea how?17:03
ssam2yeah, I guess the last release will be broken on ARMv7 hf then17:03
radiofreeit creates a stage1-gcc that segfaults when you try and run it17:04
* ssam2 slowclaps17:04
*** CTtpollard has quit IRC17:04
ssam2if we remove it, then we're back with the problem that stage1 chunks built on old ARM systems will not run on new ARM systems, but will have the same cache key17:04
radiofreebut i'd appreciate it if someone else tried flashing the current image, then building from a clean cache17:05
ssam2but I guess that's better than it not working at all17:05
ssam2radiofree: can't do today, but i'll try tomorrow or friday17:05
* ssam2 is currently trying to understand how gerrit plugins could possibly ever work, given that the ones up for download on the gerrit website trigger an error when I try to install them17:06
ssam2oh, seems that the downloads are broken if you use 'wget' instead of a browser to download them17:09
* ssam2 claps slowly again17:09
radiofreefrom baserock?17:10
radiofreei'd prefer the downloaded files had the release number in them rather than "current" as well17:11
ssam2no, from https://gerritcodereview-plugins.storage.googleapis.com/index.html17:11
radiofreewhen i downloaded the latest one it wanted to overwrite my existing "current" and i had no idea what version my existing "current" was17:11
ssam2radiofree: you can download the files with the number in if you want17:11
ssam2we link to the 'current' ones to avoid needing to update the wiki, I think17:12
* pedroalvarez was trying to merge perryl's patch for morph in definitions.git17:12
perrylpedroalvarez: oddly, i kept trying to push it to my fork of definitions.git17:12
*** CTtpollard has joined #baserock17:14
perrylthanks pedroalvarez!17:22
pedroalvarezperryl: no worries! thanks for fixing that :017:22
pedroalvarez:)17:22
mwilliams_ctpaulsherwood: radiofree has run off with your jetson. I did all I could to defend it (if by "all I could" you mean I stepped back and enjoyed the extra desk space)17:25
paulsherwoodlol17:26
*** CTtpollard has quit IRC17:26
* straycat wonders why morph's building gcc17:28
radiofreepaulsherwood: you don't mind if i completely wipe yours?17:28
radiofreei'm going to setup one running the latest release, and one running the previous release, both trying to build definitions master17:29
radiofreethe previous release should work, current should fail17:29
straycatignore me, build-log-on-stdout was blinding me17:29
straycator more likely the ci had build and transferred gcc by the time i reran build17:29
straycat*built17:30
*** gfinney has quit IRC17:30
* radiofree just notices there's a brand new, completely unboxed jetson already on his desk....17:30
mwilliams_ctno backs17:30
paulsherwoodradiofree: you lucky person, you17:30
radiofreepaulsherwood: i'll reflash yours with a working version of master tomorrow17:31
radiofreeit needs to build first17:31
paulsherwoodradiofree: ok17:32
ssam2apologies if I did indeed break master on ARM :(17:36
*** ssam2 has quit IRC17:36
*** mdizzle has quit IRC17:51
*** sambishop has quit IRC17:59
*** tiagogomes_ has quit IRC18:01
*** bashrc_ has quit IRC18:01
radiofreea clean 15.02 can't build 15.10 because of "ERROR: Unknown architecture armv8l64 in morphology systems/cross-bootstrap-system-armv8l64-generic.morph"18:10
* radiofree upgrades morph18:10
paulsherwoodradiofree: remove that file18:11
paulsherwood(is simpler)18:11
radiofreeok18:11
jjardonmmm, seems a bit odd that I have to specify in my system a extension that its going to be used only at deploy time (simple-network)18:18
radiofreeso 15.02 should be able to build 15.10, but the resulting image will be useless18:19
jjardonso if I decide to use NETWORK_CONFIG in my deploy morph, I have to rebuid my system and then deploy, instead simply deploy directly18:19
radiofreewell, i think it will be useless, i'm flashing 15.10 to a clean dev board now, so we'll find out if that can build 15.1018:20
radiofreejjardon: that doesn't sound too good18:25
pedroalvarezsimple-network should be added to all the systems once it's back18:43
pedroalvarezBuy yes, we need to change how we enable extensions for a deployment18:44
*** wschaller has quit IRC18:50
jjardonpedroalvarez: rigth, Ive added a story here: https://storyboard.baserock.org/#!/story/2318:59
*** franred has quit IRC19:01
*** edcragg has quit IRC19:23
*** edcragg has joined #baserock19:25
*** edcragg has quit IRC19:40
straycatages ago it was suggested all the extensions should be run by default, but maybe it's better to have them specified in the cluster morph instead20:15
* straycat just did an upgrade for the first time and it just worked :>20:17
jjardonanyone here knows cement? http://builtoncement.com/2.4/dev/quickstart.html I was wondering what would be more difficult: port cliApp to python3 or port morph to cement or similar21:05
straycata lot of morph would have to change if we moved from cliapp21:10
*** gary_perkins has quit IRC21:35

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!