IRC logs for #baserock for Friday, 2016-03-11

*** edcragg has quit IRC01:08
*** franred has joined #baserock05:37
*** lachlanmackenzie has quit IRC05:56
*** thrace_ has quit IRC05:56
*** lachlanmackenzie has joined #baserock06:02
*** thrace_ has joined #baserock06:02
*** gtristan has quit IRC06:03
*** gtristan has joined #baserock06:12
*** lachlanmackenzie has quit IRC07:10
*** thrace_ has quit IRC07:10
*** lachlanmackenzie has joined #baserock07:21
*** thrace_ has joined #baserock07:21
*** lachlanmackenzie has quit IRC07:47
*** thrace_ has quit IRC07:47
*** lachlanmackenzie has joined #baserock07:54
*** thrace_ has joined #baserock07:54
*** bashrc has joined #baserock08:58
*** rdale has joined #baserock09:03
*** bruce_ has joined #baserock09:04
gtristanmorning rdale ... can I trouble you for a build log or two ? they dont need to be fresh, assuming you have a cache handy...09:04
*** edcragg has joined #baserock09:04
rdaleyes ok09:05
gtristangreat :)09:05
gtristanI'm interested in coreutils and tar build logs09:05
gtristansome reason it looks like I'm building a bad coreutils, it breaks the build of sed, diff and tar in the same way09:06
gtristanbuilding sed without build-depending on coreutils passes fine :-/09:06
gtristanso hopefully I'll get a hint09:06
rdalei don't seem to have got either of those in my minimal test system09:08
gtristanoh ? your not building coreutils-common ?09:08
gtristanrdale, curiously... what is your minimal test system, is there a systems/ target ?09:09
* gtristan was shooting for foundation09:09
rdaleminimal-system-x86_64-generic.morph - which doesn't have foundation - i can add it and build09:09
gtristanif its not too much trouble, I could use the build log...09:10
gtristanI'll let you know if I figure it out... also I might lower the bar on my side and shoot for minimal-system myself09:11
* gtristan wasnt aware that there was a deployable target without foundation.morph09:11
rdaleit still needs to build quite a lot of stuff which doesn't appear in the final system, but that's why it was a good test for artifact splitting09:13
gtristanI see... there is minimal-system-armv5l-openbmc-aspeed which is a suitable target for me to try and deploy actually09:13
pedroalvarezmorph sed:
gtristanpedroalvarez, :)09:14
pedroalvarezmorph tar:
gtristanpedroalvarez, and coreutils would be nice... I can see that with sed you dont have the broken cat error (when I configure sed with a staged coreutils, I get loads of 'cat -: No such file or directory')09:15
pedroalvareznote, all of them built with morph, but they should be pretty similar09:17
gtristanyeah they should be09:18
gtristanI was worried about coreutils line 4067 from your coreutils paste, but seems that it happens there as well09:19
pedroalvarezi've seen that one in various places now09:20
pedroalvarezbut my guess is that they fall back to something else to copy the file09:21
gtristanodd, well I looked at the bootstrap script, it does cp -p (preserve attributes if possible), which is a valid option for busybox cp :-/09:22
gtristanthere are 2 cp invocation, one without options and one with -p09:23
gtristanafter " Creating po/Makevars from po/Makevars.template ... " and before " bootstrap: done.  Now you can run './configure'. "09:23
* gtristan is in his second run of tar... this time after fixing his PATH which may have effected the coreutils build a bit09:24
gtristanpedroalvarez, curiously... do you capture stderr in that log ?09:26
gtristanI found that ybd was not capturing stderr, in my build I do09:26
gtristanand have some warnings that dont appear in your log09:26
gtristanstuff like: src/dd.c:965:14: warning: assuming signed overflow does not occur when simplifying conditional to constant [-Wstrict-overflow]09:26
pedroalvarezI believe morph captures stderr09:28
gtristanIs build-system-armv5l-openbmc-aspeed built regularly ?09:29
pedroalvarezin fact, that busybox error goes to stderr I think09:29
gtristanit *could* just be that coreutils-common is armv5l broken09:30
pedroalvarezit could... :/09:30
gtristanI would think that coreutils-common is built regularly for armv7lhf though, right ?09:31
pedroalvarezthe onl;y architectures we build everyday are armv7lhf, x86_64 and x86_6409:31
pedroalvarezand yes, including coreutils-common09:31
gtristanso it's quite doubtful that this is an arch bug09:31
gtristanok here comes the build error09:32
gtristanyup, many... cat: -: No such file or directory09:32
gtristanin configure, followed by: Error: your shitty compiler cannot even create executable, you idiot09:32
gtristanprinted to the console :)09:33
gtristanI checked that though, it doesnt get as far as invoking gcc09:33
gtristanits erroring in something like cat confdefs.h - <<_ACEOF ... some code09:34
gtristanand then prints that message, but doesnt invoke gcc to try to compile anything09:34
rdalei can't seem to get ybd to build coreutils - it is giving me: [build-essential] Autodetected build system is NOT FOUND which seems a bit of an off error09:38
gtristanhmmm, I seem to recall having one of those...09:45
gtristanat one point09:45
rdaleah - found the problem - i had a hacked non-working version of - it's now building normally09:46
gtristanI suspect your build logs lack stderr as well09:47
gtristanshould probably file a bug for that09:47
*** jonathanmaw has joined #baserock09:53
*** franred has quit IRC09:56
*** N-a-N has joined #baserock09:57
rdalei still can't get ybd to work - i'm getting an error about a missing build_step even with the master10:04
*** cosm_ has quit IRC10:04
gtristanodd, how did this happen ? wasnt ybd master / definitions master working ?10:06
gtristanrdale, this is only when trying to build more than minimal-system ?10:06
rdalei think it will go wrong even when i don't have coreutils added as a stratum10:07
rdalethis is the error: there has been some discussion about adding build steps recently10:09
*** franred has joined #baserock10:37
*** tiagogomes has joined #baserock11:08
*** tiagogomes_ has quit IRC11:11
*** tiagogomes has quit IRC11:58
*** tiagogomes has joined #baserock12:10
*** fay_ has quit IRC13:27
*** fay_ has joined #baserock13:27
*** fay_ has quit IRC13:42
*** gtristan has quit IRC13:44
paulsherwoodrdale: have you specifiedbuild system? Autodetect is tecghnically no longer supported?13:51
* paulsherwood should make 'Autodetected...' a WARNING13:52
rdalethe problem was my fault - when i added coreutils-common to my minimal system, i got the syntax wrong, and ybd thought it was part of build-essential13:52
paulsherwoodah, ok13:52
paulsherwoodwell, not ok. still need to improve the definitions handling so there's more obvious feedback on user error13:53
rdalethen networking isn't working in my kvm, and i haven't had time to sort that out yet13:53
rdaleyes, the error was confusing, but i don't like the 'morph: .. name:' overloading when specifying stratum artifacts in a system13:54
paulsherwoodme neither.13:54
rdalethat is very error prone syntax apart from being ugly13:54
paulsherwoodi want to propose a load of definitions changes... but there's no way morph will keep up imo13:54
* rjek raises an eyebrow at paulsherwood13:55
paulsherwood(there's a challenge to those folks who still think morph is the bees knees)13:55
persiaCan changing the filename template from .morph to .def be one of those?13:56
rjekpersia: Why only three letters?  We don't need compatibility with Windows 3.1113:56
*** fay_ has joined #baserock13:57
persiarjek: Because ".defin" and ".define" don't parse as well for me, and I don't want to type ".definition", and I find that I say "defs" often enough in speech that I seem to have preshortened it.  Note that I'm more interested in the unisyllabic part than the trigraphic part.13:58
rjekWe have tab completion!13:58
rjekOr, we could use .yaml :)13:58
rdaleor .yml14:00
persiaI can use magic to know it is yaml: I want more semantic information, so someone could write a patch so that magic could tell me "Component Defintion (YAML)" or "System Definition (YAML)"14:00
* persia prefers .yaml to .yml for semantic clarity14:01
* paulsherwood prefers .yaml now too14:01
* richard_maw also prefers .mdwn to .md, since .md is a GCC machine description file, but Git{Hub,Lab} assume .md is markdown14:02
rdalei suggested .yml because that is what rails uses, i don't know which extension is most common14:02
paulsherwoodwell... i suggest we default to yaml, but support all the others too for a while (.def, .morph, .yml) :-)14:03
persiaPlease not .yml14:03
persiaBut why not .def?14:03
rdalei think .yaml is good because it is boring14:03
* persia sees no value to adding ".yaml" to a YAML file, because the magic already knows about YAML internals14:03
rjek.yml is totally artifical.  If something postdates 1995, there is no reason to contract extensions.14:04
richard_mawI think the bikeshed should be magenta14:04
*** richard_maw has left #baserock14:04
persia+1 for magenta14:07
rjekBut which shade?14:07
rdalei don't think getting names right is bikeshedding, it is difficult and takes a lot of time14:10
*** jmacs has joined #baserock14:17
*** gtristan has joined #baserock14:27
pedroalvarezi thought we agreed having 2 build tools was useful14:37
paulsherwoodpedroalvarez: it has some advantages, yes.14:38
jjardonpedroalvarez: about ;  did you have any time to investigate which plugin is causing the problem?15:48
pedroalvarezjjardon: the only thing I did was to try stock weston, and it did work15:48
pedroalvarezI don't have any ideas about how to debug the problem further :(15:49
jjardonpedroalvarez: by stock weston you mean the weston from genivi-weston without any plugin or the one from weston-common ?15:52
pedroalvarezgenivi-weston, without any weston.ini15:53
jjardonpedroalvarez: does it fail with both ivi-shell-weston.ini and gdp-weston.ini?15:55
pedroalvarezjjardon: hm.. I can't remember, let me check if I still have that image installed16:00
pedroalvareznope :/16:02
pedroalvarezI mean, I don't have it installed, so I can't check16:04
*** franred has quit IRC17:07
*** edcragg has quit IRC17:51
*** jonathanmaw has quit IRC17:52
*** bashrc has quit IRC17:53
rdalegtristan: ybd coreutils build log:    tar build log:
paulsherwoodjjardon: ?20:18
*** tiagogomes has quit IRC20:21
*** N-a-N has quit IRC20:26
*** bruce_ has quit IRC20:30
*** Guest24701 has joined #baserock23:35
*** Guest24701 has quit IRC23:36

Generated by 2.15.3 by Marius Gedminas - find it at!