IRC logs for #baserock for Wednesday, 2015-12-16

*** gtristan has quit IRC02:02
*** gtristan has joined #baserock02:09
*** Stanto has quit IRC03:09
*** gtristan has quit IRC04:20
*** gtristan has joined #baserock04:46
*** gtristan has quit IRC05:32
*** gtristan has joined #baserock06:20
*** VLetrmx has quit IRC07:24
*** VLetrmx has joined #baserock07:25
*** toscalix has joined #baserock08:05
*** toscalix has quit IRC08:28
*** toscalix has joined #baserock08:28
*** ctbruce has joined #baserock08:31
gtristanSotK, can you please file a bug with the GNOME upgrade issues you encountered ?08:33
gtristanI havent been doing upgrades at all, and for the short-term I'll just forget08:33
*** rdale has joined #baserock08:50
*** edcragg has joined #baserock08:53
SotKgtristan: sure, I'll do that tonight08:58
SotKits probably exacerbated by the fact I'm upgrading a really old system that isn't in the reference definitions08:59
gtristanpossible, but that's still good to be documented09:01
gtristanbetter to know what we can and cant do :)09:01
pedroalvarezmy guess is that the upgrades are not quite working for this usecase09:06
pedroalvarezwhich is sad09:06
pedroalvarezI would say that the gnome system is not working at all because it doesn't have any of the users that it needs09:07
*** tiagogomes_ has quit IRC09:08
gtristanpedroalvarez, it will be years before it actually has "users"09:08
gtristanpedroalvarez, I dont think that qualifies as "working" though, is there any system in definitions that has users ?09:09
gtristanright now working = it boots from the image and the components actually work, it *does* that09:10
pedroalvarezy meant unix users09:10
gtristanI has plenty of users in that sense09:11
gtristanadm, netdev, avahi, pulse, gdm, ....09:11
gtristanany missing users can be filed as bugs, that's totally in the scope of what I've been doing these months09:12
gtristannot that I will know about it, because... you know... storyboard doesnt send emails...09:12
gtristananyone know how to build a python module in baserock ?09:13
gtristanlike say... python-packages/rdflib for instance ? actually particularly ?09:13
paulsherwoodgtristan: can you run easy_install?09:22
*** tiagogomes_ has joined #baserock09:22
paulsherwoodactually, no.... that ckind of defeats the point, i guess09:23
gtristanI dont know, I was just looking in python-common.morph and there is a handful of types of python build-systems09:31
gtristanas I havent added any python module yet, I figured someone like jjardon (hint) who knows so much about changing everything to python3... would "just know" how it's supposed to be done :D09:32
paulsherwoodjjardon is on vacation today i think09:32
paulsherwoodpedroalvarez may know09:32
gtristanwell, "python-distutils" did something09:33
gtristanwhether it's the correct thing remains to be seen09:33
gtristanbut it was successful09:34
VLetrmxif the question is what build-system flag do you use for python modules packaged with setuptools/distutils then yes it's python-distutils09:35
VLetrmxsee the DEFAULTS file for a full list of build-system definitions, also feel free to define your own09:35
pedroalvarezi kind of assumed this "python module" couldn't be build using python-distutils09:36
VLetrmxwhy?09:36
pedroalvarezbecause the answer "use python-distutils" was really simple :)09:37
VLetrmxoh ha, yes :)09:37
pedroalvarezbut yes, never had to use anything different to build python components09:38
VLetrmxindeed python modules are usually packaged with setuptools/distutils09:40
*** ssam2 has joined #baserock09:49
*** ChanServ sets mode: +v ssam209:49
*** locallycompact has quit IRC09:53
*** franred has joined #baserock10:04
*** Stanto has joined #baserock10:18
*** ssam2 has quit IRC10:41
*** locallycompact has joined #baserock10:47
*** ssam2 has joined #baserock10:55
*** ChanServ sets mode: +v ssam210:55
gtristanSo that didnt work11:12
gtristanbuilding zeitgeist... it does this: http://paste.baserock.org/hefakogebu11:12
gtristanbuilding rdflib with python-distutils did not make that check pass :-S11:13
ssam2could it be expecting /usr/bin/python to be python2 not python3 ?11:14
gtristanssam2, there is no question about that... /usr/bin/python is *always* python211:15
gtristanI dont think there is any discrepency there11:15
SotKI thought /usr/bin/python was python3 in baserock these days?11:15
gtristanSotK, that would be a terrible mistake afaics11:16
gtristanpython apps which use python3 ask for it explicitly11:16
VLetrmxSotK, do you still have a link to that PEP that explains that?11:17
* SotK looks11:17
gtristanit's not11:18
SotKhttps://www.python.org/dev/peps/pep-0394/11:18
gtristanSotK, a chroot into the zeitgeist artifact shows 'python --version' gives the correct 2.711:18
VLetrmxyeah that's the one i think11:18
VLetrmxi kind of thought that it'd be better to have python be python2 for the sake of backwards compatibility, but we seemed to not reach an agreement last time11:19
gtristanI see whats going on11:19
gtristanVLetrmx, it's not something for us to really decide11:19
gtristanVLetrmx, installing python3 as /usr/bin/python would just be asking for trouble, running python2 apps with a python3 interpretor11:20
gtristanso anyway, rdflib installs properly11:20
gtristanbut succeeds in compiling/installing without a an isodate module installed11:21
VLetrmxgtristan, iirc the pep leaves it open, since any script using #!/usr/bin/python should be compatible with both, but yeah having python linked to python2 would seem to make the most sense for backwards compatibility right now11:21
* gtristan gets typical knee-jerk annoyance of python modules not asserting their deps properly at configure time11:21
VLetrmxone of the talks at nixcon was also about how awful packaging in python is, this seems to be a popular subject at the moment :311:25
gtristanhttps://bugzilla.gnome.org/buglist.cgi?bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&bug_status=NEEDINFO&bug_status=RESOLVED&bug_status=VERIFIED&list_id=83691&query_format=advanced&short_desc=python3&short_desc_type=allwordssubstr11:34
gtristanIf python2 programs 'just worked' with python3, I suspect that query would not have so many results11:34
gtristanactually, 'port+python3' reports almost the same count11:35
VLetrmxi don't follow, are you saying you want python to be linked to python2 always?11:36
ssam2at EuroPython this year, someone asked Guido van Rossum what his least favourite thing about Python was. didn't take him long to answer: "anything to do with packaging"11:41
gtristanVLetrmx, I am simply stating that python2 is not python3, and its not a safe assumption that a python2 program will work with python3 or vise versa11:55
gtristanif you ask my opinion, yes: I think that even in a world/system with absolutely no python2 programs present, the interpretor should still be installed at /usr/bin/python311:56
gtristanvice versa, even11:57
VLetrmxoh right, then i agree, usually python2 programs break because the 2 languages handle strings differently, str in python2 is used for text and binary data, where as python3 adds a separate type for binary data12:02
edcraggthere's a bunch of incompatibilities, isn't there?12:25
gtristanedcragg, that's my understanding anyway, gedit folks told me they had months of headaches "porting to python3" so, it can't be a drop-in replacement12:26
VLetrmxstring handling's the only thing i've encountered when porting stuff to python3, but there's other stuff here https://docs.python.org/3/howto/pyporting.html12:28
edcragghave good test coverage and see what breaks!12:29
*** mwilliams_ct has joined #baserock12:29
gtristanAny clue what's going on here: http://fpaste.org/301623/26956214/ ?12:48
gtristanContext: compiling zeitgeist12:49
gtristanautogen passes, zeitgeist seems to compile OK on my host12:49
gtristantried various branches but same error when building in baserock12:49
paulsherwood-j?12:50
gtristanno12:50
gtristanI tried that12:50
paulsherwoodor possibly a corrupted cached artifact?12:50
gtristanit's like, the first thing it does12:51
paulsherwoodor possibly out of memory?12:51
gtristanram ?12:51
gtristannah12:51
paulsherwoodi guessed not12:51
paulsherwoodmissing dependency?12:51
paulsherwoodcan you share your definition? i can try it elsewhere?12:52
gtristanpossibly... when installing rdflib (python) on my host, it installed python-pyparsing and python-somthing-about-sparql12:52
gtristanI tried building the python-pyparsing we have in openstack and stage that, no joy12:53
gtristanI can, I have a huge branch right now12:53
gtristanI can push the branch up as a wip to g.b.o12:53
paulsherwoodcool12:54
* gtristan hands paulsherwood wip/tristan/empathy-integration12:55
gtristanoh12:55
gtristansorry, you wont be able to easily build it :-S12:55
gtristanthere are 9 unmerged lorries in gerrit12:56
paulsherwoodgtristan: how do you deal with that? using file:/// ?12:56
gtristanI usually use the upstream git12:56
gtristanexcept for tarballs12:57
gtristanin which case I run lorry locally, clone the workd/<foo>/git... and point to that with file:///12:57
gtristanthere is one tarball lorry12:57
gtristanand... you wont like knowing what it is hehe...12:57
gtristanWebKitGtk1-tarball !!12:57
paulsherwoodwhy do you need *any* tarballs?12:57
gtristanFor WebKitGtk, same reason as before, it's svn with a crazy branching scheme iirc12:58
paulsherwoodgtristan: why not have gnome-apps as a separate stratum? gnome.morph is getting humungous :)12:59
gtristanso 2 reasons, the lorry (at the time at least) did not support the lorry syntax we needed, and, more importantly, svn does not support https there12:59
gtristanpaulsherwood, because then we would lose the dependency info, so I wanted to avoid that at least for now12:59
paulsherwoodgtristan: ah, ok12:59
paulsherwoodperryl: i've been thinking about your reproducibility work...13:01
paulsherwoodif there was a pipeline building components, you'd maybe want to publish the checksum(s) for reproducible components, and then (maybe) periodically force rebuild (some of) them to check that you still got the same thing?13:02
*** franred has quit IRC13:03
paulsherwoodand somehow dashboard the result?13:03
SotKISTR ssam2 doing some experimentation towards such a process earlier this year13:04
paulsherwoodbut i'm assuming that there must be existing dashboard solutions for doing fancy displays of emitted json content (which is i guess how the results would be made available?)13:04
paulsherwoodSotK: i believe perryl was involved in that too :)13:04
paulsherwood(assuming you mean http://wiki.baserock.org/projects/deterministic-builds/)13:05
perrylpaulsherwood: yes, the current system is very inefficient; it requires manual running and the results displayed is a start but...well, it could be a lot better :)13:06
*** tiagogomes_ has quit IRC13:06
paulsherwoodso could most of the software in the known universe :-)13:06
SotKpaulsherwood: I was thinking of the "Continuous builder for Baserock" section on that page, yes13:07
paulsherwoodSotK: ack13:07
paulsherwoodactually, i didn't notice those patches to morph-cache-server... they could be re-purposed into kbas, maybe13:08
paulsherwoodperryl: do you know if there is an actual instance of that variant of morph-cache-server anywhere?13:09
perrylpaulsherwood: not as far as i know, think ssam2 may be better to ask?13:10
ssam2no, i took it down ages ago13:13
ssam2easy to set up again though if there is interest13:14
paulsherwoodssam2: i'm mainly interested to see what its output looks like - i can probably infer it from the code13:15
ssam2the machine is still there, just shut off13:15
*** franred has joined #baserock13:17
ssam2http://185.98.148.127:8080/13:17
paulsherwoodgreat, thanks!13:19
paulsherwoodactually, looks like all of the artifacts there were not actually bit-for-bit?13:20
SotKpaulsherwood: there are some bit-for-bit ones in the later pages13:20
paulsherwoodah, ok13:20
ssam2if you look at the numbers, most were fixed early on, as well13:21
paulsherwoodtvm13:21
*** gtristan has quit IRC13:23
*** franred has quit IRC13:25
*** tiagogomes_ has joined #baserock13:36
*** gtristan has joined #baserock13:39
*** toscalix has quit IRC13:44
paulsherwoodperryl: i don't know if http://riemann.io is the right thing, but that's the *kind* of thing i'm thinking of for publication of results, and it seems concourse are already using it13:48
perrylpaulsherwood: great stuff, i'll take a look at it after lunch and see how well it'll gel with the repro test output13:49
*** tiagogomes_ has quit IRC13:51
*** tiagogomes_ has joined #baserock13:53
paulsherwoodmaybe this would interest locallycompact - riemann is written in clojure :)13:55
*** toscalix has joined #baserock13:59
* locallycompact wonders what the hell he's looking at13:59
*** trn has quit IRC14:01
paulsherwoodlocallycompact: how other folks process events from ongoing services... for baserock, services could include (eg) g.b.o, artifact cache, ciat etc14:04
paulsherwoodfor exampel i've been collecting elapsed build times as build logs from ybd on an ad-hoc basis for a while (https://github.com/devcurmudgeon/build-logs)14:05
paulsherwoodbut it would be more interesting to collect that kind of info as event streams i think14:05
paulsherwoodor maybe not :)14:06
*** trn has joined #baserock14:12
*** trn has quit IRC14:33
*** trn has joined #baserock14:33
*** toscalix has quit IRC14:45
*** toscalix has joined #baserock14:45
*** franred has joined #baserock14:46
*** pedroalvarez has quit IRC15:00
*** pedroalvarez has joined #baserock15:01
*** ChanServ sets mode: +v pedroalvarez15:01
*** locallycompact has quit IRC15:41
*** locallycompact has joined #baserock15:42
paulsherwoodhttp://linux.slashdot.org/story/15/12/16/040223/0-day-grub2-authentication-bypass-hits-linux15:56
gtristaneek16:24
gtristanfunny thing is that I think it was ismael himself who helped actually get grub working with the EFI boot in #grub just a week or two ago16:26
gtristanbut even if I'm effected, there's no way afaict from the article, that the 0 day exploit can get passed cryptfs without a password16:28
gtristanjust a power grub shell16:28
*** persia has quit IRC17:00
*** thrace has quit IRC17:00
*** ctbruce has quit IRC17:20
*** ssam2_ has joined #baserock17:31
*** ssam2 has quit IRC17:34
*** dabukalam has quit IRC17:54
*** dabukalam has joined #baserock17:54
*** locallycompact has quit IRC17:59
*** bashrc has quit IRC18:00
*** edcragg has quit IRC18:19
*** edcragg has joined #baserock18:21
*** edcragg has quit IRC18:32
*** persia has joined #baserock18:39
*** persia_ has joined #baserock18:39
*** toscalix has quit IRC18:44
*** ssam2_ has quit IRC19:07
* SotK goes to document his upgrade woes in storyboard19:59
SotK(actually, I'll try upgrading via a clean devel system first)20:08
mwilliams_ctdaft question (and at a daft time so I dont expect a reply!): will ybd work outside of baserock, say on ubuntu? Readme seems to suggest so20:32
SotKmwilliams_ct: I believe it should do20:33
mwilliams_ctSotK: ace, thanks :). Too sleepy to try and set it up on my vps right now, but that will save me worrying20:33
SotKneeds to be run as root though20:34
mwilliams_cthmm, that affects things. looking into running it in a concourse.ci task which arent privileged by default20:34
mwilliams_ctmwilliams_ct: note to self, you need to think about privileged tasks20:35
* SotK grumbles about morph downloading every artifact in the system, then downloading the system artifact too20:37
persiaI thought that hot fixed, for the special case where the system was cached.20:41
SotKgtristan: I feel the pain of using this old version of storyboard21:07
SotKupgrading from a clean devel system has the same trouble22:27
SotKI guess tomorrow's step is deploying a fresh VM then22:27

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