IRC logs for #baserock for Wednesday, 2015-05-27

ratmice__richard_maw: when you're talking about stripping in 'minimisable debuggable systems', and you say "Option 1 isn't suitable since some binaries shouldn't be stripped", It seems you refer to stripping symbol tables rather than just stripping debuginfo and leaving symbol tables intact?04:37
ratmice__I think the size benefits of stripping out symbol tables are much reduced compared to the latter (not that I have numbers)04:41
ratmice__so I think there is an option 1b which is plausible strip debug info, leave symbol tables intact, that can be done automatically through 'file/strip'04:42
ratmice__though its reduced but not minimal04:43
*** zoli__ has joined #baserock05:35
*** zoli__ has quit IRC05:39
*** zoli__ has joined #baserock06:37
straycat17:26  radiofree$ it's come up before, what's the fix for "Couldn't find morphology:06:40
straycat                  systems/system-that-is-clearly-there-and-commited-to-git.morph"06:41
straycatmost likely the morph file doesn't exist at the sha that's pointed to in definitions06:41
*** rdale has joined #baserock06:42
*** petefoth has quit IRC07:18
*** petefoth has joined #baserock07:18
*** mariaderidder has joined #baserock07:47
*** jonathanmaw has joined #baserock08:01
*** tiagogomes has joined #baserock08:05
*** mike has joined #baserock08:14
*** mike is now known as Guest8602108:14
*** gary_perkins has joined #baserock08:16
*** bashrc_ has quit IRC08:21
*** bashrc_ has joined #baserock08:21
*** edcragg has joined #baserock08:23
KinnisonAnother nail in the 'Docker in baserock' coffin:
* rjek sighs at people08:36
KinnisonNow if we want docker, we'll need 'go' in build-essential and we'll *require* platform-specific switches for chunk enablement because go can't do MIPS and I'll guess it doesn't distinguish between the various ARM options (though it perhaps does)08:36
* Kinnison can't bring himself to read the code to find out08:36
jonathanmaw"Go is the only language you'd need to know" fills me with anger08:40
KinnisonFor working on/with G08:40
jonathanmaweven if the statement is more reasonable in context08:40
pedroalvarez[22:02:04] <pedroalvarez> urgh, libexpat change broke everything08:45
pedroalvarez[22:26:35] -*- pedroalvarez sends to fix it08:45
pedroalvarez(I'm sure today there are more viewers than yesterday)08:46
richard_mawKinnison: AIUI they were leaving a bootstrap path so release n can build n+1, so it sounds like to bootstrap go you need gcc's go 1.4 something to build go 1.508:51
ratmice__Kinnison: at least there is still gccgo, rust for instance dropped their ocaml bootstrap compiler fairly early in development such that the language that the bootstrap compiler compiles no longer compiles the rust compiler, I believe that they hadn't ported to arm/mips at that point, although they do have the rather interesting/non-conventional ability to bootstrap new architectures through llvm IR (i'm not sure if its been done etc)08:51
KinnisonYeah, gccgo will become critical to getting this stuff bootstrapped :(08:53
*** ssam2 has joined #baserock08:55
*** ChanServ sets mode: +v ssam208:55
ratmice__pedroalvarez: hmm, isn't part of the problem with gettext depending on expat directly that gcc requires gettext and that'd bring excess stuff into the initial bootstrap?09:00
ratmice__at least that is the rationale for why it does so dynamically upstream i believe09:01
pedroalvarezratmice__: I didn't know that. It was jjardon who moved expat to the gettext dependencies09:10
pedroalvarezI'm only trying to fix what it's currently broken :)09:11
* Kinnison is very much saddened that when he asked if testing had been done he was assured it had been09:11
pedroalvarezI assumed that the behaviour was tested, but not the build. That is, building expat and gettext on the top of a system09:12
*** lachlanmackenzie has joined #baserock09:13
pedroalvarezshould we be worried about what ratmice__ said regarding gcc, gettext, ane expat?09:20
ratmice__I haven't really looked at expat's deps to see what impact it has on the size of build-essential, I hate this entire notion of optional dependencies that creeps in to everything though :(09:24
KinnisonGiven we didn't have expat in b-e before, I imagine gcc has something built in, no?09:24
ratmice__Kinnison: it doesn't use the xml stuff, I believe gnome/glade use it and do checks for gettext+xml in their configure scripts09:25
Kinnisonindeed, but my point is, there's no need to migrate gettext into b-e because it isn't there already and gcc already works09:26
pedroalvarezI hope this is not the kind of issue that we discover once we build a system with a system built with this change09:27
Kinnisongcc's compiler-behavour should not alter09:27
KinnisonI am, however, even more insistent that we need pre-merge testing from Mason09:28
ratmice__Kinnison: in the first statement you said 'expat', in the 2nd you said 'gettext', I'm suprised there is no gettext in b-e but haven't looked at this problem in depth really09:29
ratmice__Kinnison: yeah you're right gcc contains a gettext in intl/09:34
Kinnisonratmice__: sorry, I was referring in general to this shuffle-around :)09:35
ratmice__so yeah, I don't think there is any problem with making gettext depend on it, it shouldn't affect the gettext(s?) already in b-e.09:38
ratmice__binutils has one as well09:41
robtaylorsorry, that was accidental...09:42
* Kinnison just assumed you were shouting at him :)09:43
straycatpedroalvarez, franred, How come we're using python-keystoneclient 0.11.1 ?09:54
pedroalvarezstraycat: this was Juno's requirement:09:56
pedroalvarezstraycat: ooi, why?09:57
pedroalvarezstraycat: we have updated it for Kilo upgrade09:57
straycatjust noticed that we seemed to be using quite an old version09:59
radiofreeso what's the reason system-version-manager remove stops working?10:00
radiofreesomething systemd, something to do with that "machines" subvolume?10:00
ssam2oh, I forgot about that one :(10:01
pedroalvarezstraycat: we will upgrade to  1.3.110:01
ssam2that would be the best fix10:01
ssam2in the meantime you can workaround by deleting the /var/lib/machines subvolume manually I guess10:01
radiofree"Another problem is that systemd seems to create a Btrfs subvolume in /var/lib/machines, which cannot be removed with `rm`. This causes `system-version-manager remove` to fail with 'Operation not permitted' for system-versions that have this directory."10:01
radiofreethis seems pretty random10:01
radiofreechanging 'Storage' to volatile in the journald config at least gets around the lack of a log10:07
ssam2disabling systemd-machined may be the way forward for now, so the bug at least only hits people who use systemd-machined10:13
petefothSotK: I am intrigued by your change to guides/arm-howtos which says that partial deployment 'has a number of limitations.' Are those limitations listed / documented anywhere?10:13
SotKno, but they basically boil down to "installing random tarballs into your system at runtime makes reproducibility hard"10:17
SotKand also "system-integration-commands don't get run, and neither do any configuration extensions"10:18
ratmice__pedroalvarez: sorry if this is somewhat of a goose chase (anyhow summary), there is the possibility that we can include a gettext in b-e, (gcc, glibc, binutils * stages), and just cut that down to (1 * stages) gettext compiles, I'm guessing this would be 'stage1-gettext' etc, and non-impactful since I believe they would be strictly build deps?10:18
petefothSotK: Thanks.10:19
pedroalvarezratmice__: interesting10:20
pedroalvarezratmice__: but then, we will have to move the expat/gettext to b-e10:21
pedroalvarezI'll research about that when I have time :) thanks ratmice__10:22
ratmice__pedroalvarez: unsure (depends on if its strictly a build dep, or if it uses libintl as at runtime)10:22
straycatpedroalvarez, cool10:24
pedroalvarezlibexpat fix merged10:52
*** gary_perkins has quit IRC11:00
*** gary_perkins has joined #baserock11:01
* richard_maw goes to lunch while he rebuilds the world up from libexpat11:08
pedroalvarezMason x86_64 currently building glib11:11
*** sebh has joined #baserock11:15
pedroalvarezfranred: in that bug report, they say the fix is not to add a new dependency11:24
pedroalvarezhave you found anything useful?11:24
pedroalvarezare you going to just add the dependency?11:25
*** brlogger` has joined #baserock11:28
jjardonups, sorry for breaking the build. Thanks for the quick fix pedroalvarez!11:29
pedroalvarezThey also claims that this might have fixed the issue:
pedroalvarezjjardon:  np11:29
franredpedroalvarez, although that comment in on discussion - see last comment before the latest patch11:29
franredpedroalvarez, please, read the bug report thread... as you can see that patch does not fix the issue, and it is clear that it is still in master because we are facing as well11:30
*** ratmice___ has joined #baserock11:30
pedroalvarezwe are not using master though?11:31
franredpedroalvarez, fix which you refer to it is in 2015.1.0 and we are still facing the issue11:32
*** petefoth has quit IRC11:36
*** brlogger has quit IRC11:36
*** ratmice__ has quit IRC11:36
*** petefoth_ is now known as petefoth11:36
*** tiagogomes has quit IRC12:03
franredjjardon, pedroalvarez, ssam2, I've fixed the commit message and the URL
*** gary_perkins has quit IRC13:01
*** gary_perkins has joined #baserock13:08
*** tiagogomes has joined #baserock13:13
straycat"As of 12 April 2004, 44 sets of CPAN modules had names that differed only in capitalization."13:20
jmacs /o\13:22
straycatthis is actually a more serious problem on Windows apparently, due to the case insensitive file system, and perl relying on file system case sensitivity to disambiguate >.>
*** petefoth has quit IRC13:26
ratmice___very serious problem indeed, it would imply someone was using perl :D13:37
ratmice___there's also macs with ?sensitivity, depending on the filesystem13:42
richard_mawand then there's a gamut of software that uses : or , as separators for file paths13:43
DavePageAnd stuff that can't cope with perfectly normal characters like newlines in filenames13:44
richard_mawI know right!13:44
* ratmice___ imagines some language that just executes random instructions, so you try and find the right seed13:50
jmacsI know someone who was trying to do that for the ioccc13:53
richard_mawI know of someone who memorised a random seed for interviews, so he could give a rather daft answer to the traditional fizzbuzz.13:55
jmacsImplementations of rand() may vary13:55
flatmushthey rarely do though13:56
richard_mawjmacs: indeed, that was one of the problems with getting solitaire to behave right under wine13:56
* flatmush likes complementary multiply with carry13:57
ratmice___anyhow what got me there was the notion that perl (particularly visibile through its special variables) has embraced the notion of side-effects at the language level14:09
radiofreehmm.. shift+g in journalctl isn't working on an x86 image14:11
ssam2broken 'less' perhaps?14:12
radiofreethe same image works fine on arm (i think, testing now)14:13
radiofreeyes, fine on arm14:14
ssam2maybe it's your keyboard ;)14:14
ssam2i can't think what else...14:14
radiofreewell i'm ssh'ing in, so same keyboard14:15
radiofreehmm i rebased the x86 image against master14:15
ratmice___radiofree: I don't suppose it's caps-lock related? (I will note there is shady stuff with capslock/Xorg, where the light remains where X has it set, and terminal/X have independent interpretation of whether caps lock is on :)14:24
radiofreewell i'm ssh'ing in, same keyboard, same caps (if any)14:26
radiofreei'm not using exactly the same image on this jetson though, i'm deploying that now14:27
ratmice___radiofree: i guess i'm saying that doesn't matter (same keyboard), IF you're switching between virtual terminals, because each virtual terminal has independent notion of if the caps lock is on or not, and gives no indication for some unknown reason (it used to work i swear!)14:44
ratmice___thats probably not it, but to be thorough...14:44
*** zoli__ has quit IRC14:46
*** gary_perkins has quit IRC14:56
*** gary_perkins has joined #baserock14:59
*** Guest86021 has quit IRC15:05
jjardonDoes anyone know what the stars from the story list are for?!/story/list15:38
KinnisonNot a clue15:39
SotKTheir tooltip says "Subscribe", I'd guess that clicking them means you get emails when something is added to the story15:40
jjardonAh, maybe thats the intention when they actually implement that (sending emails) last time I checked was not yet done15:41
jjardonCan I have a quick review to lorry libgudev?
richard_mawjjardon: I'll give you a +1 if you can tell me what we currently have that uses it and currently gets it from systemd15:59
*** jonathanmaw has quit IRC16:02
jjardonrichard_maw: NetworkManager16:04
richard_mawjjardon: not currently used in any of our systems16:06
jjardonrichard_maw: yes, but its part of our current definitions and will break soon as gudev will be removed from systemd git tree16:11
*** zoli__ has joined #baserock16:12
richard_mawif a tree falls in the forest, and there's nobody there to hear it, does it make a sound16:13
richard_mawI suspect it may have already bit-rotted16:13
richard_mawbut it's reasonable enough justification16:13
jjardonrichard_maw: Its not, Im using it for my GNOME system as well :)16:14
jjardonalso, this is interesting because foundation doesn't have to depend on GLib anymore (need to check to be sure)16:16
jmacsCan anyone else "git clone" ? Times out for me.16:19
ssam2jmacs: timed out for me16:20
jjardonjmacs: same here16:22
jmacshttps might have worked, very slowly16:23
jjardonjmacs: I tried again and it worked now16:23
*** mariaderidder has quit IRC16:45
*** bashrc_ has quit IRC16:54
*** ssam2 has quit IRC16:55
*** edcragg has quit IRC18:05
*** gary_perkins has quit IRC18:18
*** lachlanmackenzie has quit IRC18:28
*** tiagogomes has quit IRC20:11
*** inara has quit IRC20:32
*** inara has joined #baserock20:34
*** zoli__ has quit IRC21:07
*** tiagogomes has joined #baserock22:03
*** zoli__ has joined #baserock22:16
*** zoli__ has quit IRC22:46

Generated by 2.14.0 by Marius Gedminas - find it at!