IRC logs for #baserock for Tuesday, 2015-07-14

*** zoli__ has quit IRC00:32
*** zoli__ has joined #baserock02:01
*** zoli__ has quit IRC02:06
*** zoli__ has joined #baserock06:11
*** persia_ has quit IRC06:29
*** persia has quit IRC06:29
*** persia has joined #baserock06:30
*** persia has quit IRC06:30
*** persia has joined #baserock06:30
*** persia_ has joined #baserock06:30
*** persia_ has joined #baserock06:30
*** rdale has joined #baserock07:01
*** mariaderidder has joined #baserock07:13
*** paulw has joined #baserock07:38
*** bashrc has joined #baserock08:04
*** mike has joined #baserock08:17
*** mike is now known as Guest1024308:17
perrylaugh, i've just realised why setting the class to an object and running it wasn't working. i was calling `class.function()` instead of `class().function()` *headdesk*08:20
paulsherwoodperryl: gotta love programming :)08:21
* perryl requests a tattoo gun to ink the sentence "REMEMBER YOUR PARENTHESES" on her face :)08:21
KinnisonI have a sharpie08:21
paulsherwoodperryl: -1 :)08:22
persiaperryl: You may find that programming in lisp for a bit helps build the muscle memory for pressing ( and ).08:23
bashrcperryl: does this mean you don't have rainbow parentheses?08:23
perryli don't think rainbow parentheses would help if i forget an opening brace :(08:24
perryli wonder if this is the python equivalent of forgetting a semicolon...08:24
*** ctgriffiths has quit IRC08:49
ratmice___peculiar, it seems its much easier to write backwards on your forehead standing in front of a mirror than it is to write backwards onto a piece of paper08:49
*** ctgriffiths has joined #baserock08:50
Kinnisonratmice___: what's fun, is that I believe wholeheartedly you just went to try08:50
*** ssam2 has joined #baserock08:55
*** ChanServ sets mode: +v ssam208:55
ratmice___I figured, being covered in paint and all it wouldn't hurt, painting the house :(08:57
ratmice___ahh, faulty test... you just have to write on the side of the paper facing away from you09:04
*** lachlanmackenzie has joined #baserock09:10
SotKperryl: nah, that's accidentally mixing tabs and spaces09:23
*** CTtpollard has quit IRC09:26
*** CTtpollard has joined #baserock09:29
*** rdale has quit IRC10:10
*** rdale has joined #baserock10:11
paulsherwoodperryl: not sure if it helps, but i did a bit more work on the ybd poc server... still didn't get it all working though. it's pushed to master10:19
paulsherwoodthis was my attempt to change to post rather than put, and have directories for atomicity as discussed with kinnison10:19
perrylpaulsherwood: thanks, i'll take a look!10:19
paulsherwoodi think it would be better to base off m-c-s, since that probably has more real-world thinking embodied in it10:20
paulsherwoodperryl: you're welcome. i'm doing my best to save you from the tattoo needle :)10:21
perrylhahah at the moment my issue has moved from parentheses to "global name is not defined" :(10:24
paulsherwoodcan anyone here comment on how our gerrit users/roles are setup, please?10:36
paulsherwooddo we differentiate between contributors, developers, integrators, progject owners, admin (as described for example at seems a bit complex10:37
* paulsherwood discovers
ssam2paulsherwood: the relevant bits of infrastructure.git are:10:43
ssam2that page on Gerrit you linked to is probably a better place to look though10:44
ssam2there can be hidden groups that only show up to admins, but I intended them all to be visible and I think they are10:45
ssam2they are10:45
paulsherwoodssam2: thanks!10:54
paulsherwoodperryl: do you have a working instance of your server anywhere that i could try out?10:54
paulsherwood(i mean
perrylpaulsherwood: i've got nothing set up with architecture; only the test runs10:55
* paulsherwood doesn't quite understand that10:55
paulsherwoodhave you hammered it? :)10:56
paulsherwoodand if so, what with? ybd? morph? both?10:56
perryli'm not sure what you mean by 'hammered'10:56
paulsherwoodtested thoroughly10:56
Kinnisonhammering typically implies load-testing10:58
perrylahh; i''ve not tried that yet10:58
perryli asked about setting up a cache server and was told to run `python ./cache-server --no-fcgi-server`10:59
paulsherwoodperryl: ok, but it may be worth loading it on a physical server somewhere and causing multiple builds to access it. not sure if you can easily do that on your devel machine?11:01
paulsherwoodin any case i think this is an important change, and ideally would be of interest for baserock as a whole, not just ybd11:01
paulsherwoodso maybe it would make sense to discuss it on the list?11:02
paulsherwoodi'm happy to go this way with it becoming part of ybd, but maybe it would be better as a separate cache server project in its own right?11:02
perrylpaulsherwood: indeed; if it can be modified to work with ybd and morph it should be relatively simple to extend to other uses11:03
perryli'll send an email to the list; in the meantime are there any instructions for setting up a cache server anywhere?11:03
paulsherwoodgood question :)11:03
KinnisonGiven cache servers were always part of trove, I doubt it11:04
paulsherwoodperryl: probably best to look in the trove definitions11:04
paulsherwoodperryl: does it cache repos as well? it imports RepoCache11:05
SotKtrove-setup.git contains the systemd units and lighttpd config for morph-cache-server11:05
paulsherwoodthanks SotK :)11:06
perrylSotK: thanks, i'll take a look11:06
perrylpaulsherwood: i can remove repo caching if it's not required11:06
SotK+1 from me for a separate cache server project fwiw11:07
paulsherwoodi don't think this should cache repos11:08
perrylno worries! i'll remove repo caching in that case11:09
paulsherwoodgreat, thanks! :)11:10
paulsherwoodperryl: maybe ask on the list if folks would prefer a separated cache-server project to be a baserock/baserock/ project on g.b.o, or not?11:12
perrylpaulsherwood: sure thing, i'll get on that after lunch :)11:13
paulsherwoodssam2: i notice elsewhere you've been looking at ruamel. does it really allow roundtripping of yaml? i remember we got into a knot in morph with that11:14
ssam2paulsherwood: it's not at all perfect, but I hope it will be good enough11:19
ssam2if we want to automatically migrate .morph files, I think it's the best option11:20
ssam2the alternative is... targetted 'sed' scripts, I guess? but that has quite severe limitations11:20
paulsherwood+1 for staying in python if we can11:20
ssam2here's an example of the roundtrip for all of definitions:
paulsherwoodsed is very powerful, but would be a complete mess11:21
ssam2this is with a patch I'm working on to preserve blank lines (which has added some spurious blank lines, too)11:21
ssam2most of the changes seem to be because ruamel.yaml doesn't preserve indent. But i'm hoping everyone will be OK with having a consistent level of indent throughout definitions11:22
ssam2I went to fix a bug in ruamel.yaml that was producing spurious blank lines, but discovered the spurious blank lines are actually in the input .morph files11:37
*** paulw has quit IRC11:37
*** rdale has quit IRC11:59
*** rdale has joined #baserock12:03
*** cosm has joined #baserock12:08
*** paulw has joined #baserock12:11
17SAC2BTE-- ./configure --with-nss --prefix="$PREFIX" --sysconfdir=/etc --without-fuse --without-libatomic-ops --without-libxfs12:25
17SAC2BTE+- ./configure --with-nss --prefix="$PREFIX" --sysconfdir=/etc --without-fuse --without-libatomic-ops12:25
17SAC2BTE+  --without-libxfs12:25
*** 17SAC2BTE is now known as tiagogoems12:26
*** tiagogoems has quit IRC12:27
*** tiagogomes has joined #baserock12:27
tiagogomesthat doesn't look correct12:28
*** fay_ has quit IRC12:30
*** fay_ has joined #baserock12:45
*** cosm has quit IRC12:52
ssam2in the example migrated definitions I linked to? you might be right12:56
paulsherwoodit wraps lots of things, i notice12:57
ssam2according to YAML spec it's correct:
ssam2lines can be broken like that and will be folded back together when read12:58
paulsherwoodok :)12:58
ssam2paulsherwood: yes, it line wraps to a specific width12:58
ssam2but if you set a longer width, it *un wraps* things12:58
ssam2which is annoying12:58
ssam2but at the moment definitions are inconstently line wrapped, which is also annoying, at least to me :)12:59
*** paulw has quit IRC13:20
*** paulw has joined #baserock13:23
straycathello again13:30
straycatssam2, i have some time to kill waiting for a build and figured it'd be useful to tie up the docbook-xml loose end, can we merge and update lorry on now?13:32
straycatpaulsherwood, hi!13:32
straycatfwiw, i've been running that change on my trove without incident13:33
paulsherwoodstraycat: looks good to me, and given you've tested it i've merged13:34
paulsherwoodstraycat: out of interest, how much time/effort did it take you to spin up your own trove?13:35
*** paulw has quit IRC13:35
paulsherwoodand are you running it on a cloud somewhere, or?13:35
*** paulw has joined #baserock13:35
* paulsherwood has been meaning to re-establish his own trove for a long time13:36
straycatit's just a libvirt vm, about 10 minutes or so i think13:36
paulsherwoodplus the sync time :)13:36
ssam2fine to update lorry, but please make a patch for definitions.git to actually update lorry in definitions too13:37
straycatoh i left that overnight :)13:37
ssam2otherwise it'll be de-updated again next time g.b.o is updated13:37
paulsherwoodgood point13:37
straycatssam2, okay will do, do you want to do the update or shall i? on my trove i was pretty lazy and just reran lorry's, for gbo i think we could just copy the new tar-importer in place13:40
straycatoh, except no we want zip support as well13:40
ssam2please do it if you're happy to13:40
ssam2copying the bits into place is fine by me for something as simple as lorry13:41
straycatyes, lorry, and lorry.tar-importer should be enough13:43
ssam2I guess ' install' should work13:44
straycatyou'd think so but actually i just noticed that i failed to update the when i added the zip-importer, so that will need to be done first13:46
ssam2good catch13:51
straycat fixes that, on second thoughts there are some gitano fixes in flight to fix a bug that's somehow not been noticed, perhaps we should wait for that to land then do a proper upgrade, i'd hate to break gbo just because i'm being impatient, we could always create baserock/baserock/docbook-xml to get perl 5.22?14:10
ssam2upgrades are cheap14:11
ssam2i'll try doing the lorry upgrade if you want14:12
ssam2would you mind doing a patch to update lorry in definitions.git though?14:16
straycatahh you already merged it, sure :)14:17
straycatlorry update:
*** ssam2 has quit IRC14:34
*** ssam2 has joined #baserock14:35
*** ChanServ sets mode: +v ssam214:35
*** paulw has quit IRC15:34
*** paulw has joined #baserock15:35
straycatdoes anyone know we go to some effort to put all perl stuff into /usr/lib/perl rather than default site and vendor dirs, we override in 2 places to do this, firstly in the perl chunk morph and secondly in morphlib/, i have to setup a more typical installation, but would also probably want to modify
straycat*know why15:39
*** mariaderidder has quit IRC15:43
ssam2straycat: i don't know about perl at all, but what do you think to this?
ssam2I also created!/story/5315:45
ssam2i've updated lorry on g.b.o, by the way15:49
ssam2have also rotated the database of Lorry log information, as it was 2.9GB and lorry-controller was continually at 200% CPU15:51
persiastraycat: Might the perl redirection be a habit we inherited from Debian under the impression it was best practice?16:03
* persia doesn't know the history, but does know that some of the early work in reference systems was modeled after Debian, and that some of the early engineers were comfortable in a Debian environment.16:04
*** Guest10243 has quit IRC16:04
straycatNot sure, this doesn't seem to come from Debian, I have modules in /usr/lib/x86_64-linux-gnu/perl5/5.20/ there16:07
KinnisonDebian's all multiarch remember16:07
*** paulw has quit IRC16:08
straycatsure without multiarch it'd be /usr/lib/perl5/5.20 which seems to be the default layout16:08
straycati'll continue with the default then and modify to suit, see how that goes16:10
* persia finds stuff under both /usr/bin/perl5/ and /usr/bin/perl/, but may not be running a recent release16:13
*** Guest10243 has joined #baserock16:18
paulsherwoodstraycat, ssam2 - am i missing something from the ml discussion about perryl's proposed cache server? i understood that m-c-s is no longer a separate thing (integrated into morph) but your comments seem to suggest that's not true?16:44
ssam2paulsherwood: the two source files which make it up were moved into morph.git16:46
ssam2nothing more was done, so to 'de-integrate' it we simply need to put those files in some other repo16:47
paulsherwoodright. so this is about establishing a generic cache server, as opposed to needing morph?16:47
paulsherwoodi thought that's what perryl has done...16:47
ssam2I haven't seen what perryl has done, I got the impression she was waiting to find out where to push it before pushing it16:47
ssam2I don't think the email she sent linked to any code16:47
paulsherwoodoops... i missed that, because i'd seen the code16:48
paulsherwoodperryl: need to show the code, or we get all confused16:49
* paulsherwood apologises to ssam2 and straycat for crossed wires16:49
*** Guest10243 has quit IRC16:57
ssam2apologies for sending mail to, my mail is doing really silly autocompletion for some reason17:05
*** bashrc has quit IRC17:08
*** ssam2 has quit IRC17:08
*** Guest10243 has joined #baserock17:10
*** lachlanmackenzie has quit IRC17:33
*** paulw has joined #baserock17:36
*** paulw has quit IRC17:45
*** mdunford has quit IRC17:48
*** Guest10243 has quit IRC17:58
*** brlogger` has joined #baserock18:22
*** pdar_ has joined #baserock18:25
*** persia_ has quit IRC18:30
*** pdar has quit IRC18:30
*** brlogger has quit IRC18:30
*** zoli__ has quit IRC18:31
*** _longines has joined #baserock19:26
*** _longines has joined #baserock19:28
*** zoli__ has joined #baserock21:06
*** zoli___ has joined #baserock21:13
*** zoli__ has quit IRC21:16
*** zoli___ has quit IRC23:00

Generated by 2.15.3 by Marius Gedminas - find it at!