IRC logs for #baserock for Monday, 2014-10-27

persiaThanks for showing me a usage for which the "there are two classes of systems" model doesn't quite work :)00:00
jjardonpersia: simplicity of definitions is one of the good things about baserock, dont think nobody wants to change that (only to make it event more simple like paulsherwood patches do)00:00
* persia needs to think about this for some time00:00
persiaApologies if my comments suggested I believed you were against simplicity :)00:01
jjardonpersia: another example is: "make mesa depend on x-common (X11) only is Im actually building a system with X11"00:01
persiaThat's an easy case of "branch definitions" to me.00:01
jjardon(check strata/mesa-common/mesa.morph)00:01
persiaIt's "I want to build foo for bar, baz, and quux, as a single cluster" that doesn't fit the branch model.00:02
persiaThe specific example being supporting both linux and BSD in jhbuild.00:02
persiaAnd the reason you need it as a single cluster is because otherwise it's far too painful to deploy for validation as part of a CI suite.00:03
jjardonyeah, branch the whole definitions only to change a single line in one morph file I think is overkill. I think add the ability to morph to understand conditionals is cleaner (from the maintainance and reusability of definitions)00:07
persiaThe main source of my disagreement is that "branch defintions and change one line" is the starting point for new users we describe in the "getting started" section, so this should be easy to do, and easy to maintain.00:07
persiajjardon: I've sent some arguments to the list.  In the event that I haven't strongly enough argued for the use of conditionals to support what I have termed "case B", please chime in.00:34
persia(as I don't think there's much point to you and I arguing about what I termed "case A" until we have a model for "case B")00:35
*** zoli_ [] has joined #baserock04:45
*** zoli_ [] has quit [Changing host]04:45
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock04:45
*** abdul [~abdul@] has joined #baserock05:29
abdul is now known as Guest3452005:30
*** madhu [~madhu@] has joined #baserock06:19
*** hari_ [~hari@] has joined #baserock06:20
*** hari_ [~hari@] has quit [Client Quit]06:21
*** visteon [~visteon@] has joined #baserock07:35
*** Guest34520 [~abdul@] has quit [Quit: Leaving]08:22
*** abdul [~abdul@] has joined #baserock08:23
abdul is now known as Guest2599408:23
*** Guest25994 [~abdul@] has quit [Client Quit]08:24
*** abdul_ [~abdul@] has joined #baserock08:25
*** hari_ [~hari@] has joined #baserock08:28
visteonHi Paul08:36
*** aananth [~caananth@] has joined #baserock08:40
visteonHi Paul, Gnana here08:40
aananthHi, This is Aananth C N08:40
*** VVIBIN [~VVIBIN@] has joined #baserock08:41
VVIBINhi ,my proxy issue resolved08:41
*** tiagogomes [] has joined #baserock08:44
*** foobar_ [560e67b5@gateway/web/freenode/ip.] has joined #baserock08:48
foobar_hi there baserockers08:49
*** karthi [ca004dc6@gateway/web/freenode/ip.] has joined #baserock08:50
*** violeta_ [] has joined #baserock08:50
*** bala [~bala@] has joined #baserock08:56
pedroalvarezfoobar_: hello!09:00
*** abdul__ [uid49809@gateway/web/] has joined #baserock09:01
*** ssam2 [] has joined #baserock09:01
Mode #baserock +v ssam2 by ChanServ09:01
*** jonathanmaw [] has joined #baserock09:14
radiofreejjardon: do the gtk upgrade patches work with mesa compiled --disable-glx?09:19
*** jjardon84 [] has joined #baserock09:23
De|ta_ is now known as De|ta09:24
*** sambishop [] has joined #baserock09:27
jjardonradiofree: yes, opengl support in GTK+ is implemented using libepoxy, a library that abstracts you of the specific opengl implementation present in your system:
jjardonpersia: thanks for the review! Should I assume is it ok to commit after those fixes?09:36
*** foobar_ [560e67b5@gateway/web/freenode/ip.] has quit [Ping timeout: 246 seconds]09:47
aananthAre you there?09:48
persiajjardon: The other bits look mostly sane to me, but I don't have +2, so you'll need a second opinion09:49
persiaAlso, I'm not sure how to fix the gdk-pixbuf-query-loaders issue, or even if I'm understanding it properly.09:50
aananthPaul: I am going through
aananthMy question is, do we need "SATA disk or SSD with cables". 10:11
KinnisonWe recommend you use an SSD to reduce the IO latency when compiling10:12
KinnisonBut any SATA disk will do in a pinch10:12
aananthHi Kinnison. Thanks.10:12
*** Krin [] has joined #baserock10:13
KinnisonWhat's important is that there's plenty of room on the drive for you to have builds, source trees, etc.10:13
aananthI have a WD My Passport with USB interface, will that be OK?10:13
KinnisonAgain, we recommend 128G or more, but 64G will do in a pinch10:13
KinnisonUSB is not a good idea10:13
Kinnisonit's higher latency, higher CPU load, and less reliable10:13
KinnisonIt will do in a pinch, but may cause issues10:14
KinnisonThe Jetson support is still being merged into upstream, so sticking to our suggestions in the wiki will help you to not encounter too many issues we've not seen in the past10:15
aananthLet me then buy a SATA cable for regular development. But for doing initial trial, I hope I can do without SATA cable.10:15
rdalei am looking at the qt5-tools lorries and some have a "refspecs": option, but most don't. what does that option do?10:15
Kinnisonrefspecs tell lorry what it should try and push into the trove git server10:16
Kinnisontypically they're used when upstream often rebases non-feature branches which means that we need lorry to force-push stuff10:16
rdaleah i see, thanks10:16
* persia wonders if the current recommended Jetson kernel from definitions master has decent USB 3.0 support10:18
KinnisonI'm not sure10:19
KinnisonWhen 3.18 hits properly, we might have reasonable chance of good support10:19
*** jjardon84 [] has quit [Remote host closed the connection]10:19
radiofreepersia: the 3.10 kernel might have usb 3.010:32
radiofreenot tried the 3.17ish kernel we're using10:32
radiofreewhat would be the best way to test that? plug as usb 3.0 hd into it and test how fast it is?10:33
*** ssam2 [] has quit [Ping timeout: 244 seconds]10:33
KinnisonPlug in a USB3 device and see if the kernel reports it as USB3 :-)10:34
pedroalvarezhm.. the only difference between a build system and the distbuild system is that the latter has openstack-clients. I wonder if we can add them to the build-systems and remove the distbuild system.10:34
* richard_maw thought he already saw patches on the ML doing that10:35
*** ssam2 [] has joined #baserock10:35
Mode #baserock +v ssam2 by ChanServ10:35
pedroalvarezwhen I say "the distbuild system" I say "the" because there is only one: distbuild-system-x86_64-openstack.morph10:36
SotKoops, that is what I added in my Mason patch series10:36
* pedroalvarez checks the email 10:36
radiofree[82376.569660] usb 2-1: new high-speed USB device number 46 using tegra-ehci10:39
radiofreeso no for this kernel then10:39
rjekHmm, ehci is explicitly USB 2, isn't it?10:40
rjekAren't USB3 controllers xhci?10:40
rjek(Perhaps a different kernel driver needs enabling?)10:41
*** hari_ [~hari@] has quit [Quit: Ex-Chat]10:47
ZaraI was thinking, would the occasional picture on the wiki be a good idea? (eg: an annotated tree diagram of chunks/strata/systems in baserock, to show where things go and help explain the concepts) It would probably take a while to make something like this both easily-editable and pretty, so I'd rather not start if it's not wanted. 10:52
ssam2I'd be in favour of it, as long as it was easy to keep up to date10:53
jmacsYes, I like pictures10:54
jmacs might be worth a look - ceph use this for their wiki10:54
* richard_maw would have found a diagram showing the relations between chunks, strata and systems; and the relationship between morphologies, sources and artifacts useful to help explain how things work to people10:54
richard_mawjmacs: it looks nicer to work with than plantUML10:56
persiaA diagram of the concepts would be fairly useful.  A diagram of a particular example from a particular branch of definitions,git I would find less useful (and somewhat limiting to newcomers).10:56
ssam2rdale: I don't see your branch 'baserock/rdale/add-extra-qt5-modules' in baserock/local-config/lorries.git10:58
*** franred [] has joined #baserock10:58
ssam2rdale: is it definitely pushed ?10:58
rdaleno, i don't have git access with my current public key - i just mailed the list about that11:01
ssam2ah, OK11:03
ssam2I'll merge it using 'git am, then, no problem11:03
ssam2rdale: so you used to have push access, but your key has changed, I guess11:10
ssam2richard_maw: has the necessary credentials to add your new public key to your existing account on, I think11:10
rdalessam2: yes the key was for my old laptop which was stolen11:11
richard_mawssam2: you also do now11:13
ssam2cool :)11:13
ssam2rdale: if you send me your new public key in an email I'll add it to your account11:14
rdaleok thanks11:14
ssam2you can then use the sshkey command to delete the stolen one, too11:15
Zarapersia: I think I'd find it easier to understand the concepts if they were presented via a concrete example (even if it were not a real one). However, I can only speak for myself, so I'd appreciate more opinions on this (from everyone).11:16
* paulsherwood notices that build-system can not be deployed without specifying various parameters, and wishes it would default rather than borking.11:18
ssam2paulsherwood: which parameters ?11:19
pedroalvarezI guess this is because mason.configure11:19
ssam2ah. good catch11:19
ssam2I'll look at fixing that11:19
paulsherwoodDISTBUILD_WORKER was the first one11:20
ssam2right. distbuild.configure needs fixing too, I geuss11:20
pedroalvarezindeed :(11:21
ssam2my fault, obviously I didn't test deploying a build-system before sending the patch for review11:21
pedroalvarezgit blame will blame me11:21
straycatThe only problem with defaulting is that it means that if I accidentally misconfigure my distbuild cluster I won't find out until I discover I've deployed a broken system.11:22
*** rdale [] has quit [Ping timeout: 245 seconds]11:22
paulsherwoodssam2: reproducible via scripts/ systems/build-system-x86_64.morph clusters/upgrade-devel.morph11:22
Zarastraycat: couldn't the default include a warning message?11:23
ssam2for distbuild.configure I was thinking of making it do nothing if none of DISTBUILD_WORKER, DISTBUILD_CONTROLLER and DISTBUILD_GENERIC are unset11:23
paulsherwooddo nothing? what if i just want a system for building?11:23
*** rdale [] has joined #baserock11:23
ssam2I mean distbuild.configure will do nothing, leaving you a system for doing local builds11:23
paulsherwoodssam2: i assume you meant s/are unset/is set/ ?11:24
ssam2yes :)11:24
straycatZara, Only if we attach some meaning to the system that's in the cluster, and use that to decide that user has probably forgotten to set their DISTBUILD_ options11:25
straycatI'm less sure about this, being clumsy as I am I remember deploying distbuild systems but forgetting to set the options.11:27
straycatBut *shrug* I don't have a better solution in mind11:27
paulsherwoodssam2: please can i ask what your in-flight 'resolve the ref' patch series aims at, in terms a user can understand?11:27
ssam2straycat: I think there are about 10000 ways to mess up deployment of a distbuild network11:28
ssam2anyone doing will need to be following the instructions closely11:28
ssam2*doing it11:28
straycatIf I make a mistake morph should do what it can to try and warn me of that11:28
paulsherwoodi thought it was just a 'cluster'11:28
ssam2straycat: I agree in theory, but we're very away from that for deployments right now11:29
ssam2paulsherwood: most importantly, it stops morph updating definitions.git if it doesn't need to when you run `morph build`11:29
ssam2paulsherwood: it's also a step towards being able to disable creation of temporary build branches11:29
paulsherwoodok. so from a user perspective, zero functionality change?11:30
ssam2should be11:30
paulsherwoodok lovely11:30
paulsherwoodas anyone been bothered to watch the new videos on w.b.o?11:30
ssam2I watched the setting up a VM one11:31
* paulsherwood wonders if anyone wants to comment on the 'no morph edit' workflow in the second one11:31
paulsherwoodssam2: i take your silence on the setting up a VM one as tacit approval :)11:31
* petefoth goes to watch a video11:31
* paulsherwood goes to announce the videos on the list11:32
ssam2paulsherwood: indeed, it looks useful11:32
ssam2it could possibly do with a more cheerful voice over :)11:32
ssam2but then setting up a VirtualBox VM is not much fun :)11:32
ssam2rdale: your new key should work now, you can do 'ssh whoami' to check11:33
ssam2and 'ssh help sshkey' for instructions on the sshkey command11:34
rdaleok thanks11:38
paulsherwoodssam2: i have a cold :)11:38
Kinnisonpersia: Just to let you know, I see your mail about conditionals and branching, but it requires a significantly larger amount of spare brain than I have right now.  When I've gotten the rest of the mails reviewed, and my current thoughts down on-keyboard, I'll have a re-read of your mail and try and form a cogent response :-)11:39
* Kinnison is really glad someone is bringing up the ideas11:39
petefothAre the new videos linked from ?11:42
petefothIf so. whicuh are the new ones? The 'BAserock VM Creation Tutorial" claims to have been made a yuear ago and uses BASerock 711:43
petefothOK found the new ones - I'll edit the 'Videos' page to match11:51
paulsherwoodpetefoth: probably worth culling some of the old ones11:54
paulsherwoodrichard_maw: quick python question re
paulsherwoodis the self.run_program call the same for both legs of the if?11:59
aananthThe instruction to setup VirtualBox VM is different between and
paulsherwoodaananth: that may be true. how exactly?12:02
aananthStep: "VBoxManage storageattach Baserock --storagectl "SATA Controller" --port 0 --device 0 --type hdd --medium baserock-current-devel-system-x86_64-generic.vdi"12:02
aananththrows error: "Could not find file for the medium '/home/caananth/baserock/baserock-current-devel-system-x86_64-generic.vdi' (VERR_FILE_NOT_FOUND)"12:02
paulsherwoodis there a vdi file in your /home/caananth/baserock directory?12:03
aananthThe same step is "VBoxManage storageattach Baserock --storagectl "SATA Controller" --port 0 --device 0 --type hdd --medium baserock.vdi" in the other page12:03
aananthLet me delete all and restart12:03
paulsherwoodshould not be necessary12:04
richard_mawpaulsherwood: no, it's a `hg clone` in the else, and a `hg pull` in the `if os.path.exists(hgdir):`12:04
aananthSorry, can I retry the last step alone?12:04
paulsherwoodthe vdi file is created by "VBoxManage convertdd baserock-current-devel-system-x86_64-generic.img foo"12:04
paulsherwoodaananth: so if you can see a vdi file in your directory, that's the name to use in the 'VBoxManage storageattach Baserock' command12:07
paulsherwoodrichard_maw: aha! thanks12:08
aananthOk, I understood. I retried with proper arguments this time, it went through.12:08
paulsherwoodaananth: excellent :)12:08
* paulsherwood goes to check if either page is incorrect12:08
aananthGot the following error when I tried to start the VM12:11
aananthVBoxManage: error: Nonexistent host networking interface, name '' (VERR_INTERNAL_ERROR)12:11
aananthVBoxManage: error: Details: code NS_ERROR_FAILURE (0x80004005), component Console, interface IConsole12:11
ssam2aananth: oh, nasty12:13
ssam2I guess it's something wrong with the line 'VBoxManage modifyvm Baserock --ioapic on --memory 2048 --nic1 bridged --nic2 hostonly', that's where we define the network interfaces12:13
aananthI opened the graphical view, it says "No hostonly network adapter is currently selected". Investigating...12:14
jmacsYou need a host-only network set up in virtualbox before you do that12:14
jmacsIt's something you only need to do once after you install VirtualBox, so I suspect whoever wrote the instructions didn't notice12:15
aananthjmacs: how to do that?12:15
jmacsYou can do it using the GUI quite easily, not sure how you do it from the command line12:15
aananthIn graphical view, I changed it12:15
ssam2what's the benefit of having a host-only network adapter in addition to a bridged one ?12:15
ssam2it seems easy to do, anyway ! Settings -> Network -> Host-only Networks and then click the green + sign12:16
ssam2as aananth has already spotted :)12:16
radiofreei've posted a reply to gavin regarding upgrading a jetson within a jetson, could someone with a bit more experience with the cycle script flesh out the last bit please?12:20
ssam2radiofree: i'll hopefully get the chance to try out your instructions, and at that time if I feel I can usefully add info to your reply then I will12:25
ssam2it's really useful that you're supporting Gavin on this anyway, thanks!12:26
* paulsherwood thinks he only *re*wrote the vbox instructions, but maybe not12:30
petefothI was the one who added the 'hostonly' - I did it because the previoius version set to set the port as 'nat' which didn't work for me, 'host-only' did12:30
paulsherwoodpetefoth: you're as bad as me, then :)12:31
petefothI've had VirtalBox installed and in use for ages and I have no recollection of having to set up the 'host-only' interface12:31
petefothpaulsherwood: :)12:31
jmacspetefoth: Which platform are you on?12:32
petefothI'm going to ull videos, and add some mor information abpout those that remain12:32
petefothjmacs: Mac OS X12:32
* pedroalvarez spots "describe": "baserock-14.26-365-g84398c0" in the definitions-version field of the deployment.meta file12:32
jmacsI have a suspicion that Windows and OSX Virtualbox set up a host-only adapter for you, and the Linux versions don't12:33
jmacsI can't uninstall VirtualBox to check right now though12:33
paulsherwoodvirtualbox can be a twisty maze of passages all alike12:33
ssam2I didn't have any host-only adapters in my VirtualBox12:33
ssam2which came from rpmfusion for Fedora 20, I think12:33
petefothjmacs: useful - I'll add a note that usesrs should check wheter or not they have a host only if set up (actually I may add an 'if you see this error...'12:34
VVIBINI am running VBox in Windows 7. I did install baserock VM and its running.However network is not up yet. I just now created a Baserock VM. But trouble in getting the ip assigned to the baserock VM. I have configured the Primary adapter in bridged with host and secondary adpter as Host only Adapter. The video doesnÂ’t talks much more to do with network settings....any clue?.12:35
VVIBIN Shows up only lo and "sit0 interface without any ip assigned.12:35
paulsherwoodssam2: ^^ ? 12:35
ssam2VVIBIN: I'll investigate in my own VirtualBox instance and get back to you on that12:35
VVIBINthnx and just looked into journctl log on boot and it show "baserock systemd[1] :failed to start ifup for sit0"12:37
ssam2VVIBIN: that message is harmless12:37
paulsherwoodVVIBIN: that happens to me too12:38
petefoth sugests that you might need to enable the the DHCP server useful that is built into VirtualBox12:38
ssam2could you go into the settings window for your VM12:38
ssam2onto the network tab12:38
ssam2and tell me what adapter it is bridged *to* ?12:38
paulsherwoodVVIBIN: can you check via vbox gui, that you have two network interfaces?12:38
* paulsherwood shuts up12:38
ssam2VVIBIN: it could be that it's bridged to your wireless interface but you're connected by wired only12:38
ssam2or, it could be bridged to your wired network but you're using wireless only12:39
VVIBINssam2: ok ,this may be ,i am using the same settings I kept for my ubuntu VM. I will turnoff the wireless and try with wired only,is there a way to overcome this12:41
ssam2VVIBIN: Virtualbox can only bridge one of your network adapters12:42
ssam2using bridged networking, you can choose which one it uses, but you can't automatically use whichever one is connected12:42
ssam2is that a problem? if you have a good wired connection I suggest bridging to your wired network interface12:43
VVIBINyes, i will make a try with wired connection.12:43
*** jonathanmaw [] has quit [Ping timeout: 244 seconds]13:02
*** karthi [ca004dc6@gateway/web/freenode/ip.] has quit [Ping timeout: 246 seconds]13:04
*** jonathanmaw [] has joined #baserock13:14
*** franred [] has quit [Quit: Leaving]13:19
aananthFollowing page helped me to resolve my VirtualBox booting issues caused by incorrect network:
petefothaananth: thanks - I'll add that to the rlevant wiki page13:23
*** aananth [~caananth@] has quit [Ping timeout: 258 seconds]13:37
*** abdul_ [~abdul@] has quit [Quit: Leaving]13:44
ssam2VVIBIN: are you having better luck with a wired connection ?13:44
paulsherwoodssam2: may be end of day for him already13:48
*** aananth [~caananth@] has joined #baserock13:49
paulsherwoodso with the arrival of build-system*, should i start using that by default instead of devel? what would i miss out on?13:50
ssam2right now you'd miss out on having Ruby and Node.js13:50
ssam2in future you'd miss out on the whole import tool thingy13:51
ssam2it lacks nothing compared to the old devel system, though13:51
paulsherwoodok. what was the reasoning for putting ruby and node into devel? (i'm not unhappy with this, just curious)13:52
radiofreei'd like to merge the baserock/james/jetson-3.17-rc5-cpufreq with baserock/jetson/3.17.0-rc5 branch13:52
paulsherwoodradiofree: +113:53
radiofreeis it the same review procedure for this? (i.e i submit 20odd patches to the mailing list)13:53
paulsherwoodradiofree: why not just post the branch and invite folks to test?13:53
ssam2paulsherwood: the import tool needs to have Ruby available in order to import RubyGems, node available to import NPM packages, etc13:53
ssam2I wanted some system where the import tool could work 'out of the box', and devel seemed like the best place13:53
ssam2I believe build-system was your idea originally :)13:54
radiofreepaulsherwood: ok, rdale is testing it, so am I, also gavin will be soon13:54
paulsherwoodssam2: aha. all we need next is docker by default :)13:54
ssam2paulsherwood: i'd be in favour of that :)13:54
paulsherwoodssam2: i'm full of daft ideas :)13:54
paulsherwoodssam2: i'll propose it, you can +1 :)13:54
radiofreethe benefit here is that we can i) use the same kernel/u-boot for devel images now ii) massively simplify that horrendous kernel args13:54
ssam2in that case +10013:54
paulsherwoodradiofree: looks like you already have enough to merge :)13:55
radiofreebtw pedroalvarez was looking at my drm patches on friday, are they good to merge yet?13:55
* paulsherwood would still like to try it out, though13:55
radiofreeagain, rdale is testing them... so have I13:55
radiofreepaulsherwood: follow the instructions I sent to the list13:55
rdalehow did you build the u-boot.bin image?13:56
paulsherwoodradiofree: ok will do, next time i am colocated with a jetson13:56
*** VVIBIN [~VVIBIN@] has quit [Quit: Leaving]13:56
ssam2morph-cache-server on is stuck in recvfrom() again13:58
ssam2hence builds are showing up as failing13:59
ssam2i'm going to restart the cache-server process unless anyone volunteers to investigate in the next minute or two13:59
ssam2actually .. how many concurrent connections can 'bottle' handle ?14:00
ssam2is it just one?14:00
radiofreerdale: it's built in baserock, however if you have a cross-compiler you can build it on your machine14:00
radiofreeor build it directly on the jetson itself14:00
rdaleok - i like to know how to build u-boot on the jetson itself14:02
madhupaulsherwood, how do we add packages in the vm image, also how to add proxy setting to access git. I'm following,  14:02
ssam2madhu: Baserock doesn't use packages. What do you need to do ?14:03
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]14:03
ssam2madhu: about the proxy, there's probably some environment variable that you must set14:03
ssam2let me check14:03
radiofreerdale: git pull git:// && cd u-boot && git checkout baserock/jetson/u-boot-tegra-next && make ARCH=arm jetson-tk1_config && make ARCH=arm CROSS_COMPILE=/usr/bin14:04
radiofreecreates a u-boot-dtb-tegra.bin file that is your u-boot.bin14:04
rdaleok, thanks i'll try that14:04
radiofreebtw you already have that u-boot installed14:05
rdaleyes, i know, but i wanted to know how to build it14:05
ssam2madhu: try setting HTTP_PROXY in the environment14:05
ssam2you can try it with `HTTP_PROXY= git clone xxx`14:06
radiofreemake sure you're cloning over http/https for that14:06
ssam2oh, good point14:06
madhussam2, thanks, I will try that, I was thinking about corkscrew for git14:07
madhuand it was not in the vm image by default14:08
radiofreemadhu: that's usually what i would use for ssh14:08
radiofree might be useful, using nc14:09
paulsherwoodmadhu: the closest instructions we have for adding things to baserock are at
aananth@ssam2, I tried "morph branch baserock:baserock/definitions default"14:09
aananthI get "Failed to fetch tarball, falling back to git clone.14:09
ssam2hmm, Morph tries to clone over git:// by default14:10
paulsherwoodaananth: that's a success message14:10
aananthI retried, I get "ERROR: /src/workspace/default: File exists"14:10
aananthOk, Great!14:10
*** Krin [] has quit [Remote host closed the connection]14:10
paulsherwoodaananth: it's not obvious, i know14:10
paulsherwoodupstream: would be nice to fix that message14:11
paulsherwood(and also work out why the tarballing is not working on gbo)14:11
madhuradiofree, thanks for the link, I will try them14:11
paulsherwoodaananth: so you should be able to cd default/baserock/baserock/definitions and proceed14:12
paulsherwood(sorry for the confusion)14:12
aananthYes Paul. I am on the last step....14:12
paulsherwoodssam2: what about ?14:17
pedroalvarezIs possible to execute a write extension installed in a system without the full path? (e.g. `python /usr/lib/python2.7/site-packages/morphlib/exts/rawdisk.write foo bar`)14:21
aananthThe last step appears to be "hang" at point::'2014-10-27 14:12:46 Deciding on task order'. I assume this is normal.14:22
paulsherwoodaananth: sadly, it is.14:22
aananthok :)14:22
paulsherwoodit's thinking, not hanging. now might be a good time to leave it running for the evening, check back in tomorrow :)14:22
pedroalvarezwell, I'd check after 5 minutes to see if it's running14:23
aananthCan I stop and redo it? I need to take the laptop to home and work from there. It is 7.55pm here.14:23
paulsherwoodyes, just ctrl C14:23
aananthOk. But I can wait if it will be over in few minutes.14:24
paulsherwoodyou may need to fiddle with network settings on switchin locations14:24
paulsherwoodi recommend you do the ctrlc. it will have a lot to rebuild14:24
aananthYeah, that is a learning... Ok I will do in another 10 mins.14:24
paulsherwoodaananth: normally i just reboot my vm on moving between office and home14:25
aananthI meant, "that will be learning (setting up network again in home).14:25
paulsherwoodaananth: yes understood. i think reboot may just work, though14:25
ssam2paulsherwood: +1 to that patch from me14:25
* paulsherwood looks around for another friendly maintainer14:26
paulsherwoodrichard_maw: ?14:26
paulsherwoodor pedroalvarez ? :)14:27
pedroalvarezha! :)14:27
pedroalvarezI'm not sure if it's ok to remove the "Failed to fetch tarball"14:27
paulsherwoodoh, it really is. it confuses users.14:28
paulsherwoodas the above conversation proves14:28
ssam2I've had a whole bunch of situations where people have seen that message and thought that it meant the command that they had run had failed14:28
ssam2I think this change is really useful14:28
pedroalvarezWhat  is going to show morph is fetching from tarball works?14:29
pedroalvarezthen maybe we shouldn't show anything here 14:29
richard_mawpaulsherwood: my issue with that patch is that the only way we know if it successfully fetched a tarball is absence of the message, but since we didn't have that before: +114:29
paulsherwoodbut note this 'fail' message has been appearing for as long as i can remember, and no-one has fixed it14:30
paulsherwoodrichard_maw: tvm14:30
ssam2removing the message would be OK, either.14:30
paulsherwoodwell it's user feedback. the git clone does take some time14:30
*** aananth [~caananth@] has quit [Ping timeout: 244 seconds]14:30
paulsherwood(and i assume it's slower than a theoretically successful tarball download, otherwise it would be the default?)14:31
ssam2yes, depending on the size of the repo `git clone` can be much slower14:31
richard_mawit's the repacking that a clone does which is the problem14:32
paulsherwoodso should i mail the patch to the list, or is it mergeable asis?14:32
richard_mawmergeable as-is in my opinion14:33
Kinnison+1 for text cleanup14:33
straycatoh, I noticed something in that series we just merged and forgot about it, there is a return following the yield, which must be unreachable?14:34
richard_mawthat's reachable14:34
richard_mawcontext managers must only yield once, so the return after the yield is to ensure it doesn't follow on to the rest14:35
* paulsherwood is happy he now understands context managers :)14:35
straycatOh I see okay14:41
* richard_maw loves context managers, since the majority of the code he writes requires you to set up and tear down resources14:42
richard_mawand context managers let you put the cleanup code with the setup code, rather than the code that uses the resource14:42
richard_mawa consequence of this though is that I tend to write a lot of tiny context managers for each resource I allocate, and have a massive with statement14:43
* richard_maw ponders adding to baserock14:43
richard_mawthat helps with the massive chain of context managers14:44
richard_mawand it allows more flexibility of which contexts are used14:45
richard_mawthe callback method on an ExitStack is roughly analogous to the defer statement from go, which appears to be how you tend to handle resource cleanup in go14:46
* richard_maw has been thinking about resource cleanup far too much lately14:47
*** aananth [~caananth@] has joined #baserock14:47
* pedroalvarez re-asks15:02
pedroalvarezIs possible to execute a write extension installed in a system without the full path? (e.g. `python /usr/lib/python2.7/site-packages/morphlib/exts/rawdisk.write foo bar`)15:02
ssam2they're programs15:02
*** aananth [~caananth@] has quit ["Leaving"]15:02
ssam2wait, I misunderstood you15:03
ssam2pedroalvarez: pkg_resources.resource_filename('morphlib', 'exts/rawdisk.write') to find the path to rawdisk.write15:06
pedroalvarezwow! that solves my problem of hardcoding the path15:06
richard_mawpedroalvarez: or `import morphlib\nwith morphlib.extensions.get_extension_filename('rawdisk', '.write') as path:\n    do_thing(path)`15:08
richard_mawor `path = morphlib.extensions._get_morph_extension_filename('rawdisk', '.write')`15:09
richard_mawthe former only matters if you've got extensions checked into the repository15:09
pedroalvarezI see15:10
ssam2I think this isn't done as part of a build or deploy, so there's no 'repository'15:10
ssam2I've set up a custom security group (firewall rules) for the Mason at
ssam2allowing global access to port :80 but only local access to port :808015:12
ssam2I'm suspicious that the reason the morph-cache-server keeps getting stuck is that it's exposed to the web, but it's only running in Bottle's development web server15:12
ssam2which maybe isn't up to the job, I don't know15:13
pedroalvarezI had the same feeling, that having 8080 open was causing the error15:13
pedroalvarezwell done :015:13
ssam2except now I've blocked off my own ssh access.15:14
ssam2I'll reopen that too ;)15:14
pedroalvarezWe should have various security groups for the different pieces of our infra15:14
ssam2yeah, we should15:15
ssam2the one I made is called 'private-morph-cache-server', but 'mason' would probably be a better name :)15:15
ssam2but Horizon doesn't seem to let me rename it15:15
pedroalvarezssam2: you can rename it :)15:15
ssam2oh yeah. it's now called 'public-mason' :)15:16
ssam2green mason again!15:17
Kinnisonssam2: Sounds like we need to harden the behaviour of those cache servers a little15:17
ssam2I shall merge now
pedroalvarezbtw, mason is accesible through this url as well:
ssam2oh, cool!15:20
*** zoli_ [] has joined #baserock15:26
*** zoli_ [] has quit [Changing host]15:26
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock15:26
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]15:30
*** bala [~bala@] has quit [Remote host closed the connection]15:46
*** genii [~quassel@ubuntu/member/genii] has joined #baserock16:06
ssam2seems like I broke the distbuild command :(16:07
ssam2TypeError: build() takes exactly 2 arguments (5 given)16:07
Kinnisonerm oops16:07
KinnisonHow did the test suite not catch that?16:08
KinnisonI thought richard_maw had managed to get distbuild integrated16:08
ssam2I think those tests are conditional on having flup and morph-cache-server in the system running the tests16:08
KinnisonAre those present in a devel system these days?16:08
richard_mawKinnison: yes16:08
ssam2they are now, but I've not updated all my systems since whenever it was merged16:09
ssam2my mind is too fried to work out what I did wrong here16:09
ssam2oh, it's because the InitiatorBuildCommand object is used, not the buildcommand.BuildCommand object16:09
* straycat thought distbuild used a different codepath16:13
straycatyou moved stuff out of build into build_command, but distbuild has an InitiatorBuildCommand and uses the distbuild_plugin16:14
straycaterm, into build_plugin16:14
ssam2I changed how was called without actually changing how it worked at all16:14
* straycat missed that16:15
* straycat facepalms16:16
straycatI forgot the distbuild commands were in the build plugin16:16
straycatRight and then that wraps build and yeah oops >.>16:18
ssam2seems that the distbuild code doesn't allow adding optional fields to the message types :(16:20
ssam2I can't just change the protocol, without having to redeploy all our existing distbuild networks16:20
ssam2might be easy to fix that, though16:21
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock16:22
straycatsadly, I also think that since we're going to have to rework that series anyway it would be worth making the build branch return (url, commit, ref) in all cases. Rather than (url, commit, ref) in one case and (url, ref, ref) in other cases.16:23
ssam2I think that's a separate change16:23
ssam2i'm not quite sure how it should work for temporary build branches, to be honest16:24
straycatI'm not exactly sure either at the moment, I've been looking at this code anyway so if we agree this is the right change to make I could try it out16:26
ssam2i can't get my head around it right now. It seems that if the user build a temporary build branch, its name *is* the ref that should go in the metadata16:28
ssam2because the content doesn't exist in any other ref16:28
ssam2but, that's a bit useless because the branch gets deleted as soon as it's built16:28
ssam2really, the problem is that temporary build branches are temporary :)16:28
persiaWhat are "flup" and "bottle" again?16:30
Kinnisonpart of the web framework we use for the http based protocols16:30
Kinnisonfor things like the cache server16:30
persiaZara: Sorry: disconnected.  I'm happy with a diagrammed concrete example, so long as it isn't trying to be authoritative.16:30
straycatssam2, Unless I'm misunderstanding something, it's the same either way in all cases you want the orignal_ref to go into the metadata but you want to resolve the definitions ref from a commit sha. If you have local changes which morph has committed and pushed, then the ref can be used instead of the sha, but I feel that it's more consistent for us to always use the sha.16:36
ssam2straycat: yeah, I guess you're riught16:39
ssam2straycat: so in that case it's really easy, just change what the pushed_build_branch() function returns16:44
ssam2I think I've worked out how to get distbuild working again :)16:44
ssam2no, maybe not16:50
ssam2there's no way of knowing what version of Morph the remote distbuild controller is running, so no way of knowing how many arguments can be passed to it16:51
ssam2I guess I have to bite the bullet and revert this patch16:51
ssam2and possibly spend some time making the distbuild protocol versioned16:51
ssam2no, I'm confusing myself16:52
ssam2this should work fine16:53
* ssam2 shuts up and tries actually testing it16:53
richard_mawssam2: you ought to be able to pass the commit in the build-request message, it'll get ignored if the controller is too old, but a new one will accept it, and can have code to handle not being given it16:55
ssam2yeah, that's what I've done16:59
ssam2however ... did the recent changes you guys did to change the message format to YAML change the protocol ?17:00
ssam2I guess it must have done17:00
ssam2I'm getting a strange error from 'tyrell'17:00
ssam2maybe it needs upgrading to work with latest Morph17:00
* radiofree hopes his eyes haven't been gouged out17:01
straycatI always have a cushion ready for that scene17:07
straycatssam2, I didn't think that the protocol got changed17:07
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]17:12
*** jonathanmaw [] has quit [Quit: Leaving]17:15
straycatIs okay?17:39
* ssam2 dreads the idea of deploying a new Trove, then remembers that in Openstack we can use the existing generic Trove image!17:42
*** tiagogomes [] has quit [Ping timeout: 265 seconds]17:42
ssam2ripsum: looks correct to me17:42
straycatwho's ripsum?17:47
ssam2I'd complain about people who use different IRC names in different networks, but I seem to have started doing that too :)17:47
jmacsI couldn't get "jmac" on here17:48
* richard_maw uses the same nick, but a different instance of irssi17:48
straycatI don't know who ripsum is, you must be confused17:48
straycatrichard_maw, does that change look okay to you?17:49
radiofreejjardon: i'll try your gtk patches on a jetson17:50
richard_mawstraycat: the indentation of the last line looks a little weird, as I'd align the last parameter in the tuple with the first, but +117:52
straycatOh okay, I can do that.17:53
jjardonradiofree: cheers! Ive just added a new patch on top so you only build the wayland backend17:55
Zarapersia: Ah, that's fine by me. I agree that it shouldn't be seen as authoritative, just as an explanatory tool. :)17:57
*** ssam2 [] has quit [Quit: Leaving]17:58
jjardonssam2: thanks for testing the coreutils patches! any idea how should I proceed now?17:58
jjardonhaha, he just escaped! 17:59
radiofreeok, rebuilding from glibc18:00
pedroalvarezjjardon: not sure about his opinion but I guess that adding the coreutils stratum to the build dependencies could works18:00
radiofreei'll leave this on overnight18:00
straycatAn unintended side effect of that patch is that the message "Building baserock:baserock/definitions master systems/foo-system.morph" would become "Build baserock:baserock/definitions sha systems/foo-system..."18:00
straycaterr Building18:00
straycatI haven't merged it, because there are some other variable name changes that should go along with it and that message should probably be modified so that we mention the ref rather than the sha.18:01
*** visteon [~visteon@] has quit [Remote host closed the connection]18:23
*** visteon [~visteon@] has joined #baserock18:23
*** ssam2 [] has joined #baserock18:30
Mode #baserock +v ssam2 by ChanServ18:30
*** ssam2 [] has quit [Client Quit]18:31
straycatso probably needs to go in as well19:21
*** zoli_ [~zoli_@linaro/zoli] has joined #baserock20:25
*** zoli_ [~zoli_@linaro/zoli] has quit [Remote host closed the connection]20:41
*** genii [~quassel@ubuntu/member/genii] has quit [Read error: Connection reset by peer]22:24

Generated by 2.15.3 by Marius Gedminas - find it at!