16:01:08 #startmeeting Mer Platform SDK status / planning meeting 16:01:08 Meeting started Fri Feb 24 16:01:08 2012 UTC. The chair is timoph. Information about MeetBot at http://wiki.merproject.org/wiki/Meetings. 16:01:08 Useful Commands: #action #agreed #help #info #idea #link #topic. 16:01:19 hello and welcome 16:01:29 who do we have here? 16:01:38 i'm present 16:02:14 lbt: around? 16:02:19 yes 16:02:24 good :) 16:02:51 #topic status update 16:03:23 hmmh 16:03:37 status from me: not really done anything on SDK side, except for a quick proof of concept with scratchbox2 within the sdk and building software/installing build dependencies for a target 16:04:08 I've also been quite busy with other things so haven't done much 16:04:08 A few points from me 16:04:29 Page here is reasonably up to date http://wiki.merproject.org/wiki/Platform_SDK 16:04:40 Been working on using kickstarter to prepare images 16:04:41 I did start to improve the helper script but then again lbt had already done something with it :) 16:05:01 Yeah, that's .... grown :) 16:05:19 https://github.com/lbt/sdk-kickstarter-configs 16:05:29 has the latest published sources 16:05:36 ok thats good 16:05:39 so images 16:05:47 Available under https://img.merproject.org/images/sdk/ 16:05:48 Focus on mic/osc so far - x86 project building and all image building works 16:06:05 you can essesntially follow the SDK page and build stuff 16:06:09 including the SDK 16:06:13 which is nice 16:06:18 #info sources available from https://github.com/lbt/sdk-kickstarter-configs 16:06:32 I also updated several packages too 16:06:43 #info images from https://img.merproject.org/images/sdk/ 16:06:58 #info OBS project https://build.pub.meego.com/project/show?project=Mer%3ATools%3ATesting has tools 16:07:17 and that includes latest mic/spectacle/kickstarter 16:07:22 with some bugfixes 16:07:31 I updated all relevant wiki pages (I think) 16:07:43 cool 16:07:59 so http://wiki.merproject.org/wiki/Category:About should link to sane information for all of them 16:08:16 Produced some yaml for an sb2 kickstart too 16:08:26 based on Stskeeps' PoC 16:08:37 https://build.pub.meego.com/package/view_file?file=00sdk.yaml&package=sdk-kickstarter-configs&project=Mer:Tools:Testing 16:08:43 for those who want gory details 16:08:51 There are some problems on sb2 16:09:06 essentially our cross-tools are built for ssse3 processors 16:09:08 :O 16:09:16 so my AMD desktop is sad 16:09:26 (like MeeGo all over again!) 16:09:38 * timoph was just about to say that :) 16:09:39 but Stskeeps assures me that this will be sorted before he sleeps again 16:09:43 so next week some time 16:10:05 will be fixed in next-next core release as changing that detail can break ability to release 16:10:11 :) 16:10:19 yeah - that is next week? 16:10:28 post tuesday iirc? 16:10:31 :nod: 16:10:43 OK - so it's not worth worrying about until then 16:11:00 since we have a basic platform sdk going, our documentation should revolve around that method 16:11:06 ie, no more 'mic2' 16:11:09 I have hacked a bit at the sb2 side and it looks like it'll be reasonable 16:11:11 totally 16:11:18 I've removed mic2 from tools and wiki 16:11:30 (although pages still need editing as per the bug) 16:11:40 but it also means we have a starting point for vendors, "install platform sdk, try to make an image.." 16:11:55 yep - and that's working today 16:11:59 so we can iterate developers/vendors alike on how easy it is to use mer 16:12:01 anything else for the current status? 16:12:08 we also need to extend that to include Vendor repos 16:12:16 timoph: yes, I'm reworking the enter_chroot 16:12:18 again 16:12:22 :) 16:12:33 so it has mer_sdk_chroot mount|enter|umount 16:12:37 still a shell script? 16:12:45 and you can enter it multiple times 16:12:45 lbt: mount does what specifically? 16:12:46 yes 16:12:50 .... 16:12:52 mounts? 16:12:59 yes, mounts what, a horse, a duck or /home ? 16:13:11 hey, keep it clean! 16:13:34 so it does /proc and /sys etc 16:13:44 and all data mounts 16:13:46 and /home 16:14:00 it also allows user hooks 16:14:15 so I use that because I have some / symlinks I like 16:14:32 so the default is to mount "all" but one can opt out 16:14:44 most of the complexity is traps and such like to minimise risk of stale mounts 16:15:34 can we check in 'enter' that 'mount' hasn't been done? 16:15:36 and instruct accordingly 16:15:39 yes 16:15:46 and advise upon exit, too 16:15:52 to umount before rm -rfing 16:16:20 #topic planning 16:16:21 http://pastie.org/3447279 16:16:28 mmm poor formatting 16:16:35 the rm -rf issue is hard 16:16:41 I will investigate immutable 16:17:09 if you bind mount /home to /fred/home and rm -rf /fred ... 16:17:17 there's not a lot we can do 16:17:36 we can't protect the user from everything 16:17:48 if one does stupid things then one does 16:17:50 ok, so planning: we can start with designing the sb2 side next week, with implementation after first prerelease 16:17:51 and since people can open 2 shells at once ... then whilst /home is mounted on the SDK if they rm the SDK ... game over 16:17:56 yep 16:18:22 I discussed with you this morning and I think we won't need to do much wrt more mounts 16:18:23 it's quite powerful proposition as we have a genuinely fast and easy to use build environment 16:18:38 yes 16:18:49 I'm aiming to have a variety of images available too 16:19:02 for the sdk? 16:19:03 and make it easy for a vendor to make their own which includes their code 16:19:06 yes 16:19:16 some uses will be non-SDK 16:19:22 eg mic bootstrap 16:19:51 ok 16:20:17 the sb2 stuff is actually the thing I'm most clueless about 16:20:31 yeah, I'm going to document it 16:20:31 so what do we actually need to do there 16:20:33 My SDK is installed in /opt/mer/mer-platform-sdk :) 16:20:49 i can guide through that as well, we should just get over some basic installation problems 16:20:52 alterego: yeah, I do that 16:21:15 but the premise is that you can generate a random mer image, initialize it as a SB2 target, install development headers, and build easily software within it 16:21:33 with cross compilers and fast tools 16:21:43 ok. I can document the thing while someone walks me through it 16:21:59 should be a good sanity check 16:22:30 so, a realistic goal for next week: guide that describes platform sdk install, creation of custom image with kickstarter, basic idea for how hardware adaptations can provide kickstarter yamls, building image? 16:22:32 so I think our next goal is sb2 building 16:22:57 and sb2 building, that is 16:23:07 sounds good 16:23:25 an example can be that x86 adaptation should provide a yaml installable into the sdk somehow 16:23:46 yeah - I may need to rework kickstarter somewhat 16:23:52 so we can do lego block style image building 16:23:59 "take this hardware adaptation, this mer core, this UI.." 16:24:08 exactly 16:24:30 before we go too far there though 16:24:39 I'd like to validate the SDK against Nemo 16:25:36 should be a good proof that it works if you can build nemo with it 16:25:47 build a nemo package anyway 16:25:58 ideally not using osc build 16:26:21 :nod: 16:26:24 Stskeeps: so I'd like to declare an sb2 minimal target 16:26:45 as in? 16:26:58 a cross-build-essential? 16:27:02 as you did with mer-core-arm7l 16:27:28 okay, so basic development headers and so on in there? 16:27:36 mic cr fs -A armv7l mer-core-armv7l-base.ks 16:27:49 then sb2-init it 16:27:56 then install devel headers 16:28:00 right 16:28:01 then do something like sb2 install-build-depends SMS-tool.spec 16:28:17 are we talking about a build-essential, or a basic usecase 16:28:25 usecase 16:28:28 ok 16:28:39 *then* do something like sb2 install-build-depends ophono.spec 16:28:56 and have both ophon and sms-ui build-dep in the root 16:29:09 and have both source packages in the SDK 16:29:34 makes sense 16:29:35 then see how the source in the SDK uses the ophono in the SDK 16:29:42 which will hurt I think 16:30:17 but if we solve that in a general way, I think we end up with make world ? 16:30:19 it can, i guess, but we'd need a publish-package kind of thing 16:30:30 fwiw 16:30:37 first goal is to simply ./configure, make kind of thing 16:30:37 yes make install would install to sb2 16:30:51 rpmbuilding is a little more involved as a start 16:31:03 yeah ... anyhow... some vague ideas 16:31:12 :nod: 16:31:27 I'm also keen to keep stuff out of the SD 16:31:28 K 16:31:43 use the fact that we have bindmount to the desktop 16:31:43 mm? 16:31:47 well, of course 16:31:57 sdk is supposed to be replace-able 16:32:18 your targets would go in your homedir, i guess 16:32:55 I'm thinking that git/emacs/eclipse/QtCreator - all run outside SDK 16:33:07 git maybe not ... but maybe 16:33:10 I think git should be in SDK ;) 16:33:14 :) 16:33:22 alterego: I kinda don't 16:33:30 I was actually going to play around with getting Qt Creator to work with the chroot 16:33:31 it's an editor 16:33:39 alterego: with but not in? 16:33:43 Well, I don't think it really matters too much. 16:33:47 Not in 16:33:48 With 16:33:48 lbt: weigh it with practicality 16:33:59 no - but I'm really interested in mininimising support costs 16:34:00 and let's see how people use platform sdk 16:34:10 exiting in and out of a chroot is not pleasant 16:34:13 i do that with osc 16:34:13 I'd personally like git in there. 16:34:23 I have issues with using scratchbox for harmattan development and not having git. 16:34:45 alterego: *nod* 16:34:49 (obviously I've added git to it now) but more terminals/tabs sometimes annoys bme. 16:35:21 lbt: at least our set is limited by what is in mer core anyway, and what can be used to build things in OBS 16:35:24 otoh, who says everyone uses git ;) 16:35:25 I bring it up as an edge-case 16:35:39 i wouldn't outright ban it, i would just say some tools are second tier 16:35:48 ie, primary usecases and secondary helpers 16:35:57 *nod* 16:36:21 So, the platform SDK, is that meant as an image creation and platform package development area? 16:36:25 git is a normal tool to work on mer (git clone ssh://review..) 16:36:30 I'm with Stskeeps on let's do it so that it works and see how people use it and change it accordingly 16:36:30 so i would reason that to be part 16:37:07 alterego: tools needed to develop with mer on a platform level, apps is ideally with qt creator instead 16:37:12 I'm just wondering about the application sdk, which obviously doesn't actually exist yet, but would it be a similar architecture? 16:37:38 So, would that mean we need to get sysroots for madde? 16:37:56 yes, that'll come in some form, but first we need a proper platform sdk 16:38:03 else people can't contribute or use mer properly :) 16:38:17 It would be cool if a vendor could go: "platform: packages; sdk: libqt, libofono, etc" 16:38:19 app sdk wouldn't be chroot based 16:38:41 And have the platform SDK spit out a device image and an application sysroot. 16:38:49 for example 16:38:56 yep 16:38:58 Where the sysroot only has the -dev versions of "public" API packages, like qt, ofono. 16:39:08 anyway, we're drifting 16:39:14 Sorry ;) 16:39:23 so, we have a plan for the next week? 16:39:29 so to summarize 16:39:54 next weeks targets: setup guide, image building and sb2? 16:40:00 was there something else? 16:40:31 [17:22] so, a realistic goal for next week: guide that describes platform sdk install, creation of custom image with kickstarter, basic idea for how hardware adaptations can provide kickstarter yamls, building image? 16:40:35 [17:22] so I think our next goal is sb2 building 16:40:36 [17:22] and sb2 building, that is 16:40:38 [17:23] sounds good 16:40:39 [17:23] an example can be that x86 adaptation should provide a yaml installable into the sdk somehow 16:40:42 yeah 16:41:03 also, for good measure: write up use cases as we go along 16:41:05 for QA later 16:41:12 yes 16:41:24 * sledges can't wait :) 16:41:27 timoph: can you update the wiki page at the bottom too 16:41:34 delete old stuff 16:41:38 lbt: I can 16:42:09 meet again next week? Earlier for Sage? 16:42:38 so we have enough time to spare to get those things done? 16:42:56 *g* .... nothing but time :) 16:42:59 I was thinking we could meet again on fri 16:43:13 well. I only have the evenings 16:43:53 so what would be a good time for Sage? anyone know? 16:44:00 perhaps on thursdays instead? 16:44:07 works for me 16:44:18 sure 16:44:32 let's just put it out to a discussion on mailing list, but thursday may work better for people 16:44:32 after 15 utc that is 16:44:43 i know for a fact i'd be in a bar if i was working in an office on a friday at this point 16:44:46 :P 16:44:55 :) 16:45:07 so, AOB? 16:45:14 #topic AOB 16:45:22 anything else? 16:45:43 feedback on what we have delivered so far in platform SDK/documentation/etc as well 16:46:17 The whole tools area is in rapid churn at the moment too 16:46:18 do we have any brave souls experimenting with the sdk? 16:46:33 we could ask feedback/input from the mailing list 16:46:35 oh yes 16:46:43 sonach has tried it I think 16:46:53 and others 16:47:14 would be good to hear what they would like to see in it 16:47:17 yep, jackylau this morning 16:47:53 I tried it too last week 16:48:03 the important thing is that we get the basic use cases going nice and stable first, then move on to more advanced 16:48:05 will give it another go over the weekend 16:48:15 this is going to be the primary interface to mer for many 16:48:22 I'm using it at the moment :) 16:48:33 Not done anything serious yet except build a few images. 16:48:46 Gonna be attempting to build binaries later. 16:48:51 I'm using it for images but waiting for sb2 for building 16:49:00 alterego: you got ssse3? 16:49:07 Yup 16:49:24 I can give you some sb2sdk .ks that you can try 16:49:44 Sure 16:49:53 #mer later then :) 16:49:59 ok. if there's nothing else. I think we can call it 16:50:15 do we have a SDK bugzilla section? 16:50:24 we do 16:50:46 and "ssues with the SDK itself rather than the tools within it. If a Mer tool has a problem in the SDK that's a tool bug" 16:51:03 is the comment 16:51:12 with an "I" on the front :) 16:51:18 :) 16:52:03 #info bugzilla has a section for the SDK (don't file tool bugs there) 16:52:44 we don't have any CI process for Tools yet 16:52:57 ok. I'll send email about the meeting time on thu and other one asking for sdk input/feedback 16:52:59 nor do I plan to in the next week 16:53:28 anything else or can we move to #mer 16:54:13 ok. thanks for attending 16:54:16 #endmeeting