Skip to end of metadata
Go to start of metadata

GENIVI Tools Team Meeting Starts[edit]

       klausbirken	Hi all!	12:59
       paulsherwood	hi all	12:59
       * paulsherwood won't ask for atattendees formally...	13:00

Quick IRC meeting how-to[edit]

       paulsherwood	- we can start informally for now, see how it goes	13:00
       paulsherwood	- folks are welcome to chime in at any time	13:00
       paulsherwood	- I'll paste topics from the agenda as we go	13:00
       paulsherwood	- any immediate questions?	13:00
       *** Joel_Replogle has joined #automotive	13:00
       * paulsherwood takes silence as approval :)	13:00
       *** pavel_ has joined #automotive	13:01
       paulsherwood	initial agenda was at	13:01
       paulsherwood	so first item is...	13:01


       * paulsherwood wonders if joel is here	13:01
       Joel_Replogle	yes, I am here	13:01
       paulsherwood	Joel_Replogle: welcome. please could you update us on what the 'plan' is for wikis?	13:01
       *** pavel_ is now known as pavelk	13:02
       Joel_Replogle	Hi all - We currently have 2 active wikis... the media wiki at,	13:02
       *** pavelk is now known as pavelkonopelko	13:02
       Joel_Replogle	and the new wiki	13:02
       Joel_Replogle	Based on feedback, we are migrating from mediawiki to confluence	13:03
       Joel_Replogle	Status of confluence:  We have set up with the Atlassian cloud tooling for both Confluence and JIRA	13:03
       paulsherwood	if others have managed to confirm they are happy with the new confluence approach, please could they please confirm?	13:03
       pavelkonopelko	yes, I can confirm	13:04
       paulsherwood	great, thank you	13:04
       CTtpollard	I've not had hands on experience with confluence, but it looks neat	13:04
       * jeremiah is happy.	13:04
       pavelkonopelko	we only should agree on the high-level structure, before it grows too creative	13:04
       *** scrumb has joined #automotive	13:04
       paulsherwood	agreed. Joel_Replogle - do you have thoughts on how to agree/implement the structure?	13:04
       Joel_Replogle	We've learned that the cloud-based version of Confluence/JIRA does not support having a * domain... this was viewed as a show-stopper.  So we are currently pursuing another transition, from cloud to server versions.	13:04
       paulsherwood	but that transition should (just) be a db migration i believe?	13:05
       jeremiah	Its been suggested that we look at other projects that we think organize their wiki well and try and copy.	13:05
       klausbirken	I am also happy with the new confluence approach, and I support Pavel's claim for a missing high-level structure.	13:05
       paulsherwood	jeremiah: food idea. any particular recommendations?	13:05
       Joel_Replogle	Yes, the migration from cloud to server is a smooth process, since it's within their supported tooling.	13:05
       paulsherwood	s/food/good/	13:06
       pavelkonopelko	Are there any features that will be lost/gain because of the translation?	13:06
       steve__	Not tried any copy yet :( No time	13:06
       paulsherwood	pavelkonopelko: were/are you ok with the toplevel structure at could that be the template?	13:07
       paulsherwood	steve__: i know the feeling :)	13:07
       pavelkonopelko	well, could be better	13:07
       pavelkonopelko	I would have to have a second look to be more specific	13:07
       Joel_Replogle	features lost: no config/infrastructure for GENIVI... features gained: ability to have additional plugins, such as those supporting v3.x style wiki formatting.	13:07
       paulsherwood	in the interest of timekeeping in this meeting could i suggest that a subset of attendees take a separate discussion about wiki topelevel structure?	13:08
       pavelkonopelko	should we organize a telco ? :-)	13:08
       Joel_Replogle	it's a good topic, and I will organize a call	13:08
       Joel_Replogle	:-)	13:08
       paulsherwood	if you must :)	13:08
       paulsherwood	interested parties say aye...	13:08
       pavelkonopelko	I am in	13:09
       paulsherwood	(note interest is likely to include doing some of the work itself)	13:09
       paulsherwood	jeremiah: ?	13:09
       Joel_Replogle	indeed!  it is a wiki, after all :-)	13:09
       steve__	aye	13:09
       jeremiah	Yes, I think we ought to all look at examples we like and submit suggestions	13:09
       jeremiah	Otherwise we all become UX and UI information experts	13:10
       paulsherwood	ok, steve__, pavelkonopelko, jeremiah, Joel_Replogle - can i leave you folks to move on this and report back next week?	13:10
       jeremiah is actually a pretty good front page.	13:10
       Joel_Replogle	yes, that sounds good.	13:10
       paulsherwood	ok, unless any pressing further points on wikis, i propose to move on	13:10
       Joel_Replogle	Once quick note...	13:11
       Joel_Replogle	Even though we are in transition mode, the transitions will be seamless (or close to it), so no need to delay using the new confluence/JIRA.	13:11
       paulsherwood	noted. thank you	13:11


       paulsherwood	Any reviews, comments, or shall we just press ahead?	13:11
       steve__	define press ahead :)	13:12
       paulsherwood	well... i elephants get eaten best one bite at a time	13:13
       paulsherwood	i can make simple proposals for some parts of CIAT	13:13
       paulsherwood	hopefully somee are non-controversial	13:13
       klausbirken	paul: on the CIAT page, there are candidate solutions, which tool is the current favorite, Jenkins?	13:13
       paulsherwood	particularly given that we can change them later	13:13
       paulsherwood	klausbirken: i'm unclear on that one...	13:14
       paulsherwood	but let's start at the beginning...	13:14
       paulsherwood	code review: anyone strongly object to gerrit?	13:14
       jeremiah	Currently we do code review on mailing lists.	13:14
       jeremiah	Moving to Gerrit might need to have consensus	13:14
       jeremiah	Which is fine, but we'll have to speak to maintainers	13:15
       paulsherwood	'speak' meaning email?	13:15
       jeremiah	Sure, or here or wherever	13:15
       * paulsherwood wonders if any maintainers are here and have an opinion	13:15
       CTtpollard	I think gerrit is a better choice than mailing list patches, however it will require extra overhead and might create extra friction to newcomers	13:16
       paulsherwood	jeremiah: we could use gerrit to trigger ci, and still allow ml patches	13:16
       jeremiah	Some companies currently will not submit patches publiclly but prefer to mail the maintainer directly and then new releases contain the pathces.	13:16
       *** manfred_ has joined #automotive	13:16
       KlausUhl	We are using Gerrit internally. I would not want to trade it in for email based reviews. The only thing that I don't like about it is the fact that it is hard to find newly added review comments.	13:16
       paulsherwood	jeremiah: i'd like to discourage that practice :)	13:16
       *** gan has joined #automotive	13:17
       CTtpollard	I don't think non public emailing is the correct way	13:17
       klausbirken	I am not a reviewer, anyway: gerrit is fine for me, I esp. like the concept that the automated tests of the CI build is just another review input.	13:17
       jeremiah	This is an anti-pattern I know but its hard to sometimes move people away from company policies.	13:17
       paulsherwood	KlausUhl: that may be config?	13:17
       KlausUhl	Maybe.	13:17
       KlausUhl	I am not the maintainer of that system :-)	13:17
       paulsherwood	jeremiah: any objection to us adopting gerrit while continuing to allow ml patches?	13:17
       KlausUhl	Maybe it is just a matter of the presentation in the web frontent.	13:17
       jeremiah	paulsherwood: I like the idea of gerrit for CI triggers and mailing list patches, sounds good!	13:17
       paulsherwood	any other +1s?	13:18
       KlausUhl	+1 from me	13:18
       paulsherwood	any -1?	13:18
       pavelkonopelko	I am ok as long as adopting is not mandatory	13:18
       pavelkonopelko	...for everybody	13:18
       paulsherwood	ack	13:18
       * paulsherwood takes this as a provisional decision, will attempt to make it so	13:18
       pavelkonopelko	...just a tool available for those who want to use ti	13:18
       paulsherwood	next...	13:18
       paulsherwood	git mirroring	13:18
       steve__	+1	13:19
       paulsherwood	i'd like to propose baserock's lorry process... it works, and collects upstreams on an ongoing basis	13:19
       jeremiah	Does this refer to mirroring GENIVI git repos?	13:19
       paulsherwood	does anyone know of other forss solutions?	13:19
       jeremiah	Or are you referring to upstream repos?	13:19
       *** philrobn has joined #automotive	13:19
       paulsherwood	both i believe	13:20
       CTtpollard	I would want to mirror both	13:20
       CTtpollard	the problem with deprecated upstream repos has been a problem for meta-genivi-demo for example	13:20
       jeremiah	So you're proposing (for clarity) a single instance of all the source code GENIVI uses in one location via git mirrors	13:20
       jeremiah	?	13:20
       * paulsherwood thinks that ci from the wild internet is rather upredictable	13:20
       paulsherwood	jeremiah: yes	13:20
       jeremiah	That sounds pretty awesome actually.	13:21
       jeremiah	Would likely be a good resource and backup	13:21
       paulsherwood	nb... not not, but a genivi/agl/automotive equivalent	13:21
       steve__	also stepping stone for LRT binary distro requirement	13:21
       jeremiah	One wouldn't edit this mirror directly right? It would slurp upstream repos and put them in for CI to consume?	13:21
       paulsherwood	correct	13:22
       paulsherwood	(on both points)	13:22
       jeremiah	steve__: Yes, this is a key enabler fro that.	13:22
       paulsherwood	do i i take it we have some +1s?	13:22
       jeremiah	Expensive to maintain and host though	13:22
       jeremiah	Is there a volunteer?	13:22
       CTtpollard	paulsherwood: are there any example of project troves / lorries for non baserock system purposes?	13:22
       paulsherwood	CTtpollard: i believe there are, but i don't have any to hand	13:23
       paulsherwood	jeremiah: you mean in terms of sponsoring hosting?	13:23
       jeremiah	+1	13:23
       jeremiah	paulsherwood: Yeah	13:23
       *** gan is now known as gunnarx	13:23
       paulsherwood	i believe we can find some... if folks agree this is a good approach	13:24
       jeremiah	Then I'll just leave this here . . . +1	13:24
       steve__	hard to say. not looked at tech. What options are out there and what are the requirements?	13:24
       paulsherwood	i don't want to be accused of 'pushing' baserock, though... so if anyone is aware of viable alternatives please speak up	13:24
       pavelkonopelko	+1 for upstream mirroring	13:24
       steve__	+1 for mirroring	13:25
       gunnarx	Meta: The speed of this meeting is hard to keep up with, tbh.	13:25
       paulsherwood	:)	13:25
       gunnarx	First question: I don't know what the "Lorry approach" contains.  You have not explained that Paul.	13:25
       paulsherwood	gunnarx: it's easier than the phonecalls :)	13:25
       gunnarx	Do you then intend to change the yocto recipes that download tarballs to use git instead?	13:25
       pavelkonopelko	lorry is a tool, a part of baserock suit	13:25
       paulsherwood	fair... lorry is just a program that can convert upstream repos from whateveer  vcs/tarball into git	13:26
       gunnarx	I believe yocto has a mechanism for mirroring sources locally, has it been looked at and compared.	13:26
       paulsherwood	there is an appliance (called trove) that does this on an ongoing basis	13:26
       gunnarx	I don't know it personally, but assume that if the recipe calls for a tarball it will mirror a tarball?	13:26
       steve__	yes yocto has simple mirror facility	13:26
       gunnarx	(the yocto approach that is)	13:26
       paulsherwood	gunnarx: afaik yocto doesn't have a tool for collecting updates, nor anything for converting to git	13:27
       paulsherwood	but i may be mistaken	13:27
       gunnarx	paulsherwood, my question 6 lines up...	13:27
       paulsherwood	i'm happy to put this on the list for discussion (and we could ask at yocto too for example)	13:27
       paulsherwood	you mean 14:26 < gunnarx> I believe yocto has a mechanism for mirroring sources locally, has it been looked at and compared.	13:27
       gunnarx	No 10 lines up, or so.  I'm trying to understand what it means for the yocto recipes to convert all to git	13:27
       steve__	an additional maintainer demand would be one aspect.	13:28
       gunnarx	While I think the baserock approach is good, I don't think our first step should be to fork and modify multiple various yocto layers	13:28
       paulsherwood	ok, this seems more controversial than the gerrit decision :)	13:29
       steve__	When it breaks and its different from upstream	13:29
       pavelkonopelko	paulsherwood: is lorry / trove capable of retrieving the 'original' archives?	13:29
       gunnarx	I'm guessing, without much data, that the yocto layers include tarball downloads, maybe a few odd hg and bzr etc.	13:29
       steve__	not seen any -1s for the high level requirement. It's the implementation... :)	13:29
       paulsherwood	we could take this separately - i don't want to hog the meeting on this one topic... any suggestions for best action point?	13:29
       pavelkonopelko	... that is storing in git and returtning in original format?	13:30
       steve__	yes gunnar, git, svn, tar balls etc.	13:30
       paulsherwood	ok... i suggest this topic continues at a more leisurely pace after the meeting (over coming days)	13:30
       paulsherwood	unless anyone has further urgent points, i believe other aspects of CIAT also need discussion	13:31
       gunnarx	It's not about being controversial, I just feel you are moving fast and uncertain if things are being thought through.  I'm all for good solutions	13:31
       paulsherwood	(but note that i've been previously been guided that if i just 'get it done' maybe folks would be happy to adopt)	13:31
       paulsherwood	gunnarx: i believe that there are a reasonable number of interested stakeholders that would like to see *something* for CIAT operational quickly...	13:32
       paulsherwood	even if it's only a strawman in parts	13:32
       * paulsherwood personally would like to get something lashed together by end of august, early sept at the latest, as a startpoint	13:33
       gunnarx	paulsherwood, yeah, please don't strawman me.  As I said, I'm not opposed to any good solutions, and certainly not putting them in place quickly either.	13:33
       gunnarx	But to be honest the Wiki migration is still in flux, and no one seemed to bring it down to earth on this discussion, afaict	13:34
       pavelkonopelko	that's why we agreed to continue in a conf call :-)	13:35
       steve__	Search on CIAT on open wiki only brings up hits in TT minutes. Is that because it is on new confluence?	13:35
       paulsherwood	gunnarx: i'm unclear on next step based on what you're saying (for CIAT... i thought we'd done wiki already)	13:35
       klausbirken	steve_:	13:35
       * paulsherwood proposes to move on to next agenda item, and clear up CIAT via the mailing list	13:36
       gunnarx	OK, but now the meeting is here  :)  Just going on record that I'd like to understand what mirroring all sources in git (something I like in theory) means for the yocto recipes that may fetch sources from elsewhere (forking and maintaining lots of layers I don't like so much)  Sorry for repetition.	13:36
       paulsherwood	gunnarx: noted	13:37
       paulsherwood	any objections to moving forward?	13:37
       gunnarx	ok by me	13:37
       paulsherwood	tvm	13:37


       steve__	yocto would take normally take different approach. mirror as is and use pre-mirror	13:37
       steve__	onwards.	13:37
       paulsherwood	klausbirken: please could you lead this topic... i need to rest my typing a bit :)	13:38
       klausbirken	Ok - so this is about test generation…	13:38
       klausbirken	I am looking into test generation from Franca interfaces. I created a Trello card ("doing") for this and a new page in our cool new confluence:	13:38
       klausbirken	Up to now, I collected some ideas and aspects for this topic, and looked a little deeper in what could be done with Franca and fMBT (which is quite cool).	13:38
       klausbirken	You are invited to add stuff to the page... maybe someone has used fMBT before or similar tools before and can give input	13:39
       klausbirken	Or put a new section "requirements" on that page.	13:39
       paulsherwood	klausbirken: this looks like a great start, thank you!	13:39
       klausbirken	I will try to make it more concrete, provide some examples until next week.	13:40
       paulsherwood	are others interested in working with klausbirken on this? klausbirken - are there things you awould like help with?	13:41
       klausbirken	Input on test frameworks you are using would be useful…	13:41
       KlausUhl	I would be interested to contribute to this. I am just not sure how much time I will be able to "donate".	13:42
       paulsherwood	for CIAT? or input from othrs on their preferred test frameworks?	13:42
       * paulsherwood thinks CIAT needs to support multiple frameworks	13:42
       KlausUhl	For generating tests with Franca	13:42
       klausbirken	I am not sure, fMBT for example is more than a test framework, it can generate test sequences from independent test steps	13:43
       klausbirken	there might be similar tools, which we should look into a little deeper	13:43
       *** jeremiah has quit IRC	13:44
       paulsherwood	ok. klausbirken i think it would be worth you emailing to the list specifically on this, better than it being lost in the minutes of this meeting ;)	13:44
       gunnarx	The discussion about frameworks is a never ending story.  The fact is that you need "one system" that executes tests that are written in multiple frameworks.  It's not rocket science really and typically lands the solution in some shell scripts as glue, CIAT executes.	13:44
       gunnarx	Just do it already :)	13:44
       *** jeremiah_ has quit IRC	13:44
       paulsherwood	gunnarx: yup	13:44
       paulsherwood	ok any more on CommonAPI today?	13:45
       steve__	+1=>gunnar on executor	13:45
       gunnarx	Yes	13:45
       klausbirken	not from my side	13:45


       klausbirken	gunnarx: yes, this is the CIAT aspect. I am looking into generation of tests from interfaces, not execution of tests.	13:45
       gunnarx	If I may take the cue to comment quickly on franca_install_automation.  CommonAPI install is broken still	13:45
       klausbirken	Regarding AMM: I didn't prepare the agenda proposal for Franca yet, still todo.	13:46
       paulsherwood	can someone volunteer to check that TT sessions are properly shepherded through GENIVI process?	13:46
       gunnarx	Wait, which topic are we on?	13:46
       paulsherwood	i moved onto AMM, gunnarx. but i note your comment about CommonAPI	13:46
       gunnarx	It was just introduction to a request for help, but if you want to move on...	13:47
       klausbirken	AMM first, CommonAPI afterwards?	13:47
       paulsherwood	yes please	13:47
       gunnarx	ok, go on	13:47
       * paulsherwood was a bit hasty in moving on, and apologises	13:47
       pavelkonopelko	re shepherding, it's most easily done in the PMO	13:48
       paulsherwood	pavelkonopelko: do i take that as your raising your hand? :)	13:48
       pavelkonopelko	...whomever joins on a regular basis PMO calls does	13:48
       pavelkonopelko	... raise their hands	13:49
       klausbirken	paulsherwood: TT "shepherd" - maybe we can sync with philrobn right now, he is online	13:49
       paulsherwood	pavelkonopelko: i do, but i keep having to drop off it	13:49
       paulsherwood	klausbirken: good idea, but i believe the process is not public?	13:49
       klausbirken	paulsherwood: I see, you're right.	13:49
       * paulsherwood can maybe voluntell someone, as walt called it, if all else fails :)	13:50
       paulsherwood	ok if no takers, i'll sort it separately	13:50
       paulsherwood	any more points on AMM?	13:51
       paulsherwood	5	13:51
       paulsherwood	4	13:51
       paulsherwood	3	13:51
       paulsherwood	2	13:51
       gunnarx	No the content of the GENIVI Members meeting is not planned in public.  The content of conferences that GENIVI host for the entire industry is of course published.	13:51
       gunnarx	Typical process is to add sessions to the Wiki page containing all proposed sessions, P Robin schedules them if they are ready to be scheduled	13:51
       gunnarx	So, that's the needed shepherding right now I think?  :)	13:51
       paulsherwood	yes. who will do it, though. me i guess :)	13:52
       gunnarx	Team lead is a good guess :)	13:52
       paulsherwood	any further comments on AMM before we return to CommonAPI	13:52
       gunnarx	But could be by the topic owner/speaker of course.	13:52
       gunnarx	I'm done	13:52
       * paulsherwood notes that if gunnarx had explained how much work this actually involves, he might not have been so quick to volunteer :_)	13:52

CommonAPI redux[edit]

       paulsherwood	gunnarx: have you raised a bug?	13:53
       *** manfred_ has quit IRC	13:53
       gunnarx	let me restart	13:53
       paulsherwood	(and incidentally should be be doing that using Jira now?)	13:53
       gunnarx	Now you're jumping to another topic.	13:53
       paulsherwood	sorry	13:54
       paulsherwood	gunnarx: please go ahead	13:54
       gunnarx	OK, I saw a cue to comment on franca_install_automation	13:54
       gunnarx	I want to inform:  It does not yet install commonAPI but I'd like it to.  Maybe someone can help out.	13:55
       gunnarx	I think the problem right now is C API depends on CDT Tools.   I may have been too eager to add all CDT to the install, because it seemed useful ,but it landed me in dependency hell	13:55
       gunnarx	Which is maybe normal in Eclipse ecosystem? :)	13:55
       klausbirken	gunnarx: dependency hell is not only in the Eclipse space :-)	13:56
       gunnarx	The problem is, the automation does not use Eclipse's built in dependency meta data	13:56
       gunnarx	So I must explicitly install every needed package, in the right order..  So far it's been OK, but now I feel like there are circular dependencies.	13:56
       klausbirken	I see - I created an installation locally I while ago - I can check.	13:56
       klausbirken	Is there a branch with your current state?	13:57
       gunnarx	Also difficult to find out where exactly some required packages actually have their home.	13:57
       gunnarx	Yes, I believe I pushed cpp_common_api branch	13:57
       *** jeremiah has joined #automotive	13:57
       *** jeremiah_ has joined #automotive	13:57
       klausbirken	Ok - I'll check.	13:57
       *** bbranch has joined #automotive	13:57
       gunnarx	OK, my request for help has been noted, and status of the project informed.  I'm hopeful it's solveable.	13:58
       klausbirken	Last question: CommonAPI 3.1.3?	13:58
       gunnarx	Yes, I believe that's the one I tried.  I could come back in a minute with more info, or we could actually take it outside of the meeting.	13:59
       paulsherwood	outside the meetign please. ok, if that completed the CommonAPI redux... shall we move to AOB?	13:59
       gunnarx	yes, ok	13:59
       klausbirken	Fine for me	13:59


       steve__	Short notice, but there will be an open LTSI workshop at LinuxCon	14:00
       steve__	14:00
       paulsherwood	great, tvm.	14:00
       paulsherwood	any other AOB from folks?	14:00
       gunnarx	Zzzz :)	14:00
       jeremiah	klausbirken: I'll come back to you about Franca testing, I'd like to know more and see if the tests I'm writing ought to be written with Franca.	14:00
       gunnarx	just kidding	14:00
       steve__	=) JTA test and longer term support appear to be on agenda	14:00
       klausbirken	jeremiah: Fine.	14:01
       * paulsherwood notes that he will email the list seeking people's feedback abou whether the irc format for this meeting has any future :)	14:01
       pavelkonopelko	steve__: how about LinuxCon Europe in October?	14:01
       gunnarx	paulsherwood, to answer, no bugs to go bugzilla	14:01
       steve__	Don't know. Ask them on LTSI list :)	14:01
       paulsherwood	gunnarx: ok, thanks. Joel_Replogle - what is the purpose of having the new public JIRA, please?	14:01
       gunnarx	for now.   I'm happy to jump to new infrastructure, including Wiki, when we know what we're doing.  I don't think the jump is ready.	14:01
       jeremiah	Well, Tim Bird is the chair of ELCE I believe and is on LTSI so . . .	14:02
       *** mdunford has quit IRC	14:02
       * paulsherwood notices the time, and proposes to bring the meeting to a close. folks would be welcome to continue the conversation here, of course	14:02
       steve__	yes and I can imagine many of the usual suspects will be at ELC	14:02
       gunnarx	And that goes for the Wiki too but I guess a meeting is planned to fill the spaces in an ordered manner.	14:03
       * klausbirken thinks that this kind of IRC meeting is much more exhaustive than conf calls	14:03
       paulsherwood	exhaustive, or exhausting? :)	14:03
       * gunnarx agrees	14:03
       klausbirken	exhausting - sorry	14:03
       Joel_Replogle	New JIRA has tight integration with Confluence.	14:03
       gunnarx	oops I misread.  I'd say both.	14:03
       paulsherwood	ok... well let's discuss it on the ml after folks have a chance to get their breath back :)	14:04
       gunnarx	Joel_Replogle, sure I agree. I think the new infra can be great.  I just want to avoid confusion in between.	14:04
       gunnarx	And don't stop contributing to the MediaWiki!  Content is better than no content :)	14:04
       paulsherwood	anyone object to me closing the 'official' meeting now?	14:04
       steve__	nope	14:04
       gunnarx	no	14:04
       pavelkonopelko	NOP	14:04
       Joel_Replogle	Gunnar - complete agreement!	14:04
       klausbirken	Yes both. I feel like the guy running behind a car - exhausted. :-)	14:04
       * paulsherwood thinks this was a very useful discussion, thanks to all	14:05
       paulsherwood	== Tools Team Meeting ends ==	14:05~>
  • No labels