Skip to end of metadata
Go to start of metadata
paulsherwood== GENIVI Tools Team Meeting ==13:00
paulsherwoodhi all13:00
gunnarxAh yes, I should not be searching the Wiki anymore.13:00
gunnarxhi13:00
*** pkonopelko has joined #automotive13:00
paulsherwoodgunnarx: no, i was intending to tidy the logs into minutes, and then my week happened instead13:00
gunnarxok13:01
paulsherwoodso i wonder if in general the irclogs are seen as 'good enough' or not?13:01
paulsherwoodanyways...13:01
paulsherwooddoes anyone have new items for the agenda?13:02
*** fredcadete has joined #automotive13:02
paulsherwoodok, well from last time....13:03
paulsherwood== Wikis ==13:03
paulsherwoodany news or progress on this?13:03
gunnarxjreplogle online?13:03
* paulsherwood tried to migrate, and discoveded it was going to be labour intensive13:04
gunnarxdoesn't look like it13:04
gunnarxmigrate what?13:04
*** stevel_ has joined #automotive13:04
paulsherwoodcontent from wiki.projects.genivi.org13:04
paulsherwoodjoel summarised the process as follows: http://paste.baserock.org/usatoyuleg13:05
gunnarxyes, I was just trying to understand what content, to understand what types of content is most problematic.  maybe that detail is not available right now13:05
paulsherwoodwhich i had to summarise as ' ‘cut and paste’ and frab about'13:05
gunnarxYes, the proposal there is to copy "rich text"13:05
paulsherwoodi admit i was hoping someone could just 'run a script' and it would all be done13:06
paulsherwoodbut software is rarely like that13:06
paulsherwoodany others have comments on wiki stuff? did anyone else try migrating content?13:07
gunnarxsomeone needs to write the script typically.13:07
gunnarxI fail to see a major problem with people migrating their content, it can't be more than a few minutes per major page right?  Is copy paste not working well?13:08
gunnarxGranted I haven't tried.  So that's why I'm asking, are there serious problems or just a matter of doing it?13:08
paulsherwoodit's just a matter of doing it. cost of my time, futzing around with wikis versus all the other things people want me to do :)13:09
paulsherwoodanyways, shall we move onto other topics?13:09
gunnarxyes, ok for me13:10
paulsherwoodgah... i notice i didn't actually send out a reminder for this meeting13:10
paulsherwood== CommonAPI ==13:10
paulsherwoodi need help from others to advance this topic...13:11
paulsherwoodand i fear by not notifying them, they may not be here :/13:11
gunnarxI don't know what the open question is either.13:11
*** waltminer has joined #automotive13:11
paulsherwoodi don't know that there is an open question, just that this is a sync point for discussion on it?13:12
paulsherwoods/it/CommonAPI, Franca/13:12
gunnarxwell, engaging the brain, I think last week I asked if the TT can evaluate the ability to efficiently use the Javatools as part of the build13:13
pkonopelkofwiw, my poc of common api for c will be published this week13:13
paulsherwoodpkonopelko: cool!13:13
gunnarxpkonopelko:  Cheers and congratulations13:13
pkonopelkonot that it can do much for you, but it is  a start13:13
pkonopelkowell, ... not yet13:14
gunnarxwrt Java based tools it is in the trello card https://trello.com/c/gpAMIGZZ/25-normal-build-integration-for-commonapi.  I suppose no progress?13:14
paulsherwoodgunnarx: i was just checking that mysefl13:14
*** CTtpollard has joined #automotive13:15
paulsherwoodwe would need volunteer(s)13:15
gunnarxpaulsherwood:  do you have a list of assigned actions.  is there something driving work forward (generally).  Trello cards seem more to indicate who is generally "interested"13:15
pkonopelkounfortunately, there is already a large code base in java / xtend for franca idl13:15
pkonopelkoreplacing that is a sizable effort13:16
paulsherwoodgunnarx: i do not have such a list. experience in other projects makes me wary of trying to 'assign' things in an open source environment, but i may be beign overcautious13:16
gunnarxpkonopelko:  agree and the argument was made also last week.  Therefore, I asked if the TT could confirm/deny the usability of those tools.13:16
pkonopelkoaccordingly to klaus birken, maintaining the tools in also more efficient with the current toolset13:16
gunnarxin our typical builds (baseline, GDP)13:16
gunnarxpaulsherwood:  OK, what happens typically if you assign things.  does something blow up?  ;)13:17
paulsherwoodgunnarx: iirc there was discussion in BIT some months ago and both Baserock and Yocto maintainers flagged this as 'a lot of work' (but i may be mistaken)13:18
stevel_+113:18
paulsherwoodgunnarx: do we think there is enough interest/market/value in causing this to be considered as a sponsored effort?13:19
gunnarxYeah it's not a new discussion.  But if the argument is that the amount of work to move away from the current implementations is bigger, I fall back on the request again.13:19
pkonopelkomaybe the most productive way is to put common api and baseline developers together into a sort of workshop13:19
gunnarxI'd like to see a documented evaluation.  Is there more than the 4 words "a lot of work"13:19
paulsherwoodpkonopelko: actually that sounds like a great idea13:19
pkonopelkoare there any upcoming f2f that would be suitable?13:20
paulsherwoodwell, Seoul AMM?13:20
pkonopelkoAMM in seoul is not well attended by commonapi developers, so I've heard13:20
gunnarxAMM13:20
paulsherwoodack.13:20
gunnarxtrue13:20
pkonopelkomaybe upcoming si eg f2f in france?13:20
paulsherwoodwell, we can flag the possibility of such a workshop13:20
gunnarxbut I don't know how much is unclear.  It's a bunch of Java programs.  Can they be run or not?13:21
paulsherwoodlol.13:21
gunnarxdoes the discussion need the franca developers... is what I'm saying13:21
gunnarxs/discussion/evaluation or test/13:21
stevel_gunnararx: on BIT side no. Firm push back from both maintainers, i.e. hard to do in either build system, partly because we somewhat close to a milestone.13:21
pkonopelkoit goes like that: commonapi devs: there are tools for that; baseline devs: they don't work13:21
pkonopelkoso somebody is missing something13:22
stevel_but no follow on detail on actually breaking out steps13:22
gunnarxi see13:22
gunnarxAny chance we could have a champion who digs into this a little on the mailing lists.  E.g. do the commonapi developers have a working setup.... in yocto...?13:23
gunnarxcan they share?  can they explain more?13:23
paulsherwoodgunnarx: would you mind actioning that yourself? you know the issues as well as anyone?13:24
gunnarxyes i mind, that's why I ask13:24
gunnarx...13:24
pkonopelkopersonally, I plan to do this sooner or later, just not sure about the timeline13:24
gunnarxI just feel I drive many topics in the organization already13:25
paulsherwoodgunnarx: understood13:25
gunnarxthanks.  that's what I like to hear :)13:25
gunnarxthanks to pkonopelko that is13:25
pkonopelkoI will have to figure this out for commonapi/c next13:25
gunnarxyes13:25
paulsherwoodpkonopelko: could you take the action?13:25
pkonopelkothis will likely happen as i start to put the first version of the generator together13:25
pkonopelkowhat is exactly the action? that on trello?13:26
paulsherwoodwrite to the ml13:26
stevel_can ask baseline mainteners side, but it would be low drive for the next few weeks..13:26
pkonopelkopaulsherwood: ok, once I will get to this I will do the digging (ml, etc.)13:26
paulsherwoodpkonopelko: tvm13:26
gunnarxafaik the tools compile with javac and run with java from openjdk, openjdk is open source, javac is a compiler, as is gcc which is compiled from source.  Has anyone really tried or is it just talk that this is "hard"...13:26
pkonopelkothere is a prepackaged binary available with commonapi / c++ releases; in my trial it did not launch beacause of missing dependencies13:27
stevel_main recall from maintainers that it would involve a lot of integration work13:27
pkonopelko...I hope that with some digging all missing pieces can be identified13:28
paulsherwoodon the baserock side, we did try integrating java, and found ourselves cursing a lot13:28
gunnarxpkonopelko:  tell the genivi-ipc@   We need to raise issues and remind about them.13:28
paulsherwoodok, shall we move on?13:28
gunnarxok13:28
paulsherwood== CIAT ==13:28
gunnarxoh here we go... :)13:29
paulsherwoodthere has been some ml discussion on this. i feel like i'm trying to push water uphil13:29
gunnarxhow deep are we going on today's meeting?13:29
paulsherwoodhow deep would you like to go?13:30
stevel_paulsherwood: do you propose changing all paths in all src to point elsewhere?13:30
gunnarxpaulsherwood:  I don't understand.  Is it the particular details that you want people to agree on.?13:30
gunnarxNo one is pushing back on the project as a whole13:30
gunnarxit can't all be uphill13:31
gunnarxstevel_ ++   paulsherwood, you understand the question?13:32
paulsherwoodstevel_: i believe from discussions here and elsewhere that yocto can be convinced to use a mirror, so i don't know that replacing all source paths is required13:32
stevel_agreed, but that is the conclusion I draw from your comments13:33
gunnarxpaulsherwood:  no but if the recipe is requesting a tarball you can't provide a git in the mirror (is my assumption)13:33
paulsherwoodyou'd still want to mirror it, gunnarx, i think13:33
gunnarxyes13:34
gunnarxyes?13:34
paulsherwoodand putting it into git would be better than leaving it as a tarball imo13:34
*** mdunford has quit IRC13:34
gunnarxdidn't get the point13:34
paulsherwoodbut this is minor detail13:34
stevel_which possible results in a recipe change..13:34
gunnarxthe question that is opened up is of whether recipes need to change all over the place or not.  if it means significant changes you are forking/maintaining layers13:34
stevel_yep13:34
gunnarxlayers that we should really be reusing as-is, probably13:35
stevel_yocto mirror is one for one what you would see in a local download folder13:35
gunnarxthanks stevel_ for that important fact.  we need facts -> conclusions13:35
* gunnarx runs to build server to check13:36
stevel_I can switch between true upstream and yocto project mirror of poky meta downloads in a few lines of config in my local.conf. No recipe or inc changes required.13:36
paulsherwoodok13:37
stevel_Carrying patches is a seperate but related question13:38
gunnarxquick check looks like about 50/50 % tarballs vs git in GDP13:38
paulsherwoodgunnarx: no other types?13:39
gunnarxignored :D13:39
stevel_svn13:39
gunnarxI bet a few percent svn, hg etc.13:39
stevel_navit is svn for example13:39
paulsherwoodi fear i'm not in the right frame of mind to advance this discussion today13:40
stevel_what I don't know is how well the concept of a YP mirror fits to the baserock build concept, i.e. whether a single mirror could cover both well13:41
gunnarxthat's ok, but we have some shared understanding of the issues at least then?13:41
gunnarxor a baserock mirror fitting yocto ;)13:41
stevel_yep13:42
gunnarxpaulsherwood:  we can move on if you want but I suspect the open concerns have been raised and understood, do you agree?13:42
stevel_if it means changing all the recipes, badly =)13:42
* paulsherwood doesn't know what to say13:43
paulsherwoodi suggest we move on. i'll collect my thoughts and write to the ml13:43
gunnarxyes, that's a good idea.  We were just looking that you confirm understanding the issues?  Not necessarily the severity of them, we can debate that.13:44
gunnarxbut the nature of them.13:44
stevel_just a suggestion paul, but if there is an existing open baserock mirror you can point at we can consider how well it would fit yocto13:44
paulsherwoodgit.baserock.org, you mean?13:45
stevel_upstream source mirror13:45
gunnarxstevel_:  git.baserock.org includes all the sources13:46
gunnarxyou need to click on the [n] links to move forward13:46
gunnarxthat's my understanding at least13:47
paulsherwoodwell, it has all of the sources for gdp, for example13:47
stevel_which git does Lorry dump into?13:47
paulsherwoodall of them13:47
paulsherwoodevery day13:47
paulsherwoodanyway, let's move on please.13:48
paulsherwood== AOB ==13:48
*** mdunford has joined #automotive13:48
paulsherwoodare folks still happy with irc, or prefer to return to conf calls?13:48
gunnarxso so.  both has its merits13:49
* paulsherwood prefers irc, but admits that it's easier to muddle through on calls without preparing :)13:49
gunnarxboth have their merits, more correctly13:49
CTtpollardcould the calls be on a public platform?13:50
gunnarxif you pay for it, sure :)13:50
paulsherwoodyes, they were.13:50
paulsherwood(genivi was paying)13:50
paulsherwoodbut the proposal to do them here was to facilitate extra contributions and provide written record13:51
paulsherwoodthe latter is being achieved, but i've not seen a lot of new contributors to the discussion in public13:51
paulsherwood(some folks have commented +vely in private, though)13:51
paulsherwoodany other AOB from folks?13:52
gunnarxfits my expectation tbh13:52
pkonopelkohaving something in public does not attract contributors by itself...13:52
stevel_+1 on differing merits. Don't feel the calls are any shorter tbh unless only topics are yes/no13:52
paulsherwoodstevel_: ack. however, you could be on a call *as well* as this meeting, and no-one would notice13:53
gunnarxI think we may set up a call once in a while.  This being the only IRC meeting means it serves a purpose as something different.13:53
paulsherwoodgunnarx: ack. would help us to re-establish that we are humans, not just bots :)13:53
fredcadeteI like seeing these meetings even though I don't participate13:54
gunnarxso a vote to  keep the majority of TT here for now, with maybe monthly exceptions13:54
fredcadetenot that it should change your minds, though13:54
paulsherwoodfredcadete: thanks for the feedback!13:54
gunnarxyou should participate13:54
stevel_posting on this thread is participating!13:54
paulsherwoodgunnarx: makes sense13:54
*** tpollard_ has joined #automotive13:54
paulsherwoodok, any more AOB before i let you all dive into the rest of your day?13:55
stevel_nope13:55
*** Egy has quit IRC13:55
gunnarxnone from me13:55
paulsherwood513:55
paulsherwood413:55
paulsherwood313:55
paulsherwood213:55
paulsherwood113:55
paulsherwood== Tools Team Meeting Ends ==
  • No labels