sanjeev | Having trouble to build GDP from github repos on Ubuntu 16.04 | 00:05 |
---|---|---|
sanjeev | libpulse-dev is installed. | 00:06 |
*** johnnynines has quit IRC | 00:08 | |
*** sanjeev has quit IRC | 00:11 | |
*** jlrmagnus has left #automotive | 00:15 | |
*** mdurnev has joined #automotive | 01:22 | |
*** leon-anavi has joined #automotive | 06:51 | |
leon-anavi | morning | 06:58 |
*** ctbruce has joined #automotive | 07:22 | |
*** CTtpollard has joined #automotive | 07:22 | |
CTtpollard | morning all | 07:38 |
*** toscalix has joined #automotive | 07:59 | |
*** kooltux_ has joined #automotive | 08:03 | |
*** jonathanmaw has joined #automotive | 08:16 | |
leon-anavi | toscalix, thanks for updating GDP-206 | 08:27 |
leon-anavi | I hope that with joint efforts we can solve this issue. | 08:28 |
toscalix | we will | 08:35 |
toscalix | jonathanmaw: and CTtpollard are waiting for the touchscreen | 08:36 |
toscalix | I am planning of buying myself one | 08:36 |
leon-anavi | very good :) | 08:36 |
CTtpollard | gahh merge commits | 09:14 |
*** kooltux_ has quit IRC | 09:24 | |
*** perryl has left #automotive | 10:19 | |
*** gunnarx has joined #automotive | 10:31 | |
*** gunnarx has joined #automotive | 10:31 | |
CTtpollard | hi gunnarx | 10:37 |
gunnarx | hello | 10:37 |
gunnarx | jeremiah, you there? | 10:38 |
CTtpollard | gunnarx: another downside of multiple branches now that we've moved to github is pull request, specifically merge commits | 10:38 |
gunnarx | they are harder to merge? | 10:39 |
CTtpollard | gunnarx: well, easy to merge to the branch from which they create the pull request from | 10:40 |
gunnarx | yes | 10:40 |
gunnarx | Do you mean a pull request that happens to a merge commit? Not sure I follow | 10:40 |
CTtpollard | for instance here https://github.com/GENIVI/genivi-dev-platform/commits/qemux86-64 personally I'm not a fan or merge commits anyway, but getting the actual commit + merge commit onto all branches is not pretty | 10:41 |
CTtpollard | I can cherry pick one or the other | 10:41 |
CTtpollard | however both will bring in the actual change | 10:42 |
gunnarx | You mean the topmost commit on that URL or many of them? | 10:42 |
gunnarx | don't understand yet what I should look at... | 10:42 |
CTtpollard | the top commit is the merge commit, the commit before is the actual commit | 10:43 |
CTtpollard | you can add the pull request as a branch and merge to where ever you want, but we can't handle merges like that due to conflicts in conf files. hence why I'm trying to cherry pick | 10:44 |
CTtpollard | for something as simple as adding a licence file, this is annoying :) | 10:45 |
gunnarx | I'm not really sure what jeremiah-LICENSE-patch branch contained. If it had branched off qemux86 and added one commit, the merge should be just a fast forward. If other changes were between I suppose the jeremiah branch could be rebased first | 10:45 |
gunnarx | There's something broken in the whole setup... when there are conflicts on simple merges | 10:46 |
CTtpollard | gunnarx: the issue isn't with the qemu branch, it's getting it onto other branches cleanly | 10:46 |
gunnarx | and that's where there is something not right, IMHO | 10:46 |
CTtpollard | the 'patch' is fine to apply, merging however is not | 10:47 |
gunnarx | anyhow, it seems an extension of what you have already stated which is that merges between the branches are not trivial, although I am surprised they are not. | 10:47 |
CTtpollard | I'm not sure where the surprise comes from, there's conflicting files between branches | 10:48 |
gunnarx | conflicting files = the same part of the same file edited in two different ways | 10:48 |
gunnarx | that's a surprise actually. | 10:48 |
gunnarx | Actually maybe I am not that surprised, because we never really made an effort to set it up | 10:49 |
CTtpollard | local conf & bblayers conf will never match between branches | 10:49 |
gunnarx | they don't need to | 10:49 |
gunnarx | say you have a master branch with MACHINE=qemu | 10:49 |
gunnarx | then all other branches are modifications of that branch where MACHINE is set to something else. | 10:49 |
CTtpollard | + tons of other config | 10:49 |
gunnarx | Now that diff is known by the tool and happens earlier in the config. Merges from master to other branches should not conflict | 10:50 |
gunnarx | Between the other branches, possibly | 10:50 |
gunnarx | oops, happens earlier in the git history I meant to say | 10:50 |
CTtpollard | I was half thinking about taking the approach I made with having the multiple templates for bblayers & local confs that I did in the single branch poc, but across all branches | 10:51 |
CTtpollard | I think that's a possibility that may work, and would allow us to still branch per target | 10:51 |
gunnarx | I'm disputing you have issues with merging. I just feel it should be theoretically possible to set it up cleanly to minimize that. | 10:51 |
gunnarx | Not sure I understand this hybrid approach... | 10:52 |
gunnarx | Is the difference then that you have no shared conf files that can conflict? | 10:53 |
CTtpollard | yes, and then init.sh is set per branch to extract the one for the specific branch | 10:54 |
gunnarx | ok, and the difference compared to single branch is you can have submodules locked to different hashes | 10:54 |
CTtpollard | or not included submodules at all | 10:55 |
gunnarx | right, you don't include BSPs that are not used for example | 10:55 |
CTtpollard | the .gitmodules file can be a pain to merge sometimes as well | 10:55 |
CTtpollard | as we found with meta-rust | 10:55 |
CTtpollard | appending a new module to the end of the file is different, when one ends in $bsp1 and the other $bsp2 | 10:56 |
gunnarx | yeah... I'm trying to fully grasp that as well. If we started from fresh the master .gitmodules would be a subset of all the others... with the right setup I feel it *should* merge cleanly. We are not making conflicting edits for each change, just adding or subtracting. | 10:58 |
CTtpollard | trying to make things easier for everyone and everything is hard | 10:59 |
gunnarx | (I know it's tiring hearing me say that it "should" work :) | 10:59 |
CTtpollard | gunnarx: the problem is a contributer (I was using meta-rust as an example) adds a new submodule, and uses the qemu branch. This module gets appended to the end of the .modules file. However the board branches all have the .modules file += their bsp layer on the end | 11:01 |
gunnarx | Ah ok. Yes in this case I can see it happening | 11:01 |
CTtpollard | now I can hack this manually, but it's stuff like that I want to avoid | 11:02 |
gunnarx | Hmm, maybe some clever work around... You would have to get git to understand this is not a conflicting change but a new one. | 11:03 |
CTtpollard | I still think the potential to switch to the .conf templates across all branches is a possible idea | 11:04 |
gunnarx | Maybe if .gitmodules had some comment lines so git can find its way better. Some set of placeholders for new submodules to be added. | 11:04 |
CTtpollard | gunnarx: interesting | 11:04 |
gunnarx | If git figures it is a modification of an existing line (that exists on all branches) instead of two independent additions, I think it should merge cleanly. | 11:05 |
gunnarx | But I'm just thinking out loud here. | 11:05 |
gunnarx | I'm not saying you should do this, just brainstorming. But if .gitmodules was set up with a bunch of placehodlers like #module1 #module2 #module3 #module4 etc. then git would know that you are inserting something new that is not the same as a previous insert earlier in the history. Would be cool to try. | 11:06 |
gunnarx | Another thing do you understand what is going on with https://github.com/GENIVI/genivi-dev-platform/pull/1 ? | 11:07 |
CTtpollard | I can't tell if jeremiah meant the branch was unneeded, or the patch itself | 11:08 |
CTtpollard | i.e the local branch | 11:08 |
CTtpollard | tbh I just want to put the structure stuff to bed so we can concentrate on ivi10 | 11:11 |
gunnarx | he closed the PR | 11:12 |
gunnarx | I want the README fixed... it should not say "almost ready for public" and so on any longer. Jeremiah's PR was a good first quick fix. | 11:13 |
CTtpollard | sure | 11:13 |
CTtpollard | gunnarx: we've discussed as maintainers about github merge commits, do you have a view on them? | 11:14 |
gunnarx | Generally I would recommend maintainers don't use the web interface. If you can study and merge (and possibly clean up) PRs locally and push them instead. | 11:17 |
gunnarx | You mean merge commits that Github creates by using the web interface to merge right? | 11:17 |
gunnarx | ^^ merge locally, then push | 11:17 |
gunnarx | does that address the question or was it a different one? | 11:19 |
CTtpollard | it addresses it, personally I think they just add noise (and currently make things harder) | 11:19 |
CTtpollard | but for bigger sets (e.g new functionality) it can be nice to have it to link back to the pr | 11:20 |
CTtpollard | if viewing in the web ui, git log not so much :) | 11:21 |
gunnarx | ah ok I get the issue now. | 11:22 |
gunnarx | Yeah OK, if you cherry pick and such I suppose it is not as clear. | 11:22 |
CTtpollard | if can cherry pick the pr merge commit onto another branch, or the actual commit (in the case of this licence patch) but I can't seem to fangle cherry-pick to pull both onto another branch | 11:24 |
CTtpollard | just aiming for consistency | 11:24 |
gunnarx | I would say as long as there is a connection to contributions made by others that should be enough. I don't think PRs by themselves are all that important - they are needed for review but once you have merged the actual commits matter more. Just my feeling | 11:25 |
gunnarx | I suppose github keeps the web part of the PR regardless, i.e. you can see any +1 and comments | 11:26 |
gunnarx | that's important of course, but the git history including the tree of PR branches, for me less so. | 11:26 |
CTtpollard | yeh it does keep the pr request, just losing out on the pr link in the history | 11:27 |
CTtpollard | I think we agree here | 11:27 |
CTtpollard | ughh pr request | 11:28 |
CTtpollard | lunch break I think... | 11:28 |
gunnarx | send me a request for pull request request ... | 11:29 |
CTtpollard | it's like PIN number | 11:29 |
gunnarx | I like PIN number | 11:30 |
CTtpollard | what about ATM machine? | 11:30 |
gunnarx | that too | 11:31 |
gunnarx | Speaking of trees, doesn't github put together something useful anyhow, i.e. PRs come from other people's forks and Github shows where those forks go into and out of the history since it can find common commits. It's up to the submitter to keep their local branches or not, I think. | 11:32 |
gunnarx | So as long as you don't rewrite the actual commit (but that's a choice you take locally) it should piece it together somewhat nicely I expect | 11:32 |
CTtpollard | not 100% sure tbh | 11:33 |
CTtpollard | it handles submodule commits nicely if they're also hosted on github | 11:35 |
gunnarx | I'm pretty sure it will link up identical hashes in the graphs, wherever they are found. | 11:35 |
gunnarx | yes | 11:36 |
*** gunnarx has quit IRC | 11:49 | |
*** praneeth has quit IRC | 11:57 | |
*** praneeth has joined #automotive | 11:58 | |
*** GuyS has joined #automotive | 12:08 | |
*** GuyS has quit IRC | 12:13 | |
*** gunnarx has joined #automotive | 12:27 | |
*** gunnarx has joined #automotive | 12:27 | |
*** GSgns has joined #automotive | 12:41 | |
*** GSgns has quit IRC | 12:45 | |
*** GSgns has joined #automotive | 12:46 | |
*** waltminer has joined #automotive | 12:52 | |
CTtpollard | jeremiah: you around? :) | 13:39 |
gunnarx | whoa, you have to write access CTtpollard ? :) | 13:58 |
gunnarx | no write access | 13:59 |
CTtpollard | seems that way | 13:59 |
gunnarx | I can fix it | 13:59 |
gunnarx | if Jeremiah's not around | 14:00 |
gunnarx | try onw | 14:02 |
gunnarx | now | 14:02 |
CTtpollard | gunnarx: your magical powers fixed it | 14:02 |
CTtpollard | gunnarx: thanks | 14:04 |
gunnarx | you are welcome | 14:05 |
*** AlisonChaiken has quit IRC | 14:15 | |
jeremiah | CTtpollard: Hiya | 14:32 |
jeremiah | Sorry I was out of the loop | 14:32 |
jeremiah | Thanks for fixing gunnarx | 14:32 |
gunnarx | np | 14:32 |
jeremiah | Just so everyone knows, I've given magic write powers to those in the GDP Crew team | 14:33 |
gunnarx | scary | 14:33 |
*** tjamison has joined #automotive | 14:33 | |
jeremiah | But they can only write to the genivi-dev-platform. :-) | 14:33 |
CTtpollard | :) | 14:33 |
jeremiah | So hopefully not too wicked scary | 14:33 |
gunnarx | you never know | 14:33 |
jeremiah | So far its only been me who's been randomly commiting to the wrong branch :P | 14:34 |
CTtpollard | jeremiah: your licence patch is on all targets now, however the qemu branch is the only one with the github PR merge commit for it | 14:46 |
gunnarx | CTtpollard, jeremiah opened up the README also. You could merge it (and remove the extra branch after)... | 14:47 |
gunnarx | opened the README pull request | 14:48 |
gunnarx | I was trying to look at if this could be cleanly merged but have other things that need to be done. For the moment if you need to cherry-pick, I imagine you could continue doing that to avoid the merge hassles that are there now | 14:49 |
*** waltminer has quit IRC | 14:49 | |
*** waltminer has joined #automotive | 14:50 | |
*** astrophys has joined #automotive | 14:52 | |
CTtpollard | heh, you can add .patch to the end of PR url to get a plaintext patch https://patch-diff.githubusercontent.com/raw/GENIVI/genivi-dev-platform/pull/1.patch | 14:52 |
*** mdurnev has quit IRC | 14:55 | |
*** jlrmagnus has joined #automotive | 14:58 | |
*** mvick has joined #automotive | 15:01 | |
toscalix | jeremiah: since tomorrow the news about GDP on github will be out in the newsletter and Minnow release announcement, I was thinking it could be good to mention in the How to Contribute page Giyhub | 15:04 |
toscalix | GitHub | 15:05 |
*** gunnarx has quit IRC | 15:06 | |
toscalix | jeremiah: wouldn't it be better to provide rights to push opnl maintainers? | 15:13 |
toscalix | sorry | 15:13 |
toscalix | jeremiah: wouldn't it be better to provide rights to push only to maintainers? | 15:13 |
toscalix | GDP crew is meant to be everybody involved in GDP | 15:14 |
toscalix | added admin rights to maintainers in genivi-dev-platformmaintainers | 15:16 |
toscalix | removed write rights to genivi crew to in meta-genivi-demo waiting for jeremiah confimation | 15:16 |
*** jlrmagnus has quit IRC | 15:20 | |
*** brlogger has joined #automotive | 15:28 | |
*** astrophys has quit IRC | 15:40 | |
*** ToxicGumbo-work has quit IRC | 15:41 | |
*** ToxicGumbo-work has joined #automotive | 15:47 | |
*** RzR has quit IRC | 16:01 | |
*** RzR has joined #automotive | 16:01 | |
*** ctbruce has quit IRC | 16:24 | |
toscalix | So we are ready to release GDP-ivi9 port to Intel Minnowboard tomorrow | 16:25 |
RzR | toscalix, on master branch ? | 16:33 |
RzR | toscalix, I will test it | 16:33 |
toscalix | yes | 16:34 |
toscalix | great | 16:34 |
RzR | I can try tonight if it helps | 16:34 |
toscalix | it does | 16:34 |
RzR | ok I can build master | 16:34 |
toscalix | https://github.com/GENIVI/meta-genivi-dev | 16:35 |
RzR | or qemu branch | 16:35 |
toscalix | yes | 16:35 |
RzR | ok | 16:35 |
toscalix | check this: https://at.projects.genivi.org/wiki/pages/viewpage.action?pageId=11569311 | 16:35 |
toscalix | we will release images though | 16:35 |
*** jonathanmaw has quit IRC | 16:51 | |
*** leon-anavi has quit IRC | 16:58 | |
*** ToxicGumbo-work has quit IRC | 16:59 | |
*** ToxicGumbo-work has joined #automotive | 17:03 | |
*** toscalix has quit IRC | 17:06 | |
*** waltminer has quit IRC | 17:43 | |
*** jlrmagnus has joined #automotive | 17:44 | |
*** jlrmagnus has quit IRC | 18:02 | |
*** AlisonChaiken has joined #automotive | 18:04 | |
*** mvick has quit IRC | 18:06 | |
*** jlrmagnus has joined #automotive | 18:18 | |
*** waltminer has joined #automotive | 18:36 | |
*** jlrmagnus has quit IRC | 18:47 | |
*** jlrmagnus has joined #automotive | 19:00 | |
*** waltminer has quit IRC | 19:47 | |
*** kooltux_ has joined #automotive | 19:49 | |
*** AlisonChaiken has quit IRC | 20:01 | |
*** jlrmagnus has quit IRC | 20:02 | |
*** AlisonChaiken has joined #automotive | 20:16 | |
*** jlrmagnus has joined #automotive | 20:18 | |
*** Sisco has joined #automotive | 20:52 | |
*** Sisco_ has quit IRC | 20:53 | |
*** AlisonChaiken has quit IRC | 21:01 | |
*** jlrmagnus has quit IRC | 21:17 | |
*** AlisonChaiken has joined #automotive | 21:19 | |
*** Saint_Isidore has joined #automotive | 21:37 | |
*** Saint_Is_ has quit IRC | 21:38 | |
*** chbae has joined #automotive | 21:47 | |
*** jlrmagnus has joined #automotive | 22:02 | |
*** chbae has quit IRC | 22:09 | |
*** sanjeev has joined #automotive | 22:13 | |
*** kooltux_ has quit IRC | 22:22 | |
*** tjamison has left #automotive | 22:57 | |
*** waltminer has joined #automotive | 23:12 | |
*** jlrmagnus has quit IRC | 23:25 | |
*** Saint_Isidore has quit IRC | 23:26 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!