Conversation
@sashko the build fails because there is no qemu 3.x in our builds. See |
@frznlogic same story as with #80: since it's |
Build checks:
|
One build check seems to have failed on some technicality, so I triggered a rerun. It will again fail on the #81 issue of course, unless something else is noticed instead. But please force-push an update to this to trigger the other build checks that have not run at all yet. |
#81 is now merged so we should rebase this and get a new build check. Warning @sashko, maintainers (and I) can force-push the source branch of any PR. This is a GitHub feature and it enables us to perform rebases and fixes, and we often do force push the PRs! Let me know if this is OK, or you can yourself rebase on the latest master, and push it again. |
@gunnarx you guys know your project better and I have no objections to you modifying my PRs. |
e3f38af
to
431412c
Compare
Cool, looks like you did the rebase |
Fails on ecryptfs-utils I looked into it a bit. Not totally sure why ecryptfs-utils is included in meta-ivi.
^^ @sashko |
--> If you want, you can add a patch to drop ecryptfs-utils recipe and remove it from the package group. (Make sure the commit clearly advises the reader to add the meta-security layer to find the recipe there.) |
@gunnarx yeah, my findings were similar: it is simply not adapted to the latest OpenSSL; I'll delete the recipe then, thanks. |
9ba1da4
to
c251bc9
Compare
@gunnarx I'm getting the following error. Do you know how I could fix it?
|
Hmm makes me wonder if we should warn about that in the readme. Suggest that PRs are created on unique branch for example. On one level master is just another branch but I can see how someone might think WTH if they suddenly have changes to their master. |
Yes sure - feel free if you wish. Actually this should be part of a "contribution guidelines" review, perhaps. I would add however:
|
@sashko I suppose it is as written, that a Java runtime needs to be installed on the build machine, for the code generators to run. (I think Java is not built as a -native dependency inside Yocto but it must be on the build machine) |
or does it want Oracle JDK? |
OpenJDK should be fine. Possibly it's a version issue since version 10 is very new. But the error actually states that it is looking in the sysroot-native area: | /home/oleksandr/projects/genivi/build/tmp/work/i586-poky-linux/audiomanagerplugins/7.6-r0/recipe-sysroot-native/usr/share/capicxx-core-native-3.1.12/jre/bin/java I'm not sure what the issue is - might be best to ask audiomanager or capicxx-core project for this... EDIT: For what it's worth, our build agents have a native installation (i.e. /usr/bin/java) of Java 8: openjdk version "1.8.0_171" Again I don't know if that's the one that is used. |
Yes I think that would be a good idea. I was just thinking somewhat aloud and its a small point that certainly could wait to a wider review. |
Hmm, some small conflict because of the most recent merges I made... Ugh, rebase conflict now. Let me fix and redo this. |
OK I nuked the ugly merge commit that GitHub created and rebased on latest master again instead. So you're all up to date on top of origin/master. As noted earlier in the thread, be aware that sashko/master is now rewritten (compared to your local copy master branch) |
I was wondering if it was further changes to the per recipe sysroot introduced in YP 2.3, but unfortunately the upgrade guide for YP 2.6 in the reference manual is not published yet. If not then we would expect it to be failing before 2.6/Thud. There is no explicit host runtime build dep in the AM Plugin recipe in meta-ivi for that I could see. Maybe the capicxx deps are supposed to cover it. I note that the recipe is doing some java path work in a do_configure_prepend(). |
@gunnarx it wouldn't be enough to make it build: there're few more issues with other GENIVI packages; I have made few pull request with fixes; it would be more helpful if you've made them reviewed instead. Thanks. |
@sashko If you let me know what other PRs you mean then I'll try to get contact with the maintainers if they are not reviewing. I saw the DLT proposal and will follow up on that, but I guess that's at the moment not holding up meta-ivi, thanks to the patch you added above. I see nothing more open on meta-ivi except the text-file change? As for it not being enough, we still have to deal with each issue, one at a time until it works... |
e67dac2
to
5f5add3
Compare
weston build failure that popped up now:
So as far as I can tell the tip of poky (oe-core) is failing for qemux86* targets (unless there's some conf setting that we're supposed to have but have missed?) To avoid it I think it might work to roll back poky to before EDIT: Just testing it now on a temporary PR |
Build layer against Yocto "Thud" 2.6. Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
security_flags.inc is already included in poky.conf. Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Removed recipe from the layer as it is already maintained by the meta-security. Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
FTP mirrors are not supported by Yocto anymore. Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Comparing persistence_client_library tag 1.1.0 to this one (1.2.0), there have been *lots* of changes introduced. The recipe has however updated its SRCREV to track master during this time, so the shortlog of the PCL changes corresponding to the SRCREV change *in this commit* is only: 797e314 Merge pull request GENIVI#15 from sashko/master bc4b12f20 Fix compilation with GCC 8 In any case the component has a new tag, and the recipe name therefore now also changed to PV=1.2.0 Signed-off-by: Gunnar Andersson <gandersson@genivi.org>
Comparing persistence_common_object tag 1.0.3 to this one (1.1.0), there have been a few more changes introduced. The recipe has however updated its SRCREV to track master during this time, so the shortlog of the PCO changes corresponding to the SRCREV change *in this commit* is: 6e827a9 Fix gdbus-codegen invocation 2ee245d Merge pull request GENIVI#5 from mdanilov/master 4b6b1ce Reverted: added dbus config file (there was commit to remove it due to conflict with pcl repo) 857df7a Change README to symlink f5b3b67 Fixed build, added missed dbus config file 29ec697 Merge pull request GENIVI#2 from GordanM/topic/deprecated d9235fa Merge pull request GENIVI#1 from GordanM/topic/fix-README 98a32c6 README.md: Update README daa21fe Merge pull request GENIVI#3 from GordanM/topic/remove-dbus-conf 103be9f dbus_config: Remove dbus configuration file fe82962 pers_ipc_dbus: remove deprecated glib init eeac12c Added sqlite backend; use --with-database=sqlite in configure step In any case the component has a new tag, and the recipe name therefore now also changed to PV=1.1.0 Signed-off-by: Gunnar Andersson <gandersson@genivi.org>
(A.k.a. Persistence Administration Service (PAS), but the bb component name is persistence-administrator for some historical reason) It was updated from version 1.0.9, but for some reason the repository is not tagged with that version, although the commmit message itself specifies the version. There is now a tag created for 1.0.10 Short log of changes - the gdbus-codegen update is the important one ebfb5dd README: fix headline and minor cleanups 49b7936 Makefile: fix gdbus-codegen invocation Signed-off-by: Gunnar Andersson <gandersson@genivi.org>
Specifically this includes: 5adefcdd fix gdbus-codegen invocation Note that 1.3.6 was skipped due to a bug Signed-off-by: Gunnar Andersson <gandersson@genivi.org>
Yes, let's do it. |
Woot! It's finally mergable :) |
I pushed a 1.0.10 tag for persistence-administrator, so recipe is now also correctly named. |
A total of 48 comments preparing this PR :-) Thanks for your work Oleksandr! |
Signed-off-by: Oleksandr Kravchuk oleksandr.kravchuk@pelagicore.com