Hello Richard and Alex, Richard: We tried to revert the commits which you mentioned and it didn't work.
Alex: Yes, is fully reproducible on latest master. bitbake core-image-minimal -c populate_sdk_ext eSDK installed via: poky-glibc-x86_64-core-image-minimal-cortexa15t2hf- neon-qemuarm-toolchain-ext-4.1+snapshot.sh . environment-setup-cortexa15t2hf-neon-poky-linux-gnueabi devtool modify busybox Sstate summary: Wanted 14 Local 0 Mirrors 0 Missed 14 Current 6 (0% match, 30% complete) So it started a compilation of missing components. We are assuming, that eSDK will include all build deps for all components in the image and not just a deps for image itself. Regards, Andrej On Mon, 2022-06-27 at 14:35 +0100, Richard Purdie wrote: > On Mon, 2022-06-27 at 12:32 +0000, Valek, Andrej wrote: > > I have a question related to eSDK dependencies. We're using the > > dunfell > > branch were everything related to this eSDK topic works fine. Now > > we're > > in the transition phase to new LTS branch, where were we found one > > big > > difference between eSDKs. > > > > The old variant (dunfell) includes all application build > > dependencies, > > but the new variant (kirkstone/master) doesn't. Means if I > > installed > > the eSDK and used "devtool modify my-app" (application installed on > > the > > image) it works without any additional build deps recompilation. > > But now, if I do the same on the newer version it always recompile > > all > > build deps. > > I was already looked, what could be changed, but I didn't find so > > far > > something suspicious. So the question is, what has been changed, > > and > > how to bring the old variant back? > > Guessing is hard. Since you're asking me to guess: > > https://git.yoctoproject.org/poky/commit/?id=568f62214bca3ac6d35eef8d9f4562596fb4c9ab > > which was partially reverted here: > > https://git.yoctoproject.org/poky/commit/?id=f22e1fbdf7bed111e080d176fe5a39c5139308ed > > maybe? It could be something else. It wasn't a specific change to > remove such dependencies but I suspect it could have happened as an > unforeseen side effect of something else. > > You may need to come up with a simple test case and then bisect > between > the two releases to see which change it was. Once we understand what > change caused it, working out a solution would be easier, it is > premature to even try without knowing the cause. > > Cheers, > > Richard > > > >
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#167333): https://lists.openembedded.org/g/openembedded-core/message/167333 Mute This Topic: https://lists.openembedded.org/mt/92019337/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
