On Tue, 2022-01-11 at 12:10 +0100, Alexander Kanavin wrote: > On Tue, 11 Jan 2022 at 12:08, Richard Purdie > <[email protected]> wrote: > > The data above is inserted into /etc/default/connman in the connman package > > itself. That package is installed into deploy/XXX/<tunearch>/connman.YYY. > > That > > is not the machine specific package directory. > > > > Or in different words, if I build connmand for a qemu machine, then build > > connman for a machine with the same tune that isn't qemu, command rebuilds > > but > > installs the package to the same location. This is not allowed. > > > > Machine specific data (in this case qemu specific) shouldn't be in general > > packages. This goes against yocto-check-layer and YP-Compatibility. > > > > > Right, I can fix this, but is there a testing gap then? This prompted no > warnings on either local builds or in a-full.
The autobuilder is not definitive as it does not have 100% test coverage much as I wish it were. Yes, there is clearly a gap in testing here :( We don't really have a machine with an identical tune to a qemu target that isn't qemu which we run sstate sigs tests against. Cheers, Richard
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#160395): https://lists.openembedded.org/g/openembedded-core/message/160395 Mute This Topic: https://lists.openembedded.org/mt/88311448/21656 Group Owner: [email protected] Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
