You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Dec 16, 2024. It is now read-only.
In buildbot_steps.yml, we only override the ANDROID_SDK variable for the ./mach package step, not the ./mach build step, so the current SDK symlink takes precedence
Both -cross1 and -cross2 still have the older r24.4.1 SDKs on disk, while all three have the new r25.2.3 SDK on disk.
-cross1 and -cross2 have the old SDK as current, while -cross3 has the new SDK as current.
Is it feasible that ./mach build --android --dev is failing due to getting the new r25.2.3 SDK on -cross3, while working OK with the r24.4.1 SDK on -cross1 and -cross2?
We've tried re-highstating servo-linux-cross3; jury is still out on the last highstate, but previous attempts have not fixed the problem.
I'm not well versed with Android, but I have a new idea about what might be going wrong:
/mach build --android --devstepANDROID_SDKvariable for the./mach packagestep, not the./mach buildstep, so thecurrentSDK symlink takes precedence-cross1and-cross2, only-cross3:-cross1and-cross2still have the olderr24.4.1SDKs on disk, while all three have the newr25.2.3SDK on disk.-cross1and-cross2have the old SDK ascurrent, while-cross3has the new SDK ascurrent.Is it feasible that
./mach build --android --devis failing due to getting the newr25.2.3SDK on-cross3, while working OK with ther24.4.1SDK on-cross1and-cross2?cc @larsbergstrom @MortimerGoro