Dealing with drunk BitBake

Bulb

Does anybody know where I'd get some antabus for BitBake? It seems to be drunk…

I am trying to build openjdk-8 from meta-java for poky from the pyro branch (yes, it is ancient; the OEM of the board does not have it tested with any newer).

The problem is that when it gets to build (do_compile) bsf-native, it fails with a bunch of errors in src/org/apache/bsf/engines/jacl/BSFCommand.java saying some imports don't exist. Apparently they should resolve from jacl-native, but when I go to the build/tmp/work/x86_64-linux/jacl-native-1.4.1-r0, where that should be built, the logs in temp indicate that it wasn't compiled and only the do_populate_sysroot_setscene ever ran.

However, the bsf_2.4.0.bb does say DEPENDS_virtclass-native = "jacl-native … and when I asked bitbake -g openjre-8 for dependency tree, the generated graph includes edges

"bsf-native.do_compile" -> "bsf-native.do_configure"
"bsf-native.do_configure" -> "bsf-native.do_prepare_recipe_sysroot"
"bsf-native.do_prepare_recipe_sysroot" -> "jacl-native.do_populate_sysroot"
"jacl-native.do_populate_sysroot" -> "jacl-native.do_install"
"jacl-native.do_install" -> "jacl-native.do_compile"

(the dot file is quite long, but all the above lines are copy-pasted from it)

So do_compile should have run for jacl-native before bsf-native. But it didn't. Does anybody around have idea how to make BitBake quit drinking?

Bulb

… I did clean out the tmp before doing it, but not the sstate-cache, so it might have thought it had a version in there—which would be OK if it actually worked, but it didn't. I'm trying again without cache.

brie

@Bulb said in Dealing with drunk BitBake:

Does anybody know where I'd get some antabus for BitBake? It seems to be drunk…

I am trying to build openjdk-8 from meta-java for poky from the pyro branch

The main word that I recognized in there was "drunk".

Tsaukpaetra

@Bulb said in Dealing with drunk BitBake:

drinking

It's been a few days, did the detox complete?