Think we have an issue in here where isLocal == true and s.pkg == nil; we never activate the target and it never gets built. This happens in the attached test repo when going through //a because of the double-subinclude; it works okay if the subinclude from//c gets there first (because then we are specifically parsing for that target which gets handled elsewhere).
I don't really know how to fix this without a heavier rewrite of things. I've attempted to ban it (I don't think it can ever work reliably at present so it shouldn't exist in the wild) although this is turning out to be too strict (it's firing on any local subinclude, which is okay as long as it didn't originate from the same package).
Bit unclear how we should go forward here.
Think we have an issue in here where
isLocal == trueands.pkg == nil; we never activate the target and it never gets built. This happens in the attached test repo when going through//abecause of the double-subinclude; it works okay if the subinclude from//cgets there first (because then we are specifically parsing for that target which gets handled elsewhere).I don't really know how to fix this without a heavier rewrite of things. I've attempted to ban it (I don't think it can ever work reliably at present so it shouldn't exist in the wild) although this is turning out to be too strict (it's firing on any local subinclude, which is okay as long as it didn't originate from the same package). Bit unclear how we should go forward here.