On 4/3/20 9:47 PM, Nicolas Boulenguez wrote: > Package: src:gcc-9 > Followup-For: Bug #954681 > Control: reopen -1 > > Hello. > > The initial symptom is cured, but the fix lets > debian/patches/ada-lib-info-source-date-epoch.diff modify > gcc/ada/libgnat/s-os_lib.ad[bs]. > > This invalidates the checksums embedded in the .ali files of > libgnat-9, causing a failure in all Ada compilations involving > packaged dependencies. Especially, all run time tests for packaged > libraries fail. > > Ludovic suggests to revert the fix (commit > c1d276b0e2cb60140389aa4c23ac3addb7a77af3 )in the gcc-9 branch. > > I suggest to cherry-pick the commit in the gcc-10 branch, so that it > is part of the future gnat-10 transition. > > Matthias, do you agree with these two actions? > > Then, we will search for a less intrusive work-around for this bug in > gcc-9.
soounds fine. Please note that I also see gnat cross build failures on arm64. I'll check if this is related.