bootstrap.sh isn't batteries included, it's kind of expected to fail and
you're kind of expected to work your way through it. It's just there to
help you through what is ultimately a very large, complicated, manual,
and error prone process.

On Fri, Jan 3, 2020 at 8:33 PM Drew DeVault <sir@cmpwn.com> wrote:
>
> bootstrap.sh isn't batteries included, it's kind of expected to fail and
> you're kind of expected to work your way through it. It's just there to
> help you through what is ultimately a very large, complicated, manual,
> and error prone process.

Thanks for the reply - It was your interesting blog post on porting
Alpine to RISC-V which landed me in this predicament!  I guess I'll
keep poking at it.  I've looked at the referenced line number (953)
which was not very enlightening and tried adding a missing file to the
gcc source (standard.ads.h) for no other reason that it was causing a
broken symlink,  I'm sure there's worse ways to spend a Friday
evening!

For the record, I did end up getting a working base load on an old
armel board.  I'm not convinced I could recreate what I did, but the
two things that fixed the majority of the issues I had was disabling
most of the compiler language support when invoking the bootstrap
script:

CBUILD=x86_64 LANG_D=false LANG_OBJC=false LANG_GO=false
LANG_FORTRAN=false LANG_ADA=false ./bootstrap.sh armel

and commenting out the stripbin function in /usr/bin/abuild because
the wrong strip binary was being called for many packages, not all,
but many.  Probably not the best solution but I figured I could go
back and rebuild everything natively later?