https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115743
--- Comment #8 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #7)
> install-data-local change does look correct though or at least closer to
> what it should be.
If we're going to check three possible variables instead of two, I'd prefer to
use a loop. Something like (untested):
@for str in dlname library_names old_library ; do \
libname=`sed -ne "/^$$str=/{s/.*='//;s/'$$//;s/ .*//;p;}" \
$(DESTDIR)$(toolexeclibdir)/libstdc++.la`; \
test -n "$$libname" && break; \
done; \
> What it means is gdb.py can't be generated until it gets installed it looks
> like.
Do you mean that generating gdb.py has to be part of the install-data-local
step? Is that so that it can use DESTDIR in the path we substitute for
@toolexeclibdir@?
I don't think we want to use DESTDIR there, because that would be incorrect if
the files are relocated later.
