[Bug driver/69805] [6 Regression] ICE in greater_than_spec_func, at gcc.c:9722

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69805

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to vries from comment #3)
> This changes the semantics of greater_than_spec_func slightly. Strictly
> speaking not necessary to fix the ICE. But the new semantics will perhaps be
> easier to understand.

In fact, I originally changed greater_than_spec_func to assert argc is odd, and
verify that both argv[0] and argv[argc - 3] is "-" and compared argv[argc - 2]
with argv[argc - 1].  But then I found %* and thought changing the semantics of
%:gt() might be cleaner, allow better composability of spec language.
It could be perhaps even cleaner if we add some spec grammar extension that
gives you the last word from list of words, but I didn't want to spend too much
time on this.