[Bug bootstrap/53343] [4.8 regression] options.c:9944:1: error: no previous prototype for 'common_handle_option_auto' breaks bootstrap

["mikpe at it dot uu.se" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53343

--- Comment #4 from Mikael Pettersson <mikpe at it dot uu.se> 2012-05-27 21:05:27 UTC ---
(In reply to comment #3)
> >--disable-build-poststage1-with-cxx
> 
> Is there a reason why you are using this option?

Yes,

1. It's a valid option, and it doesn't disable any functionality in the final
compiler.

2. Bootstrapping in C++ mode has been reported to increase bootstrap time, I
don't have any numbers, but this was mentioned last year around the time the
default was changed.  As I bootstrap more or less continuously on fairly
underpowered boxes (arm, m68k, even my dual-cpu sparc64 SunBlade2500 is too
slow compared to my x86 boxes), I don't want to spend cycles (and generate
heat) on something that is of no value to me.

3. The C++ FE has repeatedly failed to warn on some forms of code that the C FE
does warn about, allowing dodgy (for a C-centric definition of dodgy) code to
enter the gcc code base.  Bootstrapping in C mode ensures that such regressions
are detected.