[Bug driver/47236] Spaces introduced in option processing

["jsm28 at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47236

Joseph S. Myers <jsm28 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |jsm28 at gcc dot gnu.org
         Resolution|                            |INVALID

--- Comment #1 from Joseph S. Myers <jsm28 at gcc dot gnu.org> 2011-01-10 00:32:29 UTC ---
This has nothing to do with the specs language.  Rather, the canonical form of
an option accepting both joined and separate arguments is the form with a
separate argument, all option processing in both the driver and the core
compilers now uses common logic to process options into logical
cl_decoded_option structures, and all subsequent processing now uses those
structures - and if it needs to regenerate an option for specs processing, uses
the canonical form of the option (meaning specs no longer need to handle
noncanonical options if they are noncanonical in a way indicated in the .opt
files, e.g. through being aliases or through using joined arguments where
separate arguments are also accepted).  -D options certainly appear to be
passed in canonical (separate) form to subprocesses (e.g. by gcc -E) to me.

If an option accepts a separate argument and it is passed down to subprocesses
by specs, those subprocesses must accept the form with the separate argument.