Re: PATCH: Darwin / Mac OS X native support

[Stan Shebs <shebs at apple dot com>]

"Joseph S. Myers" wrote:
> 
> On Thu, 1 Mar 2001, Stan Shebs wrote:
> 
> > Although the patch is large, it's already been stripped down to the
> > minimum that still works, there are only a couple changes to generic
> > code (a new macro in varasm.c), and the generic rs6000.* changes are
> 
> Where is the documentation for this new macro?

Not written yet - I wanted to be sure that the new macro was an
acceptable approach first.

> > + /* wchar_t is int.  */
> > +
> > + #undef      WCHAR_TYPE
> > + #define WCHAR_TYPE "int"
> > + #undef      WCHAR_TYPE_SIZE
> > + #define WCHAR_TYPE_SIZE 32
> 
> Is the GCC default of "unsigned int" correct for wint_t?  If not, you
> should define WINT_TYPE as well.

OK thanks, I didn't know that.

> > + #undef      CPP_SPEC
> > + #define CPP_SPEC "%{!traditional: -D__STDC__}                       \
> > +                       %{mdisable-fpregs:-D__NO_FP__}                \
> > +               %{F*} %{ansi} %{fno-asm} %{header-mapfile}    \
> > +               -D__APPLE_CC__=" STRINGIFY(APPLE_CC) "        \
> > +               %{static:-D__STATIC__}%{!static:-D__DYNAMIC__}\
> > +                   %{MD:-MD %M} %{MMD:-MMD %M}"
> 
> I don't understand exactly what should go in which cpp specs, but are you
> sure that handling of -traditional and -MD belong here when they're
> already handled in a target-independent manner?

Actually, I'm not sure.  I'll check on it.  Some of the flags (like
-F for frameworks) are only interesting in connection with future
patches, technically should not be part of this patch.

> > +       %{dynamiclib:/usr/bin/libtool -arch_only ppc %{!all_load:-noall_load} %J \
> > +             %{current_version:}%{compatibility_version:}%{init*}} \
> > +      %{!dynamiclib:/usr/bin/ld -arch ppc %{@:-arch_multiple}} \
> 
> You seem to be adding various options in specs such as -dynamiclib.  These
> should be documented.  While many existing targets have such undocumented
> options, they shouldn't be allowed in for new targets.

Hmm, good point.  The GCC manual doesn't seem to have a good place
to document compiler flags that only tie into particular systems'
linkers, but I'm open to suggestions.  I could write a Darwin section
for chapter 2, or even add a new chapter a la the VMS info.

Stan