Re: GCC build of HEAD failed for native with your patch on 2004-02-12T20:48:51Z.

[Geoff Keating <geoffk at apple dot com>]

On Feb 12, 2004, at 1:48 PM, Zack Weinberg wrote:

> Geoff Keating <geoffk@apple.com> writes:
>
> > The actual error is:
> >
> > echo '/*WARNING: This file is automatically generated!*/' 
> > >tmp-c-parse.y
> > sed -e "/^@@ifobjc.*/,/^@@end_ifobjc.*/d" \
> >      -e "/^@@ifc.*/d" -e "/^@@end_ifc.*/d"
> > /Users/regress/tbox/cvs-gcc/gcc/gcc/c-parse.in >>tmp-c-parse.y
> > /bin/sh /Users/regress/tbox/cvs-gcc/gcc/gcc/move-if-change
> > tmp-c-parse.y c-parse.y
> > bison  -o c-parse.c c-parse.y
> > c-parse.y:344: type clash (`' `ttype') on default action
> > make[2]: [c-parse.c] Error 1 (ignored)
> >
> > Why are we ignoring errors from bison?
>
> No clue.  I don't see this problem at all (bison 1.875a here), but I
> believe the appended patch will fix it, could you please test it?

This works for me.  Apple is stuck with bison 1.28 until we can drop 
support for versions of GCC that require it.


>         * c-parse.in: Don't give asmdef a type.
>
> ===================================================================
> Index: c-parse.in
> --- c-parse.in	12 Feb 2004 19:14:00 -0000	1.200
> +++ c-parse.in	12 Feb 2004 21:47:57 -0000
> @@ -202,7 +202,7 @@ do {									\
>
>  %type <ttype> scspec SCSPEC STATIC TYPESPEC TYPE_QUAL maybe_volatile
>  %type <ttype> initdecls notype_initdecls initdcl notype_initdcl init
> -%type <ttype> simple_asm_expr maybeasm asmdef asm_stmt asm_argument
> +%type <ttype> simple_asm_expr maybeasm asm_stmt asm_argument
>  %type <ttype> asm_operands nonnull_asm_operands asm_operand 
> asm_clobbers
>  %type <ttype> maybe_attribute attributes attribute attribute_list 
> attrib
>  %type <ttype> any_word
--
Geoff Keating <geoffk@apple.com>