eliminate warnings from arlex.c

[Ian Lance Taylor]

"Ben Elliston" <bje@au1.ibm.com> writes:

> Okay, fair enough.  How about if I just remove the nounput parts of
> the patch, leaving:
> 
> 2004-12-08  Ben Elliston  <bje@au.ibm.com>
> 
> 	* arlex.l: (yylex, yyget_lineno): Provide extern declarations.
> 	(yyget_leng, yyget_in, yyget_out, yyget_text): Likewise.
> 	(yyset_lineno, yyset_in, yyset_out): Likewise.
> 	(yyget_debug, yyset_debug, yylex_destroy): Likewise.

I know this is not particularly helpful, but I think the right
approach here is to fix flex.  flex should be providing prototypes for
the functions which it provides.  Those functions declarations are by
and large only meaningful for flex, and they are at least marginally
vulnerable to breaking if a new version of flex ever comes out.

Any other opinions out there?

(Incidentally, I just looked at arlex.l.  Probably the best fix would
be to rewrite that in a simple C function.  It's a rather inefficient
use of lex.)

Ian