Re: [PATCH] Re: .macro behavior

[Hans-Peter Nilsson <hp at bitrange dot com>]

On Mon, 28 Feb 2005, Jan Beulich wrote:
> This is a resubmission of the patch at
> http://sourceware.org/ml/binutils/2005-02/msg00120.html to address the
> inconsistent acceptance of identifiers between the macro handling code
> and
> the rest of the assembler, including the unability to use a macro the
> name
> of which starts with a dot.
> Over the previously submitted revisions, this has just cosmetic
> changes
> (NEWS entry, change to MMIX testcase removed again based on that
> target's
> maintainer's request) and thus I did not re-build or re-run the tests.

> +++ 2005-02-28/gas/NEWS	2005-02-28 09:52:47.000000000 +0100
> @@ -1,5 +1,11 @@
>  -*- text -*-
>
> +* Macro names and macro parameter names can now be any identifier that
> would
> +  also be legal as a symbol elsewhere. For macro parameter names, this
> is
> +  known to cause problems in certain sources when the respective
> target uses
> +  characters inconsistently (like MMIX's use of ':'), and thus macro
> +  parameter references are no longer recognized as such.
> +

No, thanks, I don't like MMIX being accused of being
"inconsistent" like that.  It *is* consistent - all trailing
":"s are chopped off.  I think I now remember why; it's not
really related to ":" as separating namespaces, but the main
reason is so I could run the binutils test-suite!  Most tests
"unportably"  assumes that ":" is a label delimiter and not part
of the label.  Maybe I should insiste that tests be rewritten
and labels there be defined as "label .set ." :-)

And I still think there's greater benefit from keeping the
syntax of macro *parameter names* the same as today, something
like "[A-Za-z_][A-Za-z0-9_]+" and not dependent on the target
symbol character set.  (Heh, saying :alpha: and :alnum: would
imply it's locale-dependent.  I don't think we want *that*! ;-)

brgds, H-P