Re: MS/CW-style inline assembly for GCC

[Andrew Haley <aph at redhat dot com>]

Matt Austern writes:
 > On May 3, 2004, at 9:19 PM, Richard Henderson wrote:
 > 
 > > On Mon, May 03, 2004 at 05:35:40PM -0700, Stan Shebs wrote:
 > >> Can you be more specific about the difficulties?
 > >
 > > (1) Control flow.  MSVC allows branches to leave the asm block,
 > >     something that doesn't map onto existing gcc asm blocks.
 > >     The hack we had for this was truely odious.
 > 
 > <other horrible things snipped>
 > 
 > Thanks for the list.  As you say, awful stuff.
 > 
 > I don't think we want to emulate all of these things, which means
 > that not all code will be directly portable from MSVC to gcc.  Maybe
 > this means that "MS/CW-style" is a bad name for what we're trying
 > to do.  Perhaps a better name would just be: more natural syntax
 > for inline assembly.

I don't really undersatnd how one syntax for inline assembly may be
said to be more "natural" than another.  I don't think this makes any
sense: it's just a matter of what you are used to, or perhaps a matter
of preference.

 > We're not just trying to follow someone else slavishly, we're
 > trying to adopt a new syntax because we think the new syntax is
 > better.

Well, that's the problem.  As I understand it you're proposing a whole
new syntax that is less powerful but is to be preferred because it's
"less ugly".

Maintaining two totally different assembler syntaxes doesn't seem to
me like a sensible plan.  Perhaps, with a little thought, we could
improve what we have at the moment.  "less ugly" might, for example,
involve replacing operands like "%1" with real names in a slightly
less evil way than we do at the moment.

Andrew.