On Mon, Mar 10, 2008 at 4:33 PM, chromatic <[EMAIL PROTECTED]> wrote:
> On Monday 10 March 2008 13:19:59 Will Coleda wrote:
>
> > Note that:
> > > > + STRING* orig,repl;
> >
> > doesn't DWIM.
>
> That's why abutting the star and the type is an abomination under heaven, and
> our style guidelines recommend strongly against it.
That actually did help me find the problem.
>
> > Also, if you're not careful, pmc2c.pl tries to rewrite
> > Parrot_PCCINVOKE too, not just PCCINVOKE. (I used a more verbose style
> > than I would have otherwise to avoid this.)
>
> That should be fixable; can you give me an example?
Sure.
// The invalid:
Parrot_PCCINVOKE(INTERP, P0, "replace", "SS->", orig, repl);
// Or the valid:
STRING *method = string_from_literal(INTERP, "\\\\");
Parrot_PCCINVOKE(INTERP, P0, method, "SS->", orig, repl);
// Or.
#define REPLACE_FOO string_from_literal(INTERP, "replace")
Parrot_PCCINVOKE(INTERP, P0, REPLACE_FOO, "SS->", orig, repl);
Seems like forcing the PCCINVOKE to match the beginning of a word
should be sufficient to avoid all of these.
Which reminds me, shouldn't this be called INVOKE? (particle++)
>
> > "make test" in tcl is now passing failing only tests that already
> > failed in trunk (and fewer of them!)
>
> That lsort continuation jumping problem bothers me, but I couldn't
> immediately
> see why it's doing that.
Something about exceptions across namespaces. I'll see if I can narrow
it down under a separate thread.
> -- c
>
>
--
Will "Coke" Coleda
