This is an old message I'm replying to, so I'll copy most of it. On Fri, Mar 10, 2006 at 07:05:34PM +0100, Leopold Toetsch wrote: > Code snippet # NameSpace::name > > .sub main # '' actually parrot [1] > store_global 'x', $P0 # parrot::x > cl = newclass 'Foo' # parrot::Foo > .end > > .namespace ['Foo'] # 'Foo' but really parrot::Foo > > .sub foo :method # Foo::foo actually parrot::Foo::foo > store_global 'x', $P0 # Foo::x ... parrot::Foo::x > cl = newclass 'Bar' # Foo::Bar ... i.e. create nested name > .end > > .namespace ['Foo'; 'Bar'] # parrot::Foo::Bar > ... > > .HLL "perl5"," # new top-level namespace > .sub 'x' # perl5::x > ... > > [1] implicit HLL := 0, aka 'parrot' [2] default namespace > [2] the name might change still
I still don't like "parrot" as the default namespace, but I dislike it a lot less than when you first suggested it. So no change for now. :-, > The default HLL namespace is the pseudo-root namespace for all namespace > operations except 'get_namespace' and 'add_namespace', which work from the > actual namespace root. Change needed: All namespace opcodes should be HLL-root-relative, without exception. I never meant otherwise, but I didn't document it right in a couple cases. Should be OK now. > The biggest change to the current implementqtion will be, that > 'store_global' and 'find_global' will work relative to the *current* > namespace and not to namespace-root like now. Right. -- Chip Salzenberg <[EMAIL PROTECTED]>
