On Wed, Nov 16, 2005 at 08:11:24AM -0800, jerry gay wrote:
> from a read of the spec, the expected behavior of this code is unclear to me.
>
> in the rule, i'm matching the builtin whitespace rule, creating an
> alias with the same name as the named rule, and matching the builtin
> rule again. this brings up a few questions:
>
> does the alias replace the named rule or coexist with it?
It coexists.
> when does the binding of the alias to the pattern occur?
It essentially occurs when the rule is compiled; given
/ <ws> $<ws>:=(foo) <ws> /
we have three things capturing into $/<ws> in the same lexical
scope, thus $/<ws> will be an array of match objects.
> .sub 'main' :main
> load_bytecode 'PGE.pbc'
> load_bytecode 'PGE/dumper.pir'
> .local pmc p6rule, rulesub, match
> .local string target
> .local int try
>
> p6rule = find_global 'PGE', 'p6rule'
> rulesub = p6rule('<ws> $<ws>:=(foo) <ws>')
>
> try = 1
>
> target = ' foo '
> bsr Do_match
>
> target = ' foofoo'
> bsr Do_match
>
> target = 'foofoofoo'
> bsr Do_match
>
> goto End
>
> Do_match:
> match = rulesub(target)
> print 'match #'
> print try
> print "\n"
> match.'dump'()
> print "\n"
> inc try
> ret
> End:
> .end
>
> here's what pge currently does, but i'm not certain this behavior is
> correct, because it's implementation is not yet complete.
>
> match #1
> : < foo @ 0> 0
> <ws>[0]: < @ 0> 0
> <ws>[1]: <foo @ 1> 0
> <ws>[2]: < @ 4> 0
>
> match #2
> :0
>
> match #3
> :0
>
> is this what you expected?
Yes.
Pm
