develooper Front page | perl.bootstrap | Postings from July 2000

Re: Working Group Proposal

Thread Previous | Thread Next
Nick Ing-Simmons
July 23, 2000 08:09
Re: Working Group Proposal
Message ID:
Sam Tregar <> writes:
>On Sun, 23 Jul 2000, Nick Ing-Simmons wrote:
>> I think that a (non absolute) goal of perl6's grammar should be that it 
>> is a near to yacc/byacc/bison/yapp LALR(1) as is possible without quite so 
>> much hackery in the lexer.  Some hackery in the lexer is not uncommon 
>> even in "pure" yacc languages (even C needs to notice typedefs), but perl5
>> has gone too far.
>Does that mean you think a rewite could yeild support for perl5's
>"flexible" grammer in a better way, 

There is superficially (it is the details that hurt and I don't know them all)
_some_ scope for cleaning up perl5's grammar.

>or that you think the grammar should
>be simplified to make it easier to parse?

One of the BIG things in perl6 is that we think we can handle backward 
compatibility by translation, so the grammar is "up for grabs" - it should 
be cleaned up - Larry has said it will be "a lot more consistent".
I think we should strive to make it LALR(1) even if (as seems inevitable)
it remains fairly complex.

I think the grammar will remain fairly complex - to keep the language 
understandable. One of my pet hates about Tcl is that while the surface 
grammar is near trivial (lists of "words") this means that all the "sense"
is below the surface. So for perl6 there is a balance to be struck between 
easy to parse, semantically clean , natural for humans, ... .

(The other BIG thing is that we can re-do internals to be more efficient 
and get some of the speed back.)

Nick Ing-Simmons

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About