World: r3wp

Ah, recursive rules. :) I pushed my variables onto a stack when recursing, 
then popped them off when returning.

That tends to bloat the code a fair bit. ( push-vars [a b c] recursive-rule 
pop-vars [a b c] )

so then you get to thinking to generate this code automatically with 
a make-recursive-rule function, which takes a 

parse rule, looks for recursion in it, then surrounds it with push-vars 
and pop-vars for you (kind of macro expansion).
Or I did something like that, anyway.

I would be very interested in seeing this code of yours.  I'm planning 
on going further down the rabit hole of parse than anyone has gone 
before over the next couple of months.  ;-)

http://www.lexicon.net/antonr/rebol/library/make-recursive-rule.r

(Mmm.. it says it has a big flaw, but I know I'm using it successfully 
in a program. Maybe my input never trips the flaw, or I fixed it 
and forgot to write that down... Anyway, it gives you an idea how 
the user interface can look.)

Kewl, thanks!  I'll dig right into it. ;-)  (I'll be back, Wahahahaha!)

Mmm... I think only the old commented code has the problem. I would 
have to test that, though...

I'm forced to work with regular expressions now and I must say that 
PARSE is shining sun compared to that terrible black magic called 
regular expressions. They suck!

:-)

but they are said being powerfull tool, once you master them of course 
...

I always hear that if you need to use a UNIX OS, you need to learn 
them. So far I've done fairly well without them, but have been configuring 
procmail once, which was a complete nightmare. :-)

regular expressions are *very* powerful, as long as you *don't* compare 
them with Parse :-)

I've also been confronted with UNIX users who are disappointed that 
there is no regexp implementation in REBOL. :-) they find it hard 
to believe that there might be something superior to it

It'd be fun to compare parse and REs.....

Maybe a shootout between experts in both.   Both sides could learn 
a lot.

do you think you can do everything allowed in regexp in rebol's parse?

No....I suspect they each have core abilities they excel and other 
things they can only do with ugly workarounds, if at all.
It'd be interesting to see experts bringing out the best in both.

There was some talk about that on the ML a few years back. Joel and 
Ladislav had some great exchanges IIRC. Parse and RegEx are different 
tools. I can see the power of RegExs, but they aren't for the faint 
of heart when you start doing complex things. A RegEx can be very 
concise, and can be very handy for simple pattern matching.

http://www.compkarori.com/vanilla/display/PARSE-Versus-Regexs

Thanks Ladislav!

Is it possible to use "<" ">" and so on in parse rules? (not in string 
parse).

What are you trying to parse ?

I think the answer is no, but I am not sure of the question yet. 
Please give an example input.

I'm trying to parse something as simple as 1 < 2.

hm it's possible with any-word! ...

but that does not help me very much

Yes, I'm not having much luck here.

Probably it's something like Oldes' comma problem. :/

>> lt: to lit-word! first [ < ]
== '<
>> parse [1 < 2] [number! lt number!]
== true

great, thanks!

actually not great, but working :)

Unfortunately, there's probably no similar trick that can be used 
with comma, poor Oldes

replace/all str "," comma " ; :))

you cannot use this in a dialect!

before i load the string! But not too reliable, if you have "," in 
strings.

and I'm not such a poor:-) I am using double  parenthesis in my dialect 
instead of using commas (it's ugly, but it's working)

how does some sample code look?

I still think it's a shame, that we cannot load strings with commas, 
so we cannot forexample parse javascript or actionscript

my code looks like doSomeFunction(("a" add b) 1)

i think that is hard to do perfectly. for example "\"", how to parse 
that?

in actionscript it's doSomeFunction("a" add b, 1)

doSomeFunction("a" add b # 1)
doSomeFunction("a" add b . 1)
?

I know it woud be possible to improve my dialect, but I'm not going 
to do it in near future, I'm already used to write it this way:)

I think it's quite logical to close it in parenthesis than using 
chars like #

But chars like # are easier to search/replace. But i dont defend 
my solution, yours looks ok.
Do you write your dialect-code in rebol-blocks or in a string?

And for the Rebolek: I use in RSWF this:

slash: to-lit-word first [/]
	dslash: to-lit-word "//"
	rShift: to-lit-word ">>"
	UrShift: to-lit-word ">>>"
	_bigger: to-lit-word ">"
	_less: to-lit-word "<"
	_noteql: to-lit-word "<>"
	_lesseql: to-lit-word "<="
	_biggeql: to-lit-word ">="

and then:

'* (term-op: [ins-act #{0C}])
				| slash  (term-op: [ins-act #{0D}])	;Divide
				| dslash (term-op: [ins-act #{3F}])	;Modulo
				| rShift (term-op: [ins-act #{64}])	;RShift
				| UrShift (term-op: [ins-act #{65}]);UnsignedRShift
				| lShift (term-op: [ins-act #{63}])	;lShift
				| '| (term-op: [ins-act #{61}])	;bitwise OR
				| ['|| | 'or] (term-op: [ins-act #{11}]);OR
				| ['& | 'band] (term-op: [ins-act #{60}])	;bitwise AND
				| 'and (term-op: [ins-act #{10}])	;AND

Thanks Oldes, that should help me

Volker: blocks

it's much more easier as you can use Rebol's datatypes

i thought about something like "js-load string" which would do the 
comma-conversion. But if your dialect works, no need for extra work.