Parse does not have "not" match type.

I was playing around with the idea of using rebol to convert EBNF to Rebol parse rules. A strange occupation maybe, but I thought it might be useful one day. Anyway, I found a rule in EBNF, as described in XML1.0 - http://www.w3.org/TR/REC-xml.html, for which I believe there is no equivalent in Rebol - and I cannot see how it can be worked around. It is the A - B rule which says "matches any string that matches A but does not match B". I looked into the use of this rule in the xml spec and found that it does not relate simply to characters, so I cannot just do something with complement . Is there a way, or no? Thanks, Brett.

Hi, I think, that Parse has got a bug as in:

>> parse "a" [none skip]

Hi, try this: (but look out!) As long, as Rebol functions are CQSB with DRP, there are problems with Rule-res word local to A-B-rule function. A-B-rule: func [ "Generate an A-B parse rule" A [block!] {A-rule} B [block!] {B-rule} /local succeed fail rule-res ] [ [[B to end (rule-res: [none skip]) | A (rule-res: [])] rule-res] ] Regards Ladislav

That's brilliant Ladislav. It took a little while to understand what magic you put there, and I learnt something as a result. I've made a few tests so far and cannot fault it. Thanks, Brett.

Hi, the latest version (see below) is better (the generated rule can be used recursively if needed): A-B-rule: func [ {Generate an A-B parse rule} A [block!] {A-rule} B [block!] {B-rule} /local o ] [ o: make object! [ A-rule: A B-rule: B res-rule: none ] bind/copy [ (self) [ B-rule (res-rule: [to end skip]) | (res-rule: A-rule) ] res-rule ] in o 'self ] { Example: a: [any "a" "b"] b: ["aa"] a-b: a-b-rule a b parse "ab" a-b parse "aab" a-b } not-rule: func [ "Generate a not A parse rule" A [block!] {A-rule} /local o ] [ o: make object! [ A-rule: A res-rule: none ] bind/copy [ (self) [ A-rule (res-rule: [to end skip]) | (res-rule: []) ] res-rule ] in o 'self ] { Example: a: [any "a" "b"] not-a: not-rule a parse "ab" not-a parse "b" not-a parse "" not-a }

hi put up the question just in case i missed something+ seems could remove the (self) under bind/copy to achieve this function. or is it required for some reason? -z --- [lmecir--geocities--com] wrote:

Hi Zhang, the problem is the GC bug. If the bug didn't exist, (self) is unnecessary. As long, as the GC bug exists, (self) prevents the GC from collecting O. (for more information you can read Words, Bindings and Contexts thread) Regards Ladislav

Thanks, Ladislav so my reading - Object has local context local context could be GC'd if not referenced (self) below provide the reference Thanks, -z --- [lmecir--geocities--com] wrote: