[antlr-interest] #( root children ) syntax change proposal?

DVLeonid dvleonid at land.ru
Sat Nov 22 10:19:14 PST 2003


Hello Ric,

How about to use somethink like ## or ^^. It has no sense in C++ and I
think in another languages too. (Yes I borrowed this idea from Bjarn).

RK> Hi,

RK> On Tue, Nov 18, 2003 at 11:21:04AM -0800, Terence Parr wrote:
>> The #(...) notation derives from wanting to use parens like LISP but
>> not wanting the ambiguity with (... | ...) alternative subrules. The #
>> was the only character left in ASCII for me to use ;)  I have always
>> not liked it, but now it's ingrained.  That said, I've been thinking of
>> changing it for ANTLR 3 into
>> 
>> ^(root c1 ... cn)
>> 
>> in tree parsers and tree construction actions.  There MAY be some 
>> ambiguity with the ^ suffix operator, but I think we might wanna try it
>> out.
>> 
>> Anybody have any thoughts?  I like it as it's more intuitive that that
>> is a tree ;)  # is random.

RK> I would be inclined to make the thing target language dependant. For C++
RK> I've very often contemplated to replace the # with something different in
RK> order to have no trouble with preprocessor directives. ^ gives conflicts
RK> with xor operators in the action parser parts. This would mean the action
RK> parsers would even become bigger messes than they are now. Unless the tree
RK> operator parts are easy to separate from the target language.

RK> Cheers,

RK> Ric



-- 
Best regards,
 DVLeonid                            mailto:dvleonid at land.ru


 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/ 




More information about the antlr-interest mailing list