[lojban] Re: On parser extensions.

[Jorge "Llambías" <jjllambias2000@yahoo.com.ar>]

--- Robin Lee Powell <rlpowell@digitalkingdom.org> wrote:
> xorxes pointed out that "pabi'ipazeva'e" is currently not grammatical,
> whether you shove all the words together or not.  :-)  The reason seems
> to be that the current parsers don't wait long enough, and try to
> interpret things as "le pa bi'i <something>".

The EBNF grammar has:

tanru-unit-2 = (number | lerfu-string) MOI #
number = PA [PA | lerfu-word] ... 
lerfu-string = lerfu-word [PA | lerfu-word] ... 

Your PEG grammar seems to agree:

tanru-unit-2 <- (number / lerfu-string) MOI free* 
number <- PA (PA / lerfu-word)*
lerfu-string <- lerfu-word (PA / lerfu-word)*

So I don't see how it manages to get a {bi'i} in there.

> My questions are:
> 
> 1.  Did any of you have that kind of trouble parsing this sentence by
> hand?

Unlikely. The intention is quite obvious. But according to the grammar,
MOI takes only a pure number.

> 2.  Does this count as a change to the language, or merely cleaning up
> LR(1) issues?[1]

Do you mean allowing {quantifier MOI}? If it can be done, I think 
I'd be in favour, but it would count as a change, wouldn't it?

mu'o mi'e xorxes
 


	
		
__________________________________
Do you Yahoo!?
Yahoo! Small Business $15K Web Design Giveaway 
http://promotions.yahoo.com/design_giveaway/