Re: feedback on draft-fielding-uri-rfc2396bis-03.txt

Roy T. Fielding wrote:

...-encoding" as
> part of yet another rewrite of the characters section.
> 
>> - "...extends...to the first question-mark ("?"), number-sign ("#"), or
>>    the end of the URI string."
>>
>>   probably should be:
>>
>>    "...extends...to the first question-mark ("?") or number-sign ("#")
>>    or the end of the URI string."
>>
>>   (Also, there are several similar non-parallel sentences.)
> 
> 
> Your suggestion looks like bad grammar to me.  Is this an obscure rule?

Looking at it now, my suggestion probably could be clarified, but no,
I don't think the principal of parallel construction is obscure.

If you distribute "extends" over your "or" (like distributing
multiplication over addition as in A x ( B + C ) = A x B +  A x C),
your wording:

   extends...to the first question-mark ("?"), number-sign ("#"), or
   the end of the URI string."

yields:

- extends to the first question mark
or
- extends to the first number sign
or
- extends to the first the end of the URI string.

Note how that third branch comes out malformed.  (In addition, note
the reference to the _first_ end of the string, which doesn't
make sense, confirming that one of your "ors" got grouped wrong.)

You want the distributed version to be:
- extends to the first question mark
or
- extends to the first number sign
or
- extends to the end of the URI string.

That can be "factored" as:

    ...extends to the first question mark or number sign or to the end
    of the URI string

(clearer than my original suggestion) or as:

    ...extends to the first question mark or number sign or the end of
    the URI string

(my original suggestion, admittedly probably less clear).  (Of
course, other "factorings" are possible.)


(P.S.  "Number sign" and "question mark" aren't hyphenated (in
English...at least in American English...at least in American English
I know).)

Daniel

Received on Monday, 9 February 2004 11:23:26 UTC