Re: Some editorial matters for -syntax-

Ted,

Thanks for your comments.

I'll respond below to a selection of those comments;  no comment means that
I accept your suggestion without reservation.

At 15:45 30/11/98 -0800, Ted Hardie wrote:
[...]
>In 3.5.1, the description of substring matching as "not...relevant to
>feature set selection" should probably be simplified to "Substring
>matching (for tokens?) is excluded from these proposals"; there are
>areas like hierarchical types where it might be relevant, but we don't
>want to deal with them--so let's say that instead.

This paragraph was intended as an explanation of the first bullet point.
I'm not sure that your proposal says anything that isn't already stated
there.  I would be happy to delete that paragraph.

>I think the discussion of the consequences "implied universal
>quantification" so early is confusing in 3.5.1, so I would shorten
>that paragraph to something which says:
>
>"Within the semantic framework described by this document,
>Boolean-valued features can be used where presence tests would be used
>by some directory selection mechanisms."

Or:
 "Within the semantic framework described by this document,
  Boolean-valued feature tests can be used where presence tests would
  be used in a directory search filter."
?

>A further reference to the sections on the scoping rules (including
>7.2) could be put there, but it may not be necessary.

I don't feel that is necessary.

>In 3.6, I would eliminate the phrase "suggesting an "ideal" option
>that is equally or more preferred than any other"; the direction of
>the 0 to 1 range has already been given, and this seems to muddy the
>waters from that description.

The intent was to clarify the consequence of omitting ;q=x.  If it doesn't
do that then it should go.

>For 4, my first question is whether it is independent enough to be
>moved up in the document; there are a lot of forward reference in 3.x
>to it, and it seems cleaner to move the whole lot up.  If that isn't
>possible, the forward references should give the appropriate number
>(there is one "next section", for example, in subheading 3.4 which
>might refer to 3.4.1, 3.5 or 4).  If you do move 4, the "discussed
>above" reference may need to be changed; in either case, it should
>list the sections where it is discussed rather than indicate the
>direction to scan.

Section 3 was intended to introduce the underlying concepts and ideas,
before section 4 launches into the syntax specifics.  As I write, the
forward referencing I have noted is a result of using the notation
described later to illustrate some ideas.  I would prefer to play down the
forward referencing but leave the material in the order given if that does
not cause discomfort to other reasons.  Where forward referencing is used,
I'm happy to make it more explicit.

>In 4.2.3, I would add a brief reference to the URL tree of the
>registration document, just to discourage folks from having
>well-formed by unregistered feature sets.

I propose replacing the clause beginning ", but such registration is not
necessary..." with a new sentence:  "Unregistered feature tags should be
allocated in the URI tree, as described by registration procedure [3]"

>In 4.2.4.2, I fear that we may actually have to say the + or - sign
>must precede the expression "n/m" so we are not later asked why we
>allow "10/-2".  I realize that the very name "rational number" should
>exclude this, but it's probably best to make it very clear.

I thought that it was clear from both the syntax and section 4.2.4.2
(paragraph 2, 1st sentence) that the sign must precede both integers of a
rational value.

If you still feel this needs emphasizing then I can add something.


>In 4.3, I think there is a missing closing parenthesis after the
>q=0.8.

And before!  Well spotted.

>In bullet 7 in section 5.1, the satement that "Further, it can be used
>as a statement of the resulting feature set for possible further
>matching operations" needs either to be exanded in line or,
>preferably, to have a pointer to the sections in which this is
>discussed.

OK.  I'll work on it.

>In 5.5, would is it clearer to say "comparable" rather than related in
>the statement "(NE a b): 'a' is not equal and not related to 'b'." ?

If you think that's better.  What I really want to say is:

    'a' is not equal to, less than or greater than 'b'

but that seemed like a bit of a mouthful.

#g

------------
Graham Klyne
([email protected])