Looks great! It's a model of clarity, the examples are extremely
helpful, and I find it a relief to see such clear <context> syntax. :-)
I have just a few comments:
- General questions: Will the "shoulds" and "musts" here ultimately be
turned into numbered Rn rules? Is this document destined to be folded
into the NDR document?
- General comment: There are a number of little stylistic and
copyediting needs and the graphics could be normalized a bit. E.g., the
"Customization through other means" header (line 283) has some weird
page breaking going on around it. I might be able to help with this
sometime next week, if desired.
- One-per-context (line 114): I'm sure this is treated below, but the
rule is more subtle than expressed here, right? It's not that
particular slot (context driver) out of the eight can't be used again,
but the value in that slot (the "context") must be more specific than
the previous value supplied for that slot (if any) in the derivation chain.
- "Explicit" type definitions (line 172): I would say "named" rather
than "explicit", because both named and anonymous types are explicit
(that is, they both have a complexType or simpleType element around them).
- Requiring the use of a derived type (line 179): Doesn't the derived
type have to be bound to an element in the user's namespace in order to
require use of the new type instead of the base one? It doesn't quite
say so here (or does the next bullet say it for a different reason?).
- Deletion of required components (line 286): The general case should be
stated as x..y to x-1..y. 0..y is just an example of a particular kind
of reduction.
- Abstract ur-types (line 299): I don't quite understand this graphic.
Where does d,e,f get added on the ultimate type on the left? And
shouldn't one example derivation in the picture be shown emanating from
the abstract level rather than the derive 80/20 level?
Eve
Eduardo Gutentag wrote:
> On behalf of Matthew, who seems to be unable to post to this list (and
> come to
> think of it, I don't even know if I can, we'll see). I just saw this, so
> I have not read it at all. So, sight unseen, I'll venture that it still
> needs
> more and better examples :)
>
> Dan Vidt expressed willingness to work w/me on examples. Dan, you still
> there?
>
> Quoting from Matt's message to me:
>
> "The document seems great to me. I only made a few editorial changes. I
> think the important next step is for this to be reviewed by other
> members of the UBL TC (particularly the LC and NDR subcommittees) and to
> be looked at by potential users (not sure how to sollicit feedback on
> this). My biggest concern is that much of the content of the document
> may not be sufficiently clear for a UBL newbie; obviously as a co-author
> it is hard for me to make a judgement on this."
>
>
>
--
Eve Maler +1 781 442 3190
Sun Microsystems cell +1 781 354 9441
Web Technologies and Standards eve.maler @ sun.com