W3C home > Mailing lists > Public > xmlschema-dev@w3.org > March 2007

Re: XSD1.1 editorial proposals

From: <noah_mendelsohn@us.ibm.com>
Date: Thu, 15 Mar 2007 17:57:51 -0400
To: "Pete Cordell" <petexmldev@tech-know-ware.com>
Cc: xmlschema-dev@w3.org
Message-ID: <OF18A11D6F.4A7872FA-ON8525729F.00777BF4-8525729F.00787D1D@lotus.com>

I don't speak officially for the workgroup, but I'll give you my best 
intuition:

Pete Cordell writes:

> I've been reading XSD 1.1 and have a number of editorial comments that I 

> hope would make the spec easier to read.  I imagine that I should submit 

> these to www-xml-schema-comments@w3.org.

That's much appreciated.  Yes, thank you, the comments mailing list is the 
right place to send them.

> One thing I'm wondering is, should I do one big long e-mail 
> with all comments in there, or should I send one e-mail per comment?

I'd say, err on the side of one per comment, but don't go crazy.  If 
you've got a bunch of things that are truly typos, or very minor and 
editorial, combine them.  If they're substantive and not closely related, 
then separate them if practical.

> The really big thing I'd like to suggest is breaking the 
> complexType section 
> into multiple parts.  This has 5 complicated intertwined use 
> cases (if you 
> include the <complexType><sequence> scenario) and is a major barrier to 
> understanding XML schema.  (Whereas attributes, which are trivial by 
> comparison, have two sections; the second one not really adding much.) 
> Would such a suggestion be automatically dismissed?
> 
> Similarly, would suggesting a change in the order of sections be 
> automatically dismissed?

Again, I don't speak for the workgroup.  By all means submit the comment 
if that's what you suggest.  I will say that during the authoring of 
schema 1.0 we struggled mightily with these questions.  Multiple forms 
were tried over periods of months, some with radically different 
structures from the others.  I'm not saying that makes your suggestion a 
bad one.  I'm certainly not happy with the accessibility of the 
specification, and I expect many other workgroup members aren't either. 
Still, we've tried very hard more than once to improve things.  Also, as 
you may be aware, our team is relatively small, so everything has an 
"opportunity cost".   Given that we're on a fixed charter schedule, 
whatever we sign up to do that's new tends to mean that something else 
gets done less well or not done at all.  Speaking just for myself, I'd 
rather not bite off midsize chunks of such editorial rearrangement. 
There's a case to be made that we should either someday do a total 
restructure, perhaps into two documents, one about components and one 
about surface syntax, or else just leave it alone.  FWIW, I think the 
component design lends real architectural strength to schemas, but it 
greatly complicates telling a simple story to users about how individual 
language features work.   Consider (using shorthand syntax, since I HATE 
our syntax!):

        <simpleType name="derived" base="B">
                <maxInclusive>10</maxInclusive>
        </simpleType>

It's really hard to understand what the derived type is without 
considering >something< along the lines of components.  Clearly what's 
written above doesn't tell you much.  Is it an integer?  A short?  Is it 
an illegal derivation from string?  All that depends not just on what B 
is, but probably in turn on what B's ancestors are. Of course, another way 
to go would be to make you spell out all the facets over again at each 
derivation step.  That would make things much less convenient and in some 
senses less powerful, but the specification would get way simpler 
structurally.  We'd just say:  the type is what you see in the simple 
mapping from the syntax.

So, the language is inherently hard to describe.  Some of that is for good 
reasons IMO, such as the example above, some of it is for bad reasons that 
I wouldn't defend.  I will say we've tried very hard for a long time to 
make the specification easier to read and learn from.  I'm sorry we 
haven't done better so far.

Noah


--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------








"Pete Cordell" <petexmldev@tech-know-ware.com>
Sent by: xmlschema-dev-request@w3.org
03/15/2007 06:19 AM
 
        To:     <xmlschema-dev@w3.org>
        cc:     (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        XSD1.1 editorial proposals



I've been reading XSD 1.1 and have a number of editorial comments that I 
hope would make the spec easier to read.  I imagine that I should submit 
these to www-xml-schema-comments@w3.org.

One thing I'm wondering is, should I do one big long e-mail with all 
comments in there, or should I send one e-mail per comment?

Is it worth me presenting the comments here on xml-dev first so that peers 

have the option to say what I'm suggesting is rubbish, thus saving the WG 
some effort?

The really big thing I'd like to suggest is breaking the complexType 
section 
into multiple parts.  This has 5 complicated intertwined use cases (if you 

include the <complexType><sequence> scenario) and is a major barrier to 
understanding XML schema.  (Whereas attributes, which are trivial by 
comparison, have two sections; the second one not really adding much.) 
Would such a suggestion be automatically dismissed?

Similarly, would suggesting a change in the order of sections be 
automatically dismissed?

Thanks,

Pete.
--
=============================================
Pete Cordell
Tech-Know-Ware Ltd
for XML to C++ data binding visit
http://www.tech-know-ware.com/lmx/
http://www.codalogic.com/lmx/
=============================================
Received on Thursday, 15 March 2007 21:58:43 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 23:15:41 UTC