RE: ANN: XML Schema 1.1 Tutorial

Thank you for taking these concerns to heart.

p 28: Well, it's all up to you, but I still think you are conflating types 
with elements on slide 28, and in a tutorial that seems unfortunate.  My 
mail showed an example with two elements, <width> and <height>.   Also, I 
think calling it "two schools of thought" is a bit misleading and sets up 
unnecessary controversy.  I think there are tradeoffs and different use 
cases.  When you want to optimize something like business rules for a 
particular document type, or for a major subsection of a document, than I 
think putting the assertions on the type of the document root or the root 
element of the subtree makes sense.  In general, when you want to have 
reusable types, then it is appropriate to include in those types 
assertions appropriate to the correct use of that type, insofar as XSD 1.1 
can express them.  Those aren't so much conflicting philosophies as 
differing use cases, IMO.

p 232: there's a typo, perhaps carried from my original:  "BigDecimal or 
similar type " -> "BigDecimal or similar types "

Thanks!

Noah

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








"Costello, Roger L." <costello@mitre.org>
Sent by: xmlschema-dev-request@w3.org
07/18/2009 07:45 AM
 
        To:     "xmlschema-dev@w3.org" <xmlschema-dev@w3.org>
        cc:     (bcc: Noah Mendelsohn/Cambridge/IBM)
        Subject:        RE: ANN: XML Schema 1.1 Tutorial


 
Thanks Noah!

Excellent suggestions. I've incorporated them (see slide 28 and slide 
232):

http://www.xfront.com/xml-schema-1-1/xml-schema-1-1.ppt

/Roger




> -----Original Message-----
> From: noah_mendelsohn@us.ibm.com [mailto:noah_mendelsohn@us.ibm.com] 
> Sent: Friday, July 17, 2009 2:53 PM
> To: Costello, Roger L.
> Cc: xmlschema-dev@w3.org
> Subject: Re: ANN: XML Schema 1.1 Tutorial
> 
> Roger: overall, I think this is excellent.  A very impressive 
> piece of 
> work.  One disagreement and one suggestion:
> 
> * Disagreement: 
> 
> You suggest: 
> 
> "Due to the restrictions on what the XPath can reference, I 
> recommend all 
> assertions be placed on the document's root element. "
> 
> First of all, assertions go on types, not elements, but 
> that's not my main 
> concern as your intention is clear.  I just think this is not 
> in general 
> good advice, because it undermines the value of assertions on 
> types that 
> are usable across documents.  Let's say I have some type, 
> perhaps for a 
> measurement that was taken repeatedly.  An element of that 
> type might look 
> like:
> 
>         <width minMeasurement="3.02"  maxMeasurement="3.06" 
> meanMeasurement="3.04" />
> 
> or, in the same document:
> 
>         <height minMeasurement="4.3"  maxMeasurement="5.2" 
> meanMeasurement="4.8" />
> 
> Let's assume these elements are both of complexType measurementType.
> 
> Very possibly, I'd want to <assert> that minMeasurement <= 
> meanMeasurement 
> <= maxMeasurement.  Why would I want to do that at the 
> document level? 
> Presumably, measurements like this could be used in lots of 
> documents and 
> on lots of differently named elements.  For this case, the assertion 
> belongs on measurementType, I think.  The same might well be 
> true of an 
> address type, which could check that the first two digits of 
> a zip code 
> are consistent with the name of the state.  Indeed, there are 
> many, many 
> common examples in which you don't want to hoist assertions 
> to the root, 
> IMO.  Reuse of these things is important!
> 
> * Suggestion:  in your discussion of precisionDecimal, you 
> might indicate 
> that one of the reasons it has been added is that it embodies 
> XML support 
> for the recently adopted IEEE 754-2008 standard for floating point 
> decimal.  Java BigDecimal or similar type in other 
> programming languages 
> are examples of implementations of IEEE 754-2008, just as 
> Java double and 
> float are implementations if IEEE 754 floating point binary. 
>  The new XSD 
> precisionDecimal type supports the use case where you have 
> data in your 
> program that is represented using such a type, and want to 
> serialize it 
> and validate in in XML, while preserving the 754-2008 semantics (e.g. 
> precision matters).
> 
> Noah
> 
> 
> 
> --------------------------------------
> Noah Mendelsohn 
> IBM Corporation
> One Rogers Street
> Cambridge, MA 02142
> 1-617-693-4036
> --------------------------------------
> 
> 
> 
> 
> 
> 
> 
> 
> "Costello, Roger L." <costello@mitre.org>
> Sent by: xmlschema-dev-request@w3.org
> 07/17/2009 01:37 PM
> 
>         To:     "xmlschema-dev@w3.org" <xmlschema-dev@w3.org>
>         cc:     (bcc: Noah Mendelsohn/Cambridge/IBM)
>         Subject:        ANN: XML Schema 1.1 Tutorial
> 
> 
> 
> Hi Folks,
> 
> I created a tutorial on XML Schema 1.1:
> 
> http://www.xfront.com/xml-schema-1-1/xml-schema-1-1.ppt
> 
> 
> I am announcing it here before announcing it on xml-dev.
> 
> I am eager to get your feedback on:
> 
> 1. Are there any mistakes in it?
> 
> 2. Is it clear? Is it easy to understand?
> 
> 3. Have I missed any of the new functionality?
> 
> 
> I would like to especially thank Michael Kay and Michael 
> Sperberg-McQueen 
> for their patience in answering my endless questions.
> 
> /Roger
> 
> 
>

Received on Monday, 20 July 2009 21:18:35 UTC