My personal responses to the public review comments

["Anthony B. Coates (Miley Watts)" <abcoates@mileywatts.com>]

Hi everyone.  Below are my personal responses to the public review  
comments, which I'm providing as an input to the TC.  Cheers, Tony.
--
Relationship with ISO 11179: Peter Brown asked why there was no formal  
reference to ISO 11179.  My answer is that ISO 11179 tends to be a basis  
for modelling methodologies (in particular the naming conventions), and  
doesn't apply directly to code lists.  I note that both UN/CEFACT and ISO  
20022 have some basis in ISO 11179, and both are interested in genericode,  
which suggests that genericode is complementary to ISO 11179.  I don't  
believe there is anything in the genericode specification which requires a  
formal reference to ISO 11179, nor do I think it would be helpful to users  
to add such a reference.

Committee draft of genericode 1.0 does not support xml:base: I noted that  
the current Schema does not provide support for xml:base, and I think it  
should, although only for location URLs.  I don't think it should apply to  
canonical URIs.

The basic structure of the lists: Martin Roberts proposed a change to  
genericode that would make it possible to apply Schema simple data type  
validation to values in rows.  The reason that genericode doesn't do this  
already is because genericode allows for data type libraries other than  
the XML Schema datatype library, just as RELAX NG makes this provision  
(although I expect most people to use the Schema datatypes most of the  
time).  This may be worth discussing, to see what people feel generally.   
Note that it isn't *so* straightforward, because you still have to check  
that the values in each row match the datatype of the associated columns,  
and that isn't something you can check with an XML Schema.  So you need  
some 3rd party validation tool anyway (and I'm currently writing one), and  
the question is whether you do the simple type validation in the Schema in  
or the checking tool.

There were a number of comments around the positioning of genericode by  
OASIS, and about making it clear that it isn't designed to be a run-time  
lookup format.  I think some stronger words to this effect should be added  
to the specification.

Use of "rule:*" attributes in genericode Schema - CodeList.xsd: I noted  
that the Schema contains "rule:*" attributes that I added to make it  
possible to identify documentation for a rule, as opposed to "just plain"  
documentation.  This runs foul of some Schema validators, so I think it  
needs to be changed to use some kind of "rule:*" elements instead of  
attributes.

Need to add a conformance clause: yes indeed, we need to decide what  
conformance means, and we may need some help from the OASIS folks with  
this one.  I'm a little confused by this requirement, in truth.  The  
specification itself is, in some sense, one giant conformance clause.   
Adding one extra clause won't necessarily help things.  Of course, the  
usual issue is that no two people interpret the same paper specification  
the same way, such are the vagaries of language.  This typically means  
that what is required is some kind of active (i.e. not passive)  
conformance checking that isn't (as) open to matters of interpretation.   
Two possible avenues for this (both which introduce schedule issues) are  
(i) to put together a library of valid and invalid genericode examples,  
and (ii) to have a "reference" validation application which is designed  
primarily for correctness, and to which other applications can be  
compared.  I have actually started writing an genericode (design-time)  
validator, which I intend to make open source, and which could be used as  
a reference validator is the TC wanted to go down that path.  Note,  
however, that neither of these conformance solutions can control (i) that  
way that information from genericode code lists is used at run-time, or  
(ii) that users are correctly assigning canonical URIs to new code lists  
and/or new versions of code lists.  Oh, I forgot a third option, we could  
define conformance based on the Schema, on the understanding that it is  
not able to check all of the rules which apply to genericode code lists.

Cheers, Tony.
-- 
Anthony B. Coates
Senior Partner
Miley Watts LLP
Experts In Data
UK: +44 (20) 8816 7700, US: +1 (239) 344 7700
Mobile/Cell: +44 (79) 0543 9026
Data standards participant: genericode, ISO 20022 (ISO 15022 XML),  
UN/CEFACT, MDDL, FpML, UBL.
http://www.mileywatts.com/