Sign In/My Account | View Cart  
advertisement

Article:
XBRL: The Language of Finance and Accounting
Subject: XBRL does not use XML Schema
Date: 2004-03-14 02:51:21
From: John Schlesinger
One of my biggest complaints against XBRL is that it claims to use XML Schema and, indeed, XBRL taxonomies look a lot like schemas, but in fact they are not schemas. What this means in practice is that a validating parser that supports XML Schemas, such as XERCES, is not capable of validating an XBRL document fragment and so it is necessary to invest in other tools for validation.


In fact, I am not completely convinced that XBRL document fragments are XML at all. Sure, they use angled brackets and look a lot like XML, but they cannot be validated using either a DTD or an XML Schema.


You should be very wary of thinking of XBRL as XML as you may be mislead into thinking that existing XML tools and middleware will be useful.

No Previous Message Previous Message   Next Message Next Message

Titles Only Full Threads Newest First
  • XBRL most certainly does use XML Schema - and then some
    2004-03-15 19:24:01 Hugh Wallis [Reply]

    As so often happens, even in the most inaccurate of criticism there lurks a grain of truth. Indeed a couple of grains of truth lurk in Mr. Schlesinger’s comments on XBRL but they are completely eclipsed by the reality of the total story.


    The two grains of truth lie in the phrases: "…taxonomies look a lot like schemas, but in fact they are not schemas..." and "...it is necessary to invest in other tools for validation...". However his other assertions completely nullify the effect of these snippets by their inaccuracy. Notably "...XERCES, is not capable of validating an XBRL document fragment..." and "...Sure, (XBRL document fragments) use angled brackets and look a lot like XML, but they cannot be validated using either a DTD or an XML Schema."


    So – let me help to put this all to rest.


    First and foremost XBRL IS XML. There is no question about this. As David vun Kannon mentions, many thousands of XBRL documents have been perfectly well processed as well formed XML by almost every XML processor under the sun. Not only that but, since the XBRL standard states that XBRL is XML and since it states that XBRL instances shall be valid according to XML Schema, it is, prima-facie, XML Schema valid XML. I think that Mr. Schlesinger’s confusion might arise from a lack of understanding of what XBRL taxonomies are and what their goals are. To boil it down to fundamentals, an XBRL taxonomy is a way of describing a very rich set of semantics for business reporting information that is expressed in XML. XML Schema goes only a very small way to providing the mechanisms for expressing such semantics and is used as much as it can so that, at the most basic level, an XBRL instance can be schema validated. The additional parts of XBRL taxonomies are all expressed using the W3C RECOMMENDED XML Linking standard (they linkbases as defined by that standard) which has been found to provide a very rich and flexible way to describe complex, layered, extensible semantics of XML instances.


    In the development of the 2.1 version of the specification, three independent software vendors, using a variety of XML Schema tools, validated hundreds of conformance suite tests incorporating standard XML Schema validators including XERCES as well as those provided by a well known software vendor from Redmond, Washington into their product offerings, thereby again completely refuting Mr. Schlesinger’s assertion. However, XML Schema validation is only a part of the story. In order to validate XBRL as valid XBRL, yes - additional software is necessary. This phenomenon is true of any other standard, XML based or otherwise. The additional software is necessary to determine if the rules of the standard that cannot be expressed using XML Schema (which, even its authors would acknowledge, is not capable of many things – take co-occurrence constraints as a small example) have been obeyed. So, yes Mr. Schlesinger, it is necessary to invest in other tools for complete validation. However it is definitely NOT the case that exisiting XML tools and middleware will (not) be valuable – it’s just that they won’t do the whole job for you. I challenge anyone to name any XML based standard that provides as rich and sophisticated a semantic capability as XBRL that can be processed and completely validated using standard, off the shelf, all purpose XML tools.


    It is also worth mentioning that there are some XML tools that make marketing claims to "validate XBRL" and yet fall far short of being able to do anything more than basic XML Schema validation of XBRL instances - and some of the more well known ones fail repeatedly in even that due to bugs in their implementation of the XML Schema standard, as can be observed from discussions on various XML Schema mailing lists. Before basing any judgement of any standard on some all-purpose tool's claimed ability to process it I would suggest that one should examine the tool's claims rather than casting aspertions on the standard itself. In the case of XBRL 2.1 that question is simple to ask. It is "Has your tool passed the conformance suite published by the body that defined the standard?".


    Hugh Wallis
    Chair, XBRL Specification Working Group
    Co-editor, XBRL 2.1, Specification
    Vice-Chair, XBRL International Steering Committee


  • XBRL does not use XML Schema
    2004-03-15 08:07:04 David vun Kannon [Reply]

    John,


    Before slinging around broad assertions such as this, you should check your facts.


    Many people experienced problems with the original XBRL 2.0 schemas that were released in 2001. These schemas contained an ambiguous content model, though at the time of release not all XML Schema tools noticed or cared. Subsequently, more tools did notice, the problem was reported to XBRL International, a fix was developed and published as XBRL 2.0a in 2002.


    XBRL 2.1, which was published as a recommendation at the end of 2003, comes with an extensive conformance suite of tests, all of which are built upon the foundation of XML Schema validation. As someone who has been involved in the design of XBRL since 1999, I can say from personal experience that XML schema validation for every kind of XBRL document has always been a goal since we dropped DTDs in favor of XML Schemas in going from 1.0 to 2.0 of the core spec.


    One of the frustrations of of the XBRL design team has been the inconsistency with which XML Schema has been implemented in commercial and academic tools. XBRL relies heavily on substitution groups in XML Schema. Support for this feature is spotty at best in some products.


    I can't speak directly to your problem using Xerces, but if you had to use "other tools", were they other tools that claimed to do XML Schema validation or not? There is an implicit acknowledgement in your statement that different tools reach different conclusions because they implement the XML Schema spec differently. That is not a problem that can be laid at the feet of any spec, XBRL, SVG, or anyone else, that is trying to build on top of XML Schema.


    Complaints about processing problems that are sent to the xbrl-public group on Yahoo are always answered promptly (as are most mesages!). XBRL members have met with product development groups on problems with specific products, and have gotten assurances in some cases that the problems will be fixed in the next release.


    Your further comments about XBRL and XML say more about the author than the subject. What tool are you using that rejects any XBRL document (to be precise, from the 2.1 conformance suite) as not well-formed XML? Our team at KPMG constantly processes XBRL 2.0a and 2.1 documents using commercial XSL tools in both Microsoft and Java based environments. If you can't do as much, upgrade your tools or your thinking or both.


    Cheers,
    David vun Kannon
    KPMG LLP


Sponsored By: