Hans,

I don't think you will ever be disappointed with your choice. Beyond the
elegance of RELAX NG, one other reason you won't be disappointed is the
growing availability of tools. All these tools are written in Java, btw.

For example, you can generate a DTD with only an XML instance document using
DTDGenerator [1].

Then you can generate a RELAX NG schema from this DTD using DTDinst [2].

Then you can validate an XML instance against the RELAX NG schema using Jing
[3].

You will very soon be able to translate a RELAX NG schema into XML Schema
using Trang [4].

Sun's Multi-schema Validator (MSV) can validate documents against DTDs, XML
Schema, and RELAX NG, and RELAX [6].

So how can you lose? %^}

Mike

[1] http://saxon.sourceforge.net/dtdgen.html
[2] http://www.thaiopensource.com/dtdinst/
[3] http://www.thaiopensource.com/relaxng/jing.html
[4] http://www.thaiopensource.com/relaxng/trang.html
[5] http://wwws.sun.com/software/xml/developers/multischema/
[6] http://www.xml.gr.jp/relax/


> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
> Hans Fugal
> Sent: Wednesday, June 26, 2002 7:36 AM
> To: [EMAIL PROTECTED]
> Subject: [gdmxml] RELAX NG
>
>
> Hello all,
>
> I have decided to go with RELAX NG for the schema. I will explain my
> reasons for doing so.
>
> First and foremost, it is more readable. A developer getting his feet
> wet with XML will be able to understand the schema just by reading it.
> Understanding an XSD schema takes a little research, IMHO.
>
> Second, I appreciate the more modular approach that RNG gives;
> datatyping is handled by external datatype libraries, including the XSD
> datatype library.  This particularly fits this project because in many
> instances I will not be dealing with datatype issues (yet). Others, or
> perhaps myself in the future, can and should deal with those issues, and
> a modular approach will be best suited for this application. RNG has the
> attitude that there is no universal datatyping scheme, with the
> exception of a few very basic types; each application has its own unique
> datatyping requirements, and those requirements should not be required
> to adapt to one datatyping scheme.
>
> Third, RNG does not modify incoming documents. XSD has some features
> that change the infoset, increasing the complexity of an implementation.
> This includes default values and default elements. In genealogy changing
> the source is generally considered a bad thing. I have seen enough
> differences between genealogical programs to know that defaults and
> assumptions are things we want to steer clear of. They have wreaked
> havoc in GEDCOM for years, and I'd like to avoid that.
>
> Fourth, in the research I have done it is generally agreed that RNG is a
> superior tool for most applications, with the added bonus of being
> simpler. I am very impressed by the sound computer science background of
> RNG. I believe in using the best tool, but I also believe in using the
> standard tool. RNG is an ISO draft recommendation, and while it may not
> be a 'web standard' to the degree that XSD is, I believe it has enough
> merit as a standard to be used, due to its ISO status and general
> acceptance.
>
> If you look at the cvs archives you'll see that I've added a basic
> definition of the PERSONA, without regard to data types, in both XSD and
> RNG. In the XSD I refer to an as-yet-non-existent persona_name
> definition, and in RNG I just cop-out with text for both persona_name
> and description_comments.
>
> I welcome your comments and opinions!
> Hans :)
>
> --
> "Everybody is talking about the weather but nobody does anything
> about it."
>         -- Mark Twain
>
> _______________________________________________
> gdmxml mailing list
> [EMAIL PROTECTED]
> http://fugal.net/cgi-bin/mailman/listinfo/gdmxml
>


_______________________________________________
gdmxml mailing list
[EMAIL PROTECTED]
http://fugal.net/cgi-bin/mailman/listinfo/gdmxml

Reply via email to