Validation question

[Yves Savourel <ysavourel@enlaso.com>]

Hi Christian, all,

Looking back at the inline markup last call, I'd like to add to my answer.

I said that I hoped we could validate the format using the schema as much as possible. For example the nid attribute in the <ph> element can be present only if the content of <ph> is empty. I'd like to see that kind of validation possible with the schema.

As for verifying processing expectations, which may involve comparing 'before' and 'after' documents: We mentioned that such verification may be done using more complex XML validation tools such as RelaxNG or Schematron.

My question is: What would be the advantages of using such mechanisms over the Java application Rodolfo proposes? In all cases it's about writing some code to validate a resulting document. There are possible drawbacks in using RelaxNG/Schematron, for example:

- You need a RelaxNG or a Schematron processor, which are less likely to be pre-installed or available than a Java VM.

- Far less people know how to write RelaxNG or Schematron rules/programs than Java: this make the maintenance of the validation suite more difficult.

I'm not saying using RelaNG or Schematron is bad. I'm just wondering what are the pros/cons of the three solutions if you separate the validation of the document using XSD and the verification of the processing instructions using something more powerful.

Cheers,
-yves