[eml-dev] EML-2.1.0 Release Candidate-1 -- not valid

[inigo]

Hi Margaret,

no problem, I'll test the new changes soon. yes, you have a point 
-Altova is not the ultimate holder of the truth-,
but they have been right in pointing problems with our schemas. In 
addition, there are quite a bit of Altova users
out there, and it would be nice to provide them with a 'legal' schema.

Other helping hands would be desirable, specially since we are targeting 
further improvements to EML in the
near future.

While it is OK that you did not go through the list of changes I posted, 
I must say that I posted the list as the
sequence of edits I had to introduce to stop XMLSpy from complaining. 
While some steps may be deemed
worthwhile skipping, they may be deal breakers for this particular XML 
editor. What I can tell (and will tell you)
is whether the new edits pass the test.

The LNO has been promoting oXygen for a couple of years now as a very 
reasonable alternative to the XMLSpy,
I would be first to recommend it to anyone who wants to get some serious 
work with XML. It is also multi-platform,
and seems to be advancing faster than the Altova series. I would have 
moved completely to oXygen if it wasn't for
Altova's MapForce and other cool Altova utilities that helped me writing 
standard crosswalks (currently not available
at oXygen)

cheers, inigo


Margaret O'Brien wrote:
> Hi Inigo -
> I appreciate the work you did testing for these errors in spy2008. The 
> latest copy I can get ahold of is 2007, which agrees with the 
> validator at the W3C (http://www.w3.org/2001/03/webdata/xsv), that RC1 
> is a valid schema.  I made a few changes to the eml-documentation.xsd, 
> which I am attaching here. Can you do me the favor of trying this out 
> in your spy2008?  I want to be sure it works before I tag another rc. 
> My fix: It looked like spy was objecting to the repeated occurences of 
> <doc:module>, so I added a complexType whose elements referenced the 
> remanining globals, set cardinality = 0..many, and wrapped them in 
> sequence/choice to allow any them to appear in any combinations.
>
> This fix applies to bug 2703, again -- here is what has been going on: 
> You may have noticed that all the error reports are from the 
> "xs:documentation" sections of EML schema components. The content model
> for xs:documentation (and also for appinfo) is a wildcard, which by 
> default is processed with lax validation. If the processor can find 
> appropriate element declarations, it should use those instead. But to 
> make it unpredictable, processors are not required to perform the 
> schemas imports until they absolutely need to. So up until ~2005 was 
> processors probably did not bother to look at documentation.xsd, and 
> so the contents of <annotation> were checked only for well-formedness. 
> Fast-forward: processors get smarter, they might actually look at the 
> imported schemas, validated against them and spot errors. While 
> xmlSpy2008 happens to be ahead of the w3c in applying standards, they 
> still may not be catching everything.
>
> EML 2.0.x schemas were created according to the best recommendations 
> available for documentation and the same examples are still being 
> referenced in the literature today. All the changes I've made to 
> eml-documentation.xsd have further defined these elements to match the 
> way they were used by the original schema authors - and I hope, to 
> maintain flexibility for the future. Given that processors may or may 
> not load schema compoents, we cannot completely rely on their error 
> reports - and so our only defense in anticipating errors is our own 
> knowledge of schema construction. I would really like another person 
> to look over the changes to documentation.xsd and the use of its 
> elements in the other components for some requirement that I may have 
> missed.
>
> FYI, you can find info about XSV development on its author's home 
> page: http://www.ltg.ed.ac.uk/~ht/xsv-status.html, 
> http://www.ltg.ed.ac.uk/~ht/ . Altova (understandably) plays its hand 
> close, and so I havent run into any specifics on their implementation. 
> OxygenXML can validate using XSV as a plugin. And btw - OxygenXML is 
> FREE to anyone trying to make the world a better place: 
> http://www.oxygenxml.com/support_life.html (yes, this is a plug for 
> oxygen).
>
> Inigo - I did not address your entire list of changes (detailed in 
> your email of 2/27), although I did mention in the bug notes which 
> ones were applied and why. This note is long enough, and so I will do 
> that later
> today or tomorrow. Again, thanks for your willingness to be the tester 
> for spy.latest-version. If we need to get on vtc while you do that - 
> no problem - I'm at my desk today/tommorow.
>
> Cheers -
> Margaret
>