[eml-dev] Community opinion on minor revision of EML

[inigo san gil]

Eml-dev,

Margaret O'Brien, Mark Servilla, James Brunt,
Corinna Gries and other non-members of the
eml-dev committee have expressed a favorable
opinion about  releasing a new  version of EML
sooner than later. This new release would be
minor in scope, that is, it would address those
critical bugs that make the schema non-valid
according to the most stringent rules  of XML
schema. This release (let's call it eml-2.0.2)
would not break backward compatibility, and
would not mean a change on the EML habits of
any user or community that has adopted the
standard.

Why a minor release and not a more ambitious
release? That would be a good question. There
are plenty of documented eml-issues that go
beyond the basics, and that are worthy of further
study, discussion and eventual implementation.
However,  we feel that a minor release may test
the waters and perhaps help prepare the stage
for a more ambitious  release. Furthermore, this
minor release would fix all the complaints of strict
editors such as XMLSpy, version 2008. 

I am asking you about your opinion about a
release of this sort within the next month or two.

I am attaching a compressed archive containing
an EML instance with "fixed" critical problems
as described above. (I changed the archive
extension from .ZIP to .PIZ (ignore case) to
bypass some email filters --change it back--.) 
I DID NOT  touch any designations of  the
schema versionsing (that is, I did NOT put eml2.0.2
where eml2.0.1 is expressed, and so on) this is
just to illustrate the final shape of the schema
after changes are made. This fixed schema tested
OK for validation on XMLSpy  (up to the latest
2008) release,  as well as with oXygen8.2.

What continues below is the log of changes
introduced.

What specific changes are here? Many issues were
documented recently by Margaret in eml-dev.
As a summary, these are the changes performed

------------
Bug - 2054 the "//eml:eml/additionalMetadata/describes" bug
 
--Remove the cardinality of /eml:eml/additionalMetadata/describes

I.e: edit this:
<xs:element name="describes" type="xs:string" minOccurs="0" 
maxOccurs="unbounded">
to this:
<xs:element name="describes" type="xs:string">
--------------

(this from Margaret via bugzilla)
Bug 2703 - This is NOT a duplicate of bug 2054
<http://bugzilla.ecoinformatics.org/show_bug.cgi?id=2054>.
These changes are needed to make eml validate in xmlSpy2006+:  
eml-text.xsd:
<xs:complexType name="TextType"> => add mixed="true"   
and to the choice of this complexType add minOccurs="0"

eml-documentation.xsd:
add elementFormDefault="qualified" to the root-element

History:
Some folks who have newer versions of xmlSpy (2006, 2007)
 were not able to validate eml documents. The failures are
 mysterious and variable. Validating eml.xsd is successful,
but when validating a data package, Spy replies that the
schema is invalid. It usually fails in eml-view, but also
occasionally in  eml-project, or eml-software.  Initially,
Altova tech support  said that it was a bug in Spy. They
recently  changed their tune,  and suggested the above
changes to our schema. With these  changes, local copies
of EML can now be validated in Spy2007.  Altova did
not quote the parts of the xml specification relating
 to this error. Given the error messages from Spy, I
was not  able to see where eml failed to comply with 
the xml specs.  I have put the email trail with error
messages at the url above,  if that helps anyone else
get to the bottom of this.  It could be that this xmlSpy
 behavior is still a defect in the software, and that eml
 schema is valid as is, and that tech support is just trying
to shut me up.

Relationships to other bugs:
1.  It's possible that this was part of the behavior that Inigo
reported in bug  #2479 
<http://bugzilla.ecoinformatics.org/show_bug.cgi?id=2479>,
or described in emails to eml-dev.
2. This is not related to bug 2054
 <http://bugzilla.ecoinformatics.org/show_bug.cgi?id=2054>,
 in which eml.xsd failed due to the <any> element in
additionalMetadata. That failure was noticed first in xmlSpy2005,
whereas this behavior first appeared with Spy2006.

---Recently we noticed that the edits described above
were not enough to make EML a bit more professional
 (that is, to comply  with the XML schema rules -
mind you, many wildly used, mature XML  schemas
have deeper validation problems than EML!)


In addition, after the two edits suggested above I had to
perform  the following edits to comply with Schema rules.
None of these bugs are registered in Bugzilla.

((Format is :   Bug Error Message, description -->   Implemented 
Solution / workaround))

1 -  :   <ulink> unexpected under <para>  in eml.xsd and elsewhere--> 
add element <ulink>, and child tag (<citetitle>) to the <para> element 
(in eml-text.xsd), following the "docBook" model.
2 - :   no more elements under <moduleDocs> --> close element 
<moduleDocs> before we open <doc:module>s (group was not nested 
appropriately in eml.xsd)

3 - :   further elements required under element <doc:description>  in 
about ALL *.xsd files ---> added <section xmlns=""><para> group 
following <doc:description> and corresponding closing elements.
4 - :   further elements required under element <doc:example> in most of 
*.xsd files  --> added <section xmlns=""><para> group inside 
<doc:example> and corresponding closing elements.
5 - :   empty element <doc:example/> in "methods description" 
eml-entity.xsd --> entry <doc:example/> deleted
6 - :   element <sidebar> not allowed under element <section> in 
eml-attribute.xsd  --> removed <sidebar> and  corresponding closing element
7 - :   <title> element opened after <para> group, not allowed ---> 
closed </section>, reopen <section>, then <title> is OK. (so, added 
</section><section> before <title>)
8 - new soft-complain: namespaces added for "coverage" in eml-dataTable.xsd
9 - :  many empty <doc:description/> elements in eml-spatialRaster.xsd 
---> Many elements removed or filled with "summary" content
10 - :  many empty <doc:description/> elements in 
eml-spatialReference.xsd ---> Many elements <doc:description> filled 
with preceeding  "<doc:summary>" content
11 - :  many empty <doc:description/> elements in eml-spatialVector.xsd 
---> Many elements <doc:description/> removed from eml-spatialVector.xsd

-------------- next part --------------
A non-text attachment was scrubbed...
Name: eml-2.0.1_mod.piz
Type: application/octet-stream
Size: 4499508 bytes
Desc: not available
Url : http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/eml-dev/attachments/20080227/0bea6ee4/eml-2.0.1_mod-0001.obj