[DAS2] spec status

[Andrew Dalke]

Hi all,

   I'm slowly making my way through the spec and integrating
the changes but it's taking me a lot longer than I expected.
I've been very precise and detailed in what I've said, but
as a result it's not very fun work.

The current status is in CVS under das/das2/new_spec.txt .
It contains the content of what I'm doing and is not formated
for HTML.  Once it's fleshed out I'll move it into the
existing HTML system.

I'm nearly done with the background information which
describes the different parts of the spec.  I'm about to
start making the changes in the read part.

What I've been working on is the generic header section for
the sources document.  This is the one Andreas and others
wanted, to add server information (eg, email address of
the server maintainer).

I want to do it like this

<SERVER>
    .. fields go here ..
</SERVER>

and define a few elements like

   <maintainer name="Andrew Dalke"
               email="dalke at dalkescientific.com">

That got me to wondering about the new property fields
we were talking about for the feature records.  They
were going to look like

  <propertyTag>This is the value</propertyTag>
or perhaps
  <propertyTag value="This is the value" />

(I forgot which we decided - I think the latter.  I haven't
gone through the notes to figure out which and it doesn't
matter for this.)

The propertyTag was a fully namespaced element, which
might look like
   das2:bgcolor
   dss:glyph3d

where
   das2 == http://www.biodas.org/...
   dss == http://www.dalkescientific.com/das-extensions

As I recall, the idea was that these fields will be
searchable.  Specifically, one of the fields was
   "top-level feature regions"
used to replace the /regions information.

I've having problems making those fields searchable in
a generic sense.  There are two problems.

1) need to send the fully qualified element name to the
server for search, like

   {http://www.dalkescientific.com/das-extensions}glyph3d=sphere

2) properties are typed.  Is the search for a generic
field by substring, by whole word in string, by numeric value
(so that "1" == "1.0").


I'm beginning to think it's just not worthwhile having this
the properties extensible by element tag name.  So I emailed
one of the Atom developers.

> As to why simple extensions exist; my belief is that they are there as 
> a sop to the RDF/semantic-web community, who really REALLY wanted to 
> make Atom an RDF application; the WG wouldn't go for that but was 
> willing to bless the "simple extension" mechanism, which can 
> straightforwardly be mapped onto RDF.

Okay, time for the phone conference call.


					Andrew
					dalke at dalkescientific.com