This goes back to the discussion we had about rebranding some things for
version 2. We, at LASP, see TSDS as a *system* (or web app) that is
built around the OPeNDAP *server* (or service) that was developed at
LASP. Likewise, LISIRD and a LASP internal database usage tool are web
applications that use an independently managed configuration of LASP's
server. TSDS adds the special binary data store with specific data types
and eventually special Writers and Functions that can be packaged as
plugins. It's the OPeNDAP server (a.k.a. "TSS") plus plugins and related
services that make it a system. We might even want to consider changing
the last "S" in "TSDS" to "System".
Within LASP, the server has been known as "TSS". Now that we are moving
towards a netCDF-free implementation of the data model for the next
generation server (which is targeted for much more use within LASP in
addition to continued support for TSDS), LASP wants to rebrand the core
server as "LaTiS". So LaTiS can be thought of as "TSS 2.0". (It reminds
me of SunOS and Solaris versioning. :-)
The proposal that Anne and I (and a researcher from the computer science
department) submitted to NSF is to flesh out the Unidata common data
model and extend the LaTiS data model to cover all structural data types
culminating in a full featured OPeNDAP server that can run in the cloud.
The working name: "Stratis".
Since we decided to replace "TSML" with "NcML + TSDS conventions", there
really is no TSML 1.0 at this point. TSDS (1.0) uses TSS 1.0 which uses
NcML with the TSDS convention.
LaTiS (a.k.a. TSS 2.0) will replace NcML with the new TSML. It will
address many of the limitations we ran into with NcML and the incomplete
netCDF-Java implementation of the CDM. It will still play the same role,
though. Most of the development that goes into making a deployable
system is in the plugins.
I've attached a quick graphic to show the before and after.
On 06/18/2010 06:57 AM, Bob Weigel wrote:
> I thought that TSDS (the SourceForge project name) was the name of the
> server-side software and TSS was the old name. Is LaTiS a plug-in?
>> We are usurping the TSML name and replacing NcML and netCDF with our own
>> implementation of the data model and markup. Working around the NcML
> So will there will be TSML 2.0 which is read by TSDS 2.0 and TSML 1.0
> that is understood by all versions of TSDS?