[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: new version of inventory catalog XML

Hi Tom:

----- Original Message -----
From: "Tom Whittaker" <address@hidden>
To: "John Caron" <address@hidden>
Cc: <address@hidden>
Sent: Friday, August 31, 2001 9:21 AM
Subject: Re: new version of inventory catalog XML

> Hi John:
> In the Annotated Example...
>   * Please include something in here for collections of image data access
> ADDE and DODS - it looks like all the examples are for grids from NetCDF
> via DODS(?).  Note that whereas the DODS access points to a "file" on the
> server, the ADDE access does not...instead, the data are located through 3
> "names" - host, data group, and dataset.  [Even those are often
insufficient to
> "define" the collection of data.  Other elements, like spectral channel(s)
> also be needed (separate from "name").]

Its a good idea to get an ADDE example up and running, to see how that might
affect things, and I will do that ASAP.

Currently, the catalog "mental model" is as follows: a "dataset" is the
user-selectable "atom" of data content. Each dataset has a unique URL.
Collections are used to group datasets for clarity of presentation. (Does
that so far seem to work with ADDE?)

Is important to get the granularity right. ADDE grids, and perhaps other
data types probably need to be collected into datasets to avoid overwhelming
the catalog user. So there may need to be some middleware between the
catalog client and the ADDE server. (We are adding such a layer now in DODS
with the "Aggregation Server")

>   * In the "nested collection" (item 4), is the important point that the
> collection's serverID and dataType attributes are used by the children (of
> same element type) unless explicitly defined in the child's element tag?

You are correct that is how those attribues work. The main thing about
collections is to allow hierarchical groupings; for convenience and ease of
reading, we allow attributes to be "inherited" from parent elements.

>   * I would also like to see an example of defining a collection of Point
> as well.

We do have point data on the motherlode catalog, but no client for it yet. I
find it important to let clients drive the design, so we havent evolved much
here yet.

> In the DTD:
>   * Please add a "DataType" of TEXT as well.  I don't know about DODS
> but the ADDE protocol explicitly supports a data type TEXT and the server
> return an ASCII text file.  [My use of TEXT data right now is to retrieve
> indexed descriptors that are used by a GUI to allow reading of image data
> ADDE.  These are not, generally, directly linked to a particular dataset,
> rather contain information like the "long names" of spectral channels,
> are otherwise encoded in the data files by an index number.]

NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.