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

Re: [netcdf-java] announce: release candidate 2 for Netcdf-Java library version 4.0





John Caron wrote:
catalog2 is not ready for release, so we removed it from jar. probably should 
remove from javadoc to
avoid confusion. but i could leave it in the src so you can look at if you want.

Working with the released dev version is bleeding edge enough for me right now, but we might keep that option on the table.

- Is the opendap jar consistent with the scm.opendap.org/svn repository?

yes, but they dont seem to be releasing any jar files.

I've been in communication with James, Nathan, and Peter, so at least they know someone cares about the Java implementation.


Here's what I'm doing. I have extended the opendap java code to make a
customized time series server. I am using DAP Sequences to package the
data. Our data are in HDF5 files and I have NcML files for them which I
use to read the data into the server via the NetCDF APIs. We also have
simple binary files for which I plan to implement an IOServiceProvider
so I can use the NcML approach with them. I'm also overriding the
default ascii output for Sequences to produce csv formatted output for
easier integration with spreadsheets.

that sounds very cool. Im working on point data reading, send me a URL so I can 
see what it looks like.


I'll give you a peek as soon as I have a test release that isn't horribly broken. The NcML worked incredibly well. I'm doing some of this work for Bob Weigel (GMU, ViRBO) and really selling him on these technologies. He simply wanted some metadata to describe his files and I opened them up to the NetCDF API. He was happy to see how easily a THREDDS catalog could provide metadata and a link to the data service (my TimeSeriesServer via NcML). When I told him about cataloggen, he said that was exactly what he's been trying to write. He hears a lot of reinventing being discussed in the space physics VxO community, so I expect he'll be trying to get the word out about Unidata technologies.

as you've noticed, we arent handling sequences yet in NcDODSServlet. Is your 
code general enough
that we might want to incorporate? Or am i misunderstanding what you have done?

I was hoping that I could pull off a generic csv ascii output including sequences, but I got lazy (and agile) and did the simple thing that works. It's interesting that Nathan suggested that I need to modify opendap.servers.ascii.asciiSeq.java. It turns out that my code was using opendap.servlet.AsciiWriter which has your name on it. My server now extends AbstractServlet and overrides doGetASC to use my extension of AsciiWriter.

Is the current ascii format for Sequences written in stone? If a csv friendly format would be preferred, I might be inspired to make it more general and contribute that.


we have some plans for allowing user-pluggable HTML rendering. What if we gave 
you a way to plug in
your code that took a catalog or dataset and did the rendering for it?


Pluggable HTML rendering would be great. I'm already looking into providing a custom form from my server when *.html is requested. I imagine that I would also want a custom representation of the THREDDS metadata and service options.

We are doing a lot of refactoring right now for a 4.0 release. I cc:ing ethan 
since he's working on
that.

I'll keep that in mind if I have any suggestions.

BTW, has anyone tried putting THREDDS catalogs in an XML database like eXist?

Thanks,
Doug