Dear THREDDS and Netcdf-Java community:
After a long time of wrestling with this decision, I will be leaving
Unidata on August 7 to explore other opportunities. My intention is to
continue working on THREDDS for several more years in some other context,
possibly as an independent consultant, working directly for other
organizations to create custom solutions and deployments of the THREDDS
technology stack.
Unidata will continue to provide the home and core engineering support for
THREDDS. The THREDDS group (Christian, Dennis, Ryan, Sean) is in the
strongest position it has ever been to carry this work forward. Unidata
continues to be institutionally committed to the THREDDS project, both in
supporting and maintaining the current releases, and in development of new
capabilities.
Over the past year, we have significantly strengthened THREDDS with unit
testing, Coverity defect scanning, Github hosting, Jenkins and Travis
continuous integration, Gradle build management, and Jira issue tracking.
These software engineering practices are essential to turn the usual morass
of scientific code into a sustainable project that can survive the
departure of the founder(s).
This transition is also an opportunity for THREDDS to continue evolving
into a community supported project, in which many organizations contribute
features and fixes that benefit one and all. Only with ongoing community
support can a software project of this magnitude, based in a
publicly-funded organization, grow and thrive.
Where does THREDDS and Netcdf-Java fit into the technology landscape 16
years after they began? The software industry has unambiguously chosen
Java as the server-side technology for large, long-term, complex software
projects. The reasons for this include the strong-typing of the Java
language, portability across key hardware platforms, the richness of the
Java Virtual Machine (JVM) languages, the comprehensive Java development
tools, and the maturity and efficiency of the JVM implementations. THREDDS
and Netcdf-Java gain long term viability and sustainability by leveraging
this core Java technology.
On the client side, Java has been less successful, partly because it cannot
easily be linked with non-JVM languages such as C, and partly because the
needs of scientific programmers don’t fit Java’s strengths as well. With
Python gaining momentum throughout the geosciences, THREDDS will be
providing, along with the current library used in Java clients, new ways
for python clients to make use of TDS servers and the CDM Java software
stack. More details will be coming later this year.
When I leave Unidata, I will be leaving THREDDS in the best shape it has
ever been in. I am confident in the software, the Unidata THREDDS team,
and the larger THREDDS community to take this project successfully
forward. I will take some time to regroup after this transition to decide
what is next, and how I can best support the project.
Thank you for your ongoing commitment and patient support over these many
years.
John Caron
