Hi Martin, others, On Jul 18, 2012, at 4:01 AM, Martin Desruisseaux wrote:
> [..snip...] > >> Is there interest in the two communities to collaborate? Do our technical >> goals mesh well? etc. > On the Geotk side, some of the goals are: > > * Target both desktop applications and web services. +1, same here. We want to start out with a computational library, layer APIs on top (maybe web services, may others), build this capability into an end-to-end capability and drive out desktop applications and GUIs. > > * Follow closely the international standards. We are active members at OGC > and put large amount of effort in expressing OGC/ISO models in Java. This is one of our goals though clearly we haven't nearly gotten there yet and you guys are much further! > > * Focus on scientific needs (I'm an oceanographer). This means that we care a > lot about metadata, uncertainties and the like, and often prefer to get an > exception rather than an image that "looks good" but were computed on > inaccurate assumptions (while I admit that some users legitimately just want > an image that looks good, so it probably needs to be configurable, but I tend > to prone the exception as the default behaviour). Example: when a coordinate > transformation requires a datum change, Bursa-Wolf parameters (or something > equivalent) are needed in order to reach a 1 metre accuracy. Without > Bursa-Wolf parameters, the error may be around 1 kilometre. The default > behaviour is to thrown a "Required Bursa-Wolf parameters" exception in such > case, but the user can configure the library to ignore the lack of parameters > if they can accept a 1 km error. +1, agreed. > > * Try to keep the amount of dependencies low (favour JDK classes or some JSR) > at least for the core parts. The "coverage" modules have a JAI (Java Advanced > Imaging) dependenciy, but we are working right now on removing that > dependency. Example: we use the JDK logging framework instead that Common > Logging, but users can still redirect to Common Logging of Log4J. It is > possible to do such redirection through the standard Java API (I don't know > why Common Logging didn't do that in the first place. My guess is that JDK > 1.2 and 1.3 were important targets at the time Common Logging were designed). +1, this is fine with me too. I am fine with decisions like this, and we make them on the dev list, with general consensus amongst each other the community. > > * Extensive comments both in code and in javadoc. ++1! Cheers, Chris ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Chris Mattmann, Ph.D. Senior Computer Scientist NASA Jet Propulsion Laboratory Pasadena, CA 91109 USA Office: 171-266B, Mailstop: 171-246 Email: chris.a.mattm...@nasa.gov WWW: http://sunset.usc.edu/~mattmann/ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Adjunct Assistant Professor, Computer Science Department University of Southern California, Los Angeles, CA 90089 USA ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
