This is true that I tend to overlook commercial constraints when I
develop code... But JDK 1.4 have so many advantages, it is really
impossible for me to scale down CTS and GCS modules to JDK 1.3. Scaling
them down to JDK 1.3 would be a huge step back in the past, and would
force us to rely to a bunch of external libraries in order to do what
JDK 1.4 can currently do very well, e.g. logging and XML parsing. Among
thing that can't hardly live without JDK 1.4:
- An implementation of CoordinateSystemFactory close to OpenGIS
specification could not support informations like authority,
area of interest, etc. without JDK 1.4.
- All the GCS design I have explained in OpenGIS meeting (using NaN
numbers for separating clouds, lands etc. from geophysics values)
would be close to impossible without JDK 1.4! I really need supports
for float and double image in order to make this design possible, and
this kind of support has been added only in JDK 1.4. To the very, very
best case (relying only on JAI's FloatDoubleColorModel), a JDK 1.3
implementation would display an image totally black or totally white
(see new methods in ColorSpace; a correct display of images with
geophysics values is impossible without them).
- There is also the brand new API for image I/O, which never existed
prior to JDK 1.4 and is not available separatly. All my I/O stuff for
images is based on that. The JAI codecs are deprecated. Only the image
I/O API gives enough flexibility for loading only a portion of an
image (instead of the full image), selecting bands of interests,
reusing a destination image, plug or own codecs, and all thing like
that in a single API.
Actually, scaling CTS and GCS modules down to JDK 1.3 would sacrifice so
many functionalities that I would not be able to do my work anymore. I
would be forced to move my development elsewhere...
The J2SE 1.4 release is available for about 8 months now. Maybe a
significant amount of customers are planing to do the move, if they have
not already done?