VisAD update

• To: VisAD List <visad@xxxxxxxxxxxxxxxx>
• Subject: VisAD update
• From: Tom Whittaker <tomw@xxxxxxxxxxxxx>
• Date: Mon, 02 May 2005 12:07:33 -0500

<ftp://ftp.ssec.wisc.edu/pub/visad-2.0/>

The javadoc associated with this release is at:
<http://www.ssec.wisc.edu/visad/docs/>

Here is a list of code changes since the March 14th update:

In visad:

2005-04-27 14:23  donm

        * DateTime.java: make a new ctor that takes a value and a Unit.
          Don't know why this wasn't there before

2005-04-26 16:28  curtis

        * data/bio/OpenlabForm.java: Tweak PICT support to work with latest
          QTForm fixes.

2005-04-26 13:46  curtis

        * data/qt/: PictForm.java, QTForm.java: Switch QTForm from a single
          reflected universe to multiple reflected to multiple universes
          with a shared custom class loader. This continues to work
          properly with QTJava but fixes bugs that occur with multiple
          QTForms accessing multiple files simultaneously.

          Unfortunately, there is still a serious efficiency issue with
          multiple simultaneous file access. (This fix addresses
          correctness, not efficiency.)

2005-04-26 13:32  curtis

        * util/ReflectedUniverse.java: Add constructor signature for
          passing in a custom class loader.

2005-04-22 15:41  tomw

        * Makefile: Set the target jar compatability to version 1.2.  Add a
          'windowtitle' to the javadocs of "VisAD", and increase the max
          heap for doing javadocs.

2005-04-21 16:03  donm

        * RealTuple.java: add new ctor to reduce memory and time and allow
          user to more easily shoot themselves in the foot.

2005-04-21 08:57  donm

        * Tuple.java, data/mcidas/PointDataAdapter.java: add a new ctor
          that takes a checkType boolean to avoid checking whether the type
          matches the data types.  use this in PointDataAdapter

2005-04-18 15:37  curtis

        * data/tiff/TiffTools.java: Handle WHITE_IS_ZERO more efficiently.

2005-04-18 15:32  curtis

        * data/tiff/TiffTools.java: Handle WHITE_IS_ZERO photometric
          interpretation properly.

2005-04-15 13:55  rink

        * Gridded3DSet.java: a couple minor tweaks in valueToGrid to fix an
          IDV problem

2005-04-08 13:25  tomw

        * data/units/DefaultUnitsDB.java: Added "Dobson" as a valid unit
          .01mm.  This is how ozone amounts are defined.  Abreviation "DU"
          also added.

2005-04-05 15:17  curtis

        * util/Util.java: Tweak comment to note that ImageJForm also
          supports PNG.

2005-04-05 15:17  curtis

        * data/ij/ImageJForm.java: ImageJForm supports PNG, so add png to
          list of suffixes.

2005-04-04 14:51  rink

        * Gridded3DSet.java: Fix two problems in valueToGrid discovered
          with the IDV: (1) change tests for (dot-products == 0) to
          dot-products     as close to zero as the precision allows, ie.
            Util.isApproximatelyEqual(dp,0).

          (2) If distance between the point and initial guess exceed
          about half the diagonal distance, find a closer guess     point:
          the algorithm seem to struggle in curvey grids     if the point
          to box vector crosses grid boundaries.

          Note: the methodology of this algorithm is the same as before,
          we're just giving it a little 'heurisitc' help, and fixing an
          arithmetic problem.

2005-03-19 05:17  billh

        * util/SelectRangeWidget.java: combine Don's fix with mine, which
          is used for ranges of less than 0.1 (using floor and ceil for
          very narrow ranges could make it difficult for users to select
          within their actual ranges of values)

2005-03-18 19:15  billh

        * util/SelectRangeWidget.java: Don Flagg's modification of my fix,
          that still fixes his problem but does not visibly widen the range
          in SelectRangeWidget.

2005-03-18 15:52  billh

        * util/SelectRangeWidget.java: My brute force fix for Don Flagg's
          problem, which I think is simply the limit of single precision
          floating point.  This fix extends the range of a
          SelectRangeWidget by a factor of 0.001 on both the low and high
          ends. Thus if your RealType values have range (0.0, 1.0) a
          SelectRangeWidget for this will have range (-0.001, 1.001).
          Please let me know if you see a problem with this in your
          applications.

2005-03-17 22:20  curtis

        * browser/RangeSlider.java: Fix floating point rounding bug Don
          Flaggs found in gripToValue.


2005-03-17 11:14  curtis

        * browser/RangeSlider.java: Potential fix for bug Don Flaggs found
          with SelectRangeWidget. Err on the side of caution (wider
          selected data) using Convert.shortString with ROUND_UP and
          ROUND_DOWN. There is probably a more accurate way to fix the bug,
          but this solution should prevent his problem from occurring.

2005-03-17 10:58  curtis

        * util/RangeWidget.java: Fix range inaccuracies in range widget by
          making use of new Convert.shortString(double, int) signature with
          ROUND_UP and ROUND_DOWN.

2005-03-17 10:47  curtis

        * browser/Convert.java: Add Convert.shortString(double value, int
          mode) method, for indicating which rounding mode to use.
          Possibilities are ROUND_NEAREST (default), ROUND_UP and
          ROUND_DOWN.

2005-03-16 04:46  donm

        * data/vis5d/Vis5DCoordinateSystem.java: fix a typo for rotated
          grids that was causing problems with Greg Tripoli's file.  At
          some point, we need to add in support for the MERCATOR projection
          (type 5) and maybe EPA (type 6).



In edu:


2005-03-28 10:21  rink

        * wisc/ssec/mcidas/AreaDirectory.java: fix a problem when
          (numbands>32)

--
Tom Whittaker (tomw@xxxxxxxxxxxxx)
University of Wisconsin-Madison
Space Science and Engineering Center
Cooperative Institute for Meteorological Satellite Studies
Phone/VoiceMail: 608.262.2759