The new CDAT 5.0 Beta has been released. The  older version CDAT 4.3 is still available, although it is not supported anymore, and eventually it will become obsolete.

New features included in CDAT 5.0: 

 - new installation (configure)
 - new VCDAT  - Lite mode 
 - interactive plot Edit Mode
 - TrueType fonts system
 - direct postscript, PNG, pdf, other
 - MPEG animation
 - improved CDAT Demo
 - new inline CDAT HELP

CDAT 5.0 uses NumPy module (the Numeric is no longer supported!) and it  works with Python 2.5+ versions.

Because of the changes coming with NumPy, CDAT 5.0 has new modules cdms2 and MV2 and the old cdms and MV modules are now deprecated. The automatic conversion program is supplied to help you with the transition from the pre-CDAT 5.0 to the post CDAT 5.0 programming with the new modules. 

The major change in version 5 is the transition from the Numeric scientific computing package to the NumPy package. NumPy is a full replacement and enhancement of Numeric, which is no longer supported. Since many of the CDMS array objects, such as Variables and Axes, are extensions of the Numeric array, this transition is nontrivial. Fortunately NumPy is very similar to Numeric. In many cases a simple change of module name will be sufficient to update a script. 

Namespace Changes

In CDMS Version 5 the cdms2 module interfaces to numpy and numpy.ma (the replacement for MA). For an interim period, the cdms, MA and Numeric modules will be retained for backward compatibility, but are deprecated. The change of namespace reflects the fact that the array objects are fundamentally different, and in certain instances behave differently. Also, MV is now MV2, cdms2.MV is an alias for cdms2.MV2, and regrid is changed to regrid2. The following table summarizes these namespace changes for the affected CDAT modules:

Converting scripts with convertcdms.py

The changes needed to convert from cdms/Numeric to cdms2/NumPy are relatively few and can mostly be automated. You can use the script convertcdms.py, installed in the bin directory to scan a Python script or directory and make the automated changes:

Automatic changes:
1. import cdms => import cdms2 as cdms
2. import regrid => import regrid2 as regrid
3. import MV => import MV2 as MV
4. import Numeric => import numpy.oldnumeric as Numeric
5. import MA => import numpy.oldnumeric.ma as MA
6. import cdms.MV => import cdms2.MV2
7. from cdms import XX => from cdms2 import XX (similarly for regrid, MV, MA, and Numeric)
8. from cdms.XX import YY => from cdms2.XX import YY (similarly for regrid, MV, MA, and Numeric)
9. import cdms as XX => import cdms2 as XX (similarly for regrid, MV, MA, and Numeric)
10. import cdms.XX => import cdms2.XX (similarly for regrid, MV, MA, and Numeric)
import XX, cdms, YY => import XX, cdms2 as cdms, YY (similarly for regrid, MV, MA, and Numeric)
11. MA.Float => Numeric.Float, similarly for MA.Int, MA.NewAxis
12. MA.Numeric => Numeric
13. The 'typecode=' argument in MA and MV functions is changed to 'dtype='
14. XX.mask() => XX.mask
15. XX.mask is None => ((XX.mask is None) or (XX.mask is MV2.nomask))
16. A keyword argument 'axis=0' is added to MA.sum, MA.average, MA.product, and MA.repeat
17. The translations in numpy.oldnumeric.alter_code1. This module is used for most Numeric and MA-related translations. alter_code1 methods do not have to be executed separately. See the sample chapters of “Guide to NumPy” at http://numpy.scipy.org .
18. array.typecode() => arr.dtype.char
19. array.size() => arr.size
20. array.mask() => arr.mask
21. Some typecode characters have changed. For example, ‘s’ => ‘h’.

Changes to be done by hand:
1. Slicing a singleton value from an array returns a scalar, not a 0-D array. Such scalars have a shape but no length.
2. To test for the type of a scalar extracted from an array, use isinstance() instead of type().
3. Masked arrays return a special nomask object instead of None when there is no mask on the array. This applies to getmask() and the array.mask attribute.
4. Masked array functions have a default axis of Nonel. The default in previous versions was axis=0.
5. The default datatype in numpy/ma/cdms2 is float. In Numeric/MA/cdms it was int.

To run convertcdms.py:

%   convertcdms.py foo.py
saves the original file in foo.orig (if translations are made) and converts the script.

%  convertcdms.py -r directory
converts all Python and C source code in the directory, and recursively in subdirectories.

% convertcdms.py -h
shows all options.

See  appendix D of the CDMS 5.0 for the detailed info on the transition from Numeric to NumPy.

We are excited to introduce the new CDAT Help application into the CDAT distribution. 

CDAT Help lets you browse and find the documentation about the methods and classes from the CDAT source code.

To start CDAT Help, in your shell prompt  just type

%   cdathelp

This will start a new application

You can find the method you are searching for by typing its name with a module name in front of it, in the editable text line

There are currently two ways to search the cdat software help text: through the Index capability (1) or the Search engine (2).

(1) In the example above, we chose to look for axis description in cdms2 module through the Index tab. As you type the word 'cdms2' the cursor and the indexed names will position themselves on the closest alphabetic name to the name you typed.

When searching for the documentation inside the 'Index' tab, please keep in mind that you are browsing through the index of modules, classes and methods or functions in CDAT source code. So you should be searching for the name of a valid method or function.   

Notice: 
You need to type the full name with the  proper namespace (module name). We are working on enhancements to allow searching for the name itself, but this will appear in next version of CDAT Help. 

(2) You can also search through the Search tab. This time you can type the 'axis' itself and you will find all the documents that have that word inside the doc string. The word will be highlighted in the right panel where the documentation is displayed, see picture below.

Future enhancements

The search option is rather simple for now, it simply parses the help text and returns all entries that contain that search word.

The future release of this utility will have enhanced search capability with more selective search options. We will add a 'find' function that will allow to search within a document in the right panel and highlight the search pattern in the resulting text. 

Please fell free to let us know how you would like to enhance this utility.

We hope you will find it useful and come to use it everyday.