Extending%20WDSS-II

1
Extending WDSS-II
C

• Valliappa Lakshmanan
• National Severe Storms Laboratory University of
  Oklahoma
• lakshman_at_ou.edu

2
Meant to be extended

• WDSS-II is meant to be extended
• Few things are set in stone
• We use factories every where
• You can introduce your implementation objects
  into those factories
• Plug in your own writers, readers, builders, data
  formats, etc.
• Existing algorithms will transparently pick up
  your changes.

3
New algorithm

• Adding a new algorithm or tool
• Is the most common extension
• Write a new executable that links to the WDSS-II
  libraries
• Will automatically be pluggable into a WDSS-II
  deployment system
• Can ingest data from an existing system
• Can provide products instead of products already
  being used.
• Use w2algcreator dont need to be a C expert
  to do this
• You need to know C very well in order to extend
  the WDSS-II framework in any other way

4
Adding functionality to WDSS-II

• Use the factories to extend WDSS-II
• Make sure that your introduceSelf() function is
  called by Baselineinitialize()
• For example, this is how gzip files are read
• FactoryltReadergtintroduce(gz, new GzipReader()
  )
• GzipReader implements the Reader interface

5
Available factories

• These are some of the available factories in the
  infrastructure
• Reader
• To read from something other than file, gz, bz2
• Writer
• To write to something other than file, gz, bz2
• IndexRecordNotifier
• To notify to something other than LB (such as SQL
  data base or CORBA)
• DataEncoder
• To write formats other than netcdf,xml,awipsnc
• Builder/Formatter
• To read formats other than netcdf, xml, level-III
• Factories at the algorithm level
• FilterFactory (w2imgproc)
• New Kernel filters for smoothing, etc.
• VolumeCreatorFactory (w2merger)
• New strategies for blending

6
Adding to factory

• Write a class that implements the appropriate
  interface (Reader,Writer, etc)
• Create a static introduceSelf() method that adds
  itself the factory with a name
• Call this from a static initialize() method
  common to the shared library
• Directly call this initialize() method from
  Baselineinitialize() to add directly to WDSS-II
• Use a static initializer if not every body needs
  it or if you dont have access to the code of
  Baselineinitialize()
• More on this later

7
Writing ingest programs

• To read a new data format, write an ingest
  program
• Use DataDirectoryManager
• http//cimms.ou.edu/lakshman/WDSSII/software/doc/
  w2dox/html/classcode_1_1DataDirectoryManager.html
• Create a DataHandler
• Implement processData
• Data will be handed to you in chunks
• Use BOIStream if your data are binary
• Does byteswapping
• Use AsciiLineReader if your data are textual
• Look at example ingest programs in w2algs/ftptonc
  and w2algs/ldm2netcdf

8
Writing native readers

• You can make algorithms read your new data format
  natively
• Ingest programs take the input data and convert
  it into netcdf/XML
• The Level-III format is read in this manner
• Use w2ext/w2nexrad as an example
• Some what more complex choose whether you really
  want to do this.

9
ExtensionLoader

• Use a static intializer if you can not modify
  Baselineinitialize()
• Look at documentation of ExtensionLoader
• Look at w2ext/w2nexrad for examples
• Populate WDSS-II factories in this initializer
• Package up all your classes into a shared library
• Add library to environment variable
  W2_EXTENSION_LIBS at run-time

10
Using the documentation

• When writing new algorithms
• Consider if the functionality already exists
• Can you link together existing tools to get your
  functionality?
• Experiment, then use those algorithm classes to
  form your new algorithm (see the algorithm
  documentation)
• Start with the data type documentation
• LatLonGrid, etc.
• Process your data
• Use DataEncoderwriteData to write all your
  intermediate outputs