Iteration 3

• Phase: Elaboration
• Dates: March 18 - April 21, 2010

Tasks

Architecture

• Complete GDAL integration into the spatial library
  • Simplify the library's architecture by removing the raster-driver layer.
  • Determine which GDAL formats allow direct writing, forward information to Rob
  • Depending upon how many formats can be directly written (as opposed to being written to a within-memory raster and subsequently converted), either reconfigure to allow many formats (internal raster conversion) or a limited suite of formats (direct writing).
  • Rob prefers simpler and faster approach: direct write with predetermined formats/file extensions, but need to evaluate whether there are sufficient formats.
  • If direct write, formats determined by file extension. Need to create dictionary associating file extension with formats.
  • If many formats (internal conversion), perhaps format could default to the associated extension with an optional file type parameter that overrides the default? E.g., "MyMap sppMap.img (jpeg)"
  • Note that the above applies to file writing only. File reading is done via a brute-force approach within GDAL.
  • Work on the utility library will wait until later - there may be opportunities for students to contribute.

Students

• Finish hiring students
• Set students up on computers and training
• Begin assigning tasks

Assessment

Architecture

• The GDAL integration was not completed.
• However, the GDAL proof-of-concept was completed.
• Raster-driver layer removed from architecture. This is being completely replaced by GDAL.
• Formats determined and determined whether writing directly or via copy. The good news is that we can avoid using CreateCopy. Using only Create this will require the smallest change to the architecture. Using Create alone will be closest to the current implementation. The relevant formats that can be written using Create are: BMP, ENVI (.hdr), GeoTIFF, Erdas Imagine Images (.img), and many others.
• Merged the v6.0 branch back into the trunk on the repository.
• Cleaned up documentation branch from 5.1. Merged into trunk and branch deleted.
• Replaced NAnt and FLEL Dev kit files with IDE project and solution files. Took more time than expected but many lessons learned.
• Assessment of SVN box in Madison completed.

Students

• Two M.S. Comp Sci students hired and they are currently undergoing training. Computers set up, keys to the lab requested.

Lesson learned

• Decision to move to IDE is very good (Jimm's opinion): overall streamlining of operations. Rob has not yet tested.
• GDAL integration is doable. No major barriers foreseen.
• File formats (write) are sufficient for LANDIS-II purposes. GDAL can read just about anything.