Asterias wiki pages
Asterias would gladly accept new developers and contributors.
Short and midterm
The following is an incomplete list of features/additions where people could immediately contribute:
- SignS: a couple of new methods, and speeding up the threshold gradient\ descent code. Use C/C++ and improve parallelization.
- In most R code (specially SignS and ADaCGH): parallelize figure creation.
- All calls from CGI to the rest of the code: how to allow use of job schedulers (e.g., SGE)?
- Virtualization: would allow easier testing, increased security and download of pre-canned "distributions" for both real multinode and single node cluster; Linux VServer and/or Xen look like appropraite candidates with low overhead.
- Security: all of Asterias. Throughly test and poke Asterias for security holes and provide patches.
- Python Web Application Frameworks. We have started changing the code of applications to use Pylons.
- Parallelization et al.: allow use of other MPI frameworks (MPICH, OpenMPI), improved usage of clusters of SMP machines (Unified Parallel C, MPI + OpenMP) and NetworkSpaces with Sleigh (note that the later can be used with both Python and R).
Open design and achitecture issues
- Alternatives to having a server per node? What do we gain from having a (pair of) centralized web server node which distributes jobs to other nodes?
- Grid: make Asterias capable of using grids.
- Result annotation via a notebook-like interface that allows for distributed editing.
- Turn into web services.
- Pre-canned binnaries (e.g., Debian, Ubuntu, FC).