SDP: Runtime Algorithm

[joelmartinez]

Memory usage and IO performance are areas that mdoc has opportunities for improvement. Currently, each XML file is read and written multiple times when there are multiple "frameworks" involved. Additionally, each framework is processed serially, which means that the same type must be processed separately from all other instances, which makes determining things like alternate parameter names and other differences difficult.

Goals

This project's goals are to come up with an algorithm for API discovery that can handle the rigors of existing workloads, and be able to scale to future increases.

• Customization options:
  • Moniker, instead of framework
  • Option to use string other than "To Be added." for empty elements
• Optimizations
  • Write grouping, to avoid multiple reads/writes of the same file
  • Parallelization (? if possible)
  • Should be able to process an API the size of the BCL