Skip to content

Latest commit

 

History

History
40 lines (32 loc) · 1.81 KB

scoring-software.md

File metadata and controls

40 lines (32 loc) · 1.81 KB

Setting the Bar for Software Scoring

I propose that CIVL set the bar for approving software used for scoring with GAP.

I suggest the following list of requirements:

  1. The program can read a common input format and produce a common output format, currently *.fsdb + *.kml + *.igc in and *.html out.
  2. The code that does the scoring is open source. Other parts of the application can be closed.
  3. There is somewhere to report and review issues with the code, eg. github issues.
  4. There is a way to contribute fixes, eg. pull requests.
  5. The scoring works on windows, linux and osx.
  6. There are build instructions that work on windows, linux and osx.
  7. There is documentation that explains the implementation.
  8. There are releases with release notes and change logs and the source is tagged at release points.
  9. The code that does the scoring can be run in isolation, apart from the application as a whole, in batch mode from the command line.
  10. Working steps can be written to a file format that is human readable.
  11. There are tests.
  12. The build and tests are run on a public continuous integration service.

Points 1., 9. and 10. allow for a way to run the scoring program from the command line (headless), reading common inputs, writing a common output and allowing for scrutiny of the output and workings by humans.

Point 2. software that is open source allows for more scrutiny and is fairer to pilots as nothing is hidden.

Points 5. & 6. makes an expectation that the software builds and runs on the three most common operating systems around today.

Points 3., 4., 8., 11. and 12. are needed for onboarding and working with contributors and scrutineers.

We cannot yet produce a set of automated tests that allow CIVL to pass or fail an implementation of GAP but I think that is a worthy and reasonable goal.