KA10: Vidissector

[larsbrinkhoff]

The vidissector is a video input device for the AI lab PDP-10. (Actually, there were two or three hardware iterations.)

I'd like to implement this with input from a host "webcam". Or whatever the camera device on a typical laptop is called.

@rcornwell, @markpizz, there may not be any cross-platform API to do this. What are your thoughts?

[larsbrinkhoff]

I'd like to look into OpenCV: https://opencv.org/

[rcornwell]

For this and the other video device I am leaving it up to you to implement. When you have it working we can talk to Mark about moving it into main simH tree if needed.

[markpizz]

Well OpenCV might be OK, but there will probably be a complicated set of build dependencies.
On *nix hosts the makefile, as always, can detect and use what is available. On Windows, adding dependencies is dramatically more complicated. The complexity on Windows is worse by the fact that all the Windows binaries are statically linked. Static linking allows easy binary distribution packaging (which is common on Windows) without having long chains of related dependencies that need to be also installed. The windows building process pre-gathers the 'supported' dependencies into a repo which gets dropped 'next' to the simh directory tree. This allows the developer(s) to work out the complexities one time without it being a user barrier to adoption.