Programmers aren't so bad. There is a paper coming from the Berkeley University, which mathematically proves that a multithreaded program cannot be thoroughly tested, so after a very strict testing procedure, there can still be deadlocks or other synchronization problems in the program.
I am just working on a framework which would practically hides the need of mutexes and other synchronization primitives, and tries to provide a "fool-proof" parallel programming framework. Do you think that making it an open-source project (eg. on SourceForge) could be useful? I am not sure whether a new paradigm which does not come from a software giant could gain any attention...
Slashdot Mirror
Programmers aren't so bad. There is a paper coming from the Berkeley University, which mathematically proves that a multithreaded program cannot be thoroughly tested, so after a very strict testing procedure, there can still be deadlocks or other synchronization problems in the program. I am just working on a framework which would practically hides the need of mutexes and other synchronization primitives, and tries to provide a "fool-proof" parallel programming framework. Do you think that making it an open-source project (eg. on SourceForge) could be useful? I am not sure whether a new paradigm which does not come from a software giant could gain any attention...