3 ms·
To answer "why not?": Because, like a lot of things I've tried to pilot at work, management doesn't see the value when compared against other techniques and the
by Jtsummers 5y ago
To answer "why not?": Because, like a lot of things I've tried to pilot at work, management doesn't see the value when compared against other techniques and the cost of training and maintaining the skillset.
I used it on a system to show:
1. Where a problem was occurring in a concurrent system's design, and how to fix it by changing the protocol the two processes used to communicate.
2. Where a problem was occurring in the data bus of a hardware system. Specifically, which part of the spec had been mis-implemented.
Both took as long or longer to do compared to the traditional techniques of hooking up an oscilloscope and looking at results, writing test-specific programs to exercise things, and writing tests to find other problems. It could have been faster if I weren't a TLA+ novice at the time, I think, but then I'd be the only one who understood it. It also wouldn't have been terribly useful for the majority of the time, think 1-2 problems (issues found or new designs) that it may have applied to. I'm willing to work on my own skills outside of work hours (though the last 3 years I've definitely slowed down on that), but the majority of my coworkers aren't (for various reasons, mostly valid, I was single at the time, most of them were married and many had kids, that's a lot of time that's no longer yours). Management isn't willing (mostly the right thing) to spend time and money on training that doesn't have an obvious payback when other techniques work as well (from their perspective) and don't require any extra training or skill maintenance effort.