Toolbox does not correctly handle TLC's -continue flag

[lemmy]

The -continue flag causes TLC to keep exploring the state graph even if a counterexample is found. As a result, TLC can find N counterexamples instead of just one. This confuses the Toolbox which expects exactly one counterexample.

I do not think "-continue" is used widely enough to warrant a Toolbox fix. If the community feels strongly about it, please step up and provide a patch/pull request.

[lemmy]

LL: "I'm not sure what should happen if the user wants to use -continue from the Toolbox. As I recall, TLC just keeps running with that option. I don't see doing that from the Toolbox being useful. For using with the Toolbox, TLC should be stopped when an error is reported and the user allowed to then tell the Toolbox to resume running the model from where it left off. Implementing that would be difficult--especially if two different worker threads concurrently discover an error. (But that situation must be handled now, so maybe that's not a problem.) Anyway, we don't have to worry about that now."