I do wonder if maybe it would make sense to revise the verification system to be based off of a list of allowed/forbidden functions that may be called during a command.

e.g. for open the requirements are actually more complicated than this diff establishes. It should be: CMD_OPEN must result in exactly 1 opened() call OR exactly 1 error() call AND must not ever call finished(). This is a less specific variant of what m_finalityCommand represents. Where m_finalityCommand simply marks a command as allowed/forbidden to call error and finished, a more dynamic system would allow us to tightly control different calls.

Simply put: Any given command has 0:N functions it may or may not call. And currently we have no system to codify that.

So maybe something like this would make sense:

void validateCall(QString name)
{
q_assert(m_CurrentlyAllowedCalls.include?(name), "mustn't call this function");
}

void opened()
{
validateCall("opened");
...
}

void finished()
{
validateCall("finished");
...
}

[spam comment removed by sysadmin]

I've actually started work on a more general system for this. Will update soon.