Expand description
TLC’s error classes for another engine’s error endings.
The Core execution engine finds
where an evaluation error ends a search and reproduces the error with the
evaluator (dive-eval). It then classifies the error by its site with
this crate’s table (class.rs), so that Core reports TLC’s class, message
and exit. These are thin public views of that table;
the rules and their evidence are documented there.
Functions§
- assumption_
error - An
ASSUMEthat fails to evaluate, as TLC classifies it (TLC_ASSUMPTION_EVALUATION_ERROR, or the error’s own class).messageis the evaluator’s text of the error. - init_
error - An error raised while generating or checking the initial states.
stateis TLC’s text of the initial state being checked, for an error in a check of a state, andNonefor an error while generating them. - initial_
check_ error - Classifies a failed check of an actual initial state, including the bounded diagnostic replay of a pure invariant’s newly constructed value receiver. The replay retains the role’s real scope and all previously computed values; unrelated, effectful or stored-receiver failures gain no source authority.
- recurring
- An invariant that failed to evaluate on a state outside the model, which TLC’s replay checks again: the replay decides.
- replay_
error - The class once TLC’s call-stack replay meets the further error
e2after an invariant failed to evaluate on a new state. - search_
error - An error raised at
sitewhile exploring, as TLC classifies it.nameis the invariant, constraint or property, where there is one.