Assertions can be turned on and off on a class-by-class basis, with different levels: preconditions, postconditions, invariants, etc. As a result, it would be tricky and unreliable to report when they are enabled or not (consider, for example, inherited code: the checks might be on in one case and off in another). On a methodological level it would also break the idea that a correct program works the same way regardless of assertion monitoring.
What are workarounds?
If assertions are complex, they can be factored out to dedicated queries and look like
check
is_valid: complex_query
end
An alternative is to use debug statements:
debug ("check_this", "check_that")
... some complex code, including assertions
end
where "check_this"
and "check_that"
are debug keys that can be turned on when compiling for debugging.
There are hacks that could work now, but not in the future:
If a complex state needs to be computed and then checked after some operation, it can be saved in an object passed to some function with complex calculations and used later again:
check
is_valid_before: valid_pre (state) -- The state is computed by `valid_pre`.
end
code_that_does_the_work
check
is_valid_after: valid_post (state) -- The state is checked by `valid_post`.
end
Some global flag can be used to keep track about assertion monitoring:
check
is_monitoring_checks
end
where query is_monitoring_checks
has side effects:
is_monitoring_checks: BOOLEAN
-- Record whether assertion checks are turned on.
do
is_check_on := True
Result := True
end
Then, subsequent code could be written as asked in the question:
if is_check_on then
... -- Do some complex calculations when assertions are turned on.
end