I've depended, since the earliest days of Daffodil, on being able to click on
one of the
def test_foo() = { Runner.runOneTest("foo") }
And do "debug test" and have it run that single test in the breakpoint debugger.
I do this dozens of times a day when working on Daffodil using both
I've used dynamic tests in a couple of schema projects I was working on.
My recollection is that the IDE integration made it difficult to run a single
test, as you first need to generate all the tests; and there was not a good way
to cache the results.
In the case of Daffodil, we also take
Skimming the Runner code, I'm not entirely convinced this is all working
as expected. But I think Ian is looking into that as part of this bug.
Seems like a reasonable time to confirm the complexity of our TDML
runner and all the caching is actually working right before updating all
the tests to
The point of runners and the object constructing them, was to share the
XML-loading of the TDML, and computation of all the test suite objects from it,
which was otherwise being loaded repeatedly and computed repeatedly.
It is sensible to question this. There's a bunch of requirements we may