Hi,
I've noticed a couple of minor annoyances related to test running as
I've been building my first buildfile for a largish project. I wonder
if anyone has any workarounds for them.
The project I'm working on has 5 modules with java code, plus a
separate integration test module. When I'm trying to get something to
work in module 4 (which depends on 1, 2, and 3), it's time consuming
to wait for the tests for 1, 2, and 3 to run every time I change the
buildfile. Don't get me wrong -- I think that the general idea of
both tracking the last successful test run and invalidating it when
the buildfile changes (or any other relevant dependency) is
excellent. But when I'm just trying to figure out buildr, it's a bit
excessive. Most of the time, I can do test=no and be spared. But
when the thing I'm trying to get to work is the tests for module 4,
test=no isn't an option. Is there a way to turn off the dependency on
the buildfile for tests?
Another solution to that problem would be to have an option to only
run the tests for a single module -- sort of like test:pattern, except
that the constraint would be the module. This would help with another
minor issue I've noticed: when you specify a pattern, all the tests in
all the modules that match that pattern are run, even if you invoke
test (or integration) from a submodule. Pattern execution appears to
ignore the successful test check, too. The case where this has been a
problem for me is running integration tests. I have several
integration tests with "activity" in the filename. I also have a
large number of unit tests with activity in their filenames. So if I
want to run just the integration tests for "activity," I have to wait
for several dozen unit tests, across several other modules, to complete.
The latter issue is less of a problem than the former, but I mention
it since they have a possible common solution. Is there an existing
solution for any of these?
Thanks,
Rhett
- Suppressing retesting when the buildfile changes and/or onl... Rhett Sutphin
-
