On Wednesday, 24 August 2016 at 06:32:54 UTC, Jacob Carlborg
wrote:
Sure, but one might not figure why a test is failing after only
a single run. One might use "dub test" to run integration tests
that depends on some external dependency that is causing the
test to fail.
I always considered dub test to be only for running unittests and
not really suited for integration tests, but if that is a use
case then fair enough.
There are many reasons to run the test without recompiling.
I might be dense, but the only other thing than integration tests
that I can think of is if you use random data for testing, but
that would be more correctly solved by using more random data
during the unittests. Nothing is worse than tests that only
sometimes fail.
What is the point of recompiling when nothing has changed?
I'd be all for for dub to analyse whether there are any changes,
but AFAIK dub's focus is as a packaging system, not as a build
system. In that case I'd prefer it to ere on the side of caution
and recompile more often in case there might be a change. Maybe
the key would be to replace dub's limited build system by an
external build system such as reggea.
