permutation tests

Hello,
I am interested in your opinion about the usefulness of permutation
tests (if I interpreted them correctly):
If I do a permutation test I only test for independence but not that
the parameter is different from 0 (e.g. If I do a premutation test
for a rank correlation using spearman's correlation coefficient rho,
I do no test that rho is different from 0 but I use rho to test for
independence). Therefore, a permutation test is not really testing
what I want to
test. The result of a permutation test seems to be rather useless to
me and a bootstrap confidence interval (or a test based on the limits
of a confidence intervals) seems to be more appropiate. Did I
misunderstood something?
Best wishes, Felix

.