Running a standard Perl module's test suite is very easy using prove. If the module needs compilation, just run make before hand and pass the -b flag to prove, otherwise use -l to add lib to @INC.
If you pass --state=save then information about the last run will be saved in a .prove file in the current directory. This data can be used later to run only failed tests, or to reorder tests.
My testing workflow was greatly improved using this feature. I use two shell aliases to run prove:
alias ts="prove --state=slow,save -j3 -r" alias tf="prove --state=failed,save -v -r"
I start by running ts -l t (or ts -b t if this module requires make to run before testing). The -r flag tells it search for .t files recursively in t. The -j option enables 3 concurrent jobs, and the --state flag orders tests from slowest to fastest, and causes the .prove file to be updated at the end of the run.
When I get failing tests I switch to using tf -l or tf -b. Since this has --state=failed it only runs tests that failed in the previous run. The -v causes the TAP output to be printed (with coloring).
I fix my tests or code one by one, and keep running tf. When tf has nothing left to run everything is passing, so I switch back to ts to run the full test suite again.
These two simple aliases let me run the test suite much more easily and effectively than make test, so a great many thanks to Ovid and Andy Armstrong [Update: and Andy Lester] for writing and maintaining these tools.