I went through the test suite and identified places where either the test suite was probably out of date (or working with old documentation) or insserv was giving improper results. Most of the test cases which were failing can, I believe, be addressed by warnings rather than outright failures or blocking. A missing virtually facility, for example, could be provided by another script and results in a warning rather than insserv refusing to add the script to a runlevel. I feel this is the more kind approach.
The end result is running "make check" in the insserv source tree now runs the test suite. If everything goes well (which is does on my machine) the check completes successfully and prints a summary. If an error is encountered, the test suite should bomb out with a warning and leave the test data intact so it can be reviewed. (In the past test data was erased, making it harder to investigate.) The test suite now works with insserv's new output directory location (/var/lib/insserv) so that's good news. Running "make distclean" cleans up any test data. I think this bug can probably be marked as fixed upstream. I hope to publish a new beta around the end of May and a new stable release of insserv in June. With that release of insserv-1.20.0 we should be able to properly close this bug. - Jesse
