I would be incredulous if using the wrong "make" lead to just a single test step failing. Especially as that is the same test step that fails in automated tests.

(Update: Oops. I see now that those two things are distinct issues here. Note that you shouldn't have to "use make" to get the module to install, just read up on whatever tool you are using to figure out how to force it to install even though the unit test results weren't perfect. Also, which 'make' to use can be determined by running "perl -V:make".)

Humorously, if you look at the differences between v0.03 and v0.05, you'll see this:

+TEST "shouldn't be able to slurp failed pipe";
+
+eval{slurp "-|", "non-existent_prog"}
+       ? ok 0, $desc
+       : like $@,
+                  qr/^Can't open '-|non-existent_prog'/,
+                  $desc;
[download]

Yeah, I really doubt v0.05 in fact works any less well on Windows than v0.03 did. :)