I wouldn't think that failing to run 'make' prior to 'make test'
would be the problem (make checks and builds target dependencies after all).
Even the install docs for this module just say: 'perl Makefile.PL'
then 'make test' then 'make install'. And, in the spirit of
empiricism (and to attempt to reproduce the problem), I just grabbed
a clean version of MIME-tools-5.411a.tar.gz, unpacked it, ran 'perl
Makefile.PL' followed by 'make test' and everything worked just fine.
(on slackware linux with perl 5.6.1).
You say you hit this segfault with *some* other modules too. Can
you detect any commonalities with the ones that segfault at this
particular target versus ones that don't? Being unable to reproduce
the problem I can only guess that: a) your perl
installation/configuration is borked, b) you are hitting some kind of
limit (memory, processes, or command line length (pm_to_blib can
produce lengthy command lines, but that's not usually a problem on
unix and I believe ExtUtils:: already works around that on Win32),
c) your 'make' program is borked, d) something else entirely.