CC: | ALEXBIO [...] cpan.org |
Subject: | Subtle changes between 3.17 and 3.22 break testsuite of Acme-Compress-0.04 |
It started with that I observed weird results of the Acme::Compress
testsuite among the cpantesters.
With T:H 3.17 we have only PASS, with 3.21 and 3.22 we have mixed
results. My observations on my own machine with 3.21+ are that when a
working version of Acme::Compress is already installed, then testing it
yields a PASS.
So if you want to reproduce the interesting results make sure that
Acme::Compress is not already installed on your testing perl.
Bisect on the Test::Harness repository doesn't lead to a clear finding
because there were so many different types of fail since 3.17 wrt
Acme::Compress.
Inspecting the code what Acme::Compress is doing is a bit annoying to
debug because you only see encrypted source code and everything happens
at compile time. That makes debugging not trivial. But my impression is
that Acme::Compress isn't doing anything wrong.
I would welcome more eyeballs looking into this.
Thanks && Regards,