Closed mephinet closed 8 years ago
It looks like this is having problems when URI
isn't installed. It might be better to inline a package with an overload that you can use right in the test.
Yes, you are right - the final version of the unit test should not use none-core modules - I'll check if I find a better example in core that shows the same behavior... Furthermore, we need to check whether OverloadedStringify is available in all versions of Perl that you support, and decide whether you are fine with using it given that it's not documented in the overload POD...
As suggested, I replaced the dependency to URI with an inline package.
Excellent! Thank you. I will push out a dev release for CPAN Testers right now, and announce people to test it tomorrow.
Any idea why your release wasn't picked up by the CPAN testers yet? http://matrix.cpantesters.org/?dist=Log-Any+1.043-TRIAL
Because cpantesters has been catching up from a 2-week stint of downtime. The "fast matrix" uses the incoming Metabase directly: http://fast-matrix.cpantesters.org/?dist=Log-Any%201.043-TRIAL
Ah, I wasn't aware of that (2 weeks on vacation B-) - so all good from this point of view, great!
1.045 (stable) is now released to CPAN. Thanks for all your help!
This patch uses
overload::OverloadedStringify
to detect whether a reference has its own stringification - cf https://metacpan.org/source/RJBS/perl-5.24.0/lib/overload.pm#L88 If it has, it will be used instead of using Data::Dumper.This patch would fix #48.