On 21 February 2017 at 08:19, Leon Timmermans <fawaka@gmail.com> wrote: > I think it would be helpful to explicitly mention that the whole «encode if > is_utf8» is essentially always wrong. And to link to a longer explanation in > perlunisomething of why it is wrong. This idiom appears in Perl core modules like this: https://metacpan.org/source/SHAY/perl-5.24.1/cpan/Archive-Tar/lib/Archive/Tar.pm#L1501 https://metacpan.org/source/SHAY/perl-5.24.1/cpan/JSON-PP/lib/JSON/PP.pm#L1446 as well as CPAN modules like this: https://metacpan.org/source/LEEJO/CGI-4.35/lib/CGI/Util.pm#L235 If this is considered wrong, perhaps a useful first step would be to remove it from the core modules and file bug reports against the erroneous CPAN modules such as CGI above, or this one: https://metacpan.org/source/KENTNL/Test-CPAN-Changes-ReallyStrict-1.000003/lib/Test/CPAN/Changes/ReallyStrict/Object.pm#L249 Briefly glancing through this list, http://grep.cpan.me/?q=utf8%3A%3Ais_utf8 I don't see evidence for the statement about "broken XS modules" which was used to justify this change, since none of the above seems to be dealing with outputs from XS modules.Thread Previous | Thread Next