On Wed, Feb 27, 2013 at 10:15 PM, Bill Moseley <[email protected]> wrote:
> I had forgot about this until today until I saw errors in some new code.
>
> Is there anything below that needs clarification? IIRC, I got stuck
> because I didn't really follow what the tests were attempting to test.
> Does the test need to do anything more than make sure wide characters get
> turned into octets when freezing and then when thawed it's character data
> again (i.e. the utf8 flag survives round trip)?
>
> Thanks,
Looking at the JSON::Any code this *may* be a bug with that. At some
point in time (someone) had to add the line:
$conf->{utf8} = !$conf->{utf8}
for both the JSON and JSON::XS 2.x backends (they're handled
differently in 1.x). Can you remove that line and make sure that you
still see the issue otherwise this is a JSON::Any bug and I'll need to
investigate why that flag was neccessary (again) and whether we need
better resolution on when to apply that flag.
-Chris
