Hello Eric,

On 12/30/21 20:17, Eric Wong wrote:
> Uwe Kleine-König <uwe@kleine-koenig.org> wrote:
>> Hello,
>>
>> adding the Debian Perl Group to Cc, maybe they can help here.
>> (for context look at https://bugs.debian.org/1002219)
>>
>> On 12/30/21 10:12, Uwe Kleine-König wrote:
>>> I got a bug report against the public-inbox 1.6.1 package about a
>>> failing test, see below for the whole output. I didn't have time yet to
>>> look into it, so this is just a heads up to make you aware. If someone
>>> has a hint what to do, this would be greatly appreciated. Maybe just
>>> updating to 1.7 will help?
>>>
>>> Best regards
>>> Uwe
>>>
>>> On 12/21/21 17:34, Lucas Nussbaum wrote:
>>>>> #   Failed test 'filename decoded'
>>>>> #   at t/eml.t line 407.
>>>>> #          got: '=?utf-8?q?vtpm-makefile.patch?='
>>>>> #     expected: 'vtpm-makefile.patch'
>>>>> # Looks like you failed 1 test of 146.
>>>>> t/eml.t ......................
>>
>> I can reproduce this problem with 1.6.1 checked out. I played a bit around
>> (suffering from my weak perl foo) and found that when I downgrade
>> libemail-mime-perl from 1.952-1 (i.e. Debian unstable's version) to 1.949-1
>> (i.e. Debian stable's version), this works.
>>
>> The reproducer is:
>>
>> $ perl -e 'use Email::MIME; print Email::MIME->new("Content-Type:
>> text/x-patch; name=\"=?utf-8?q?vtpm-fakefile.patch?=\"\nContent-Disposition:
>> attachment; filename=\"=?utf-8?q?vtpm-makefile.patch?=\"\n\n")->filename;'
>>
>> which emits "vtpm-makefile.patch" with 1.949-1 (as public-inbox expects),.
>> but =?utf-8?q?vtpm-makefile.patch?= with 1.952-1.
>>
>> So the key question is: Is the test correct and his is a regression in
>> libemail-mime-perl, or is the test wrong and we need to fix the test (and
>> PublicInbox::Eml)?
> 
> I thought I sent a fix to this; but I nuked the root FS on one of
> my workstations on accident :<  Still recovering...

Oh, good luck in restoring your precious data.

Thanks for your patch. I just uploaded public-inbox with this change to 
fix the bug.

I still wonder if this is a regression in Email::MIME, or just a wrong 
expectation. In the first case I'd open a bug against libemail-mime-perl.

Bisecting in https://github.com/rjbs/Email-MIME.git yields 
https://github.com/rjbs/Email-MIME/commit/8a7cffd2b1091ff1a750d541a85e1813a6747b72 
as breaking commit. So this looks like an intended change.

Best regards
Uwe