Bug #7175: unpack('M*') changed behavior between patchlevels

• Status: Third Party's Issue
• Assignee: -
• Target version: -
• ruby -v: ruby 1.9.3p194 (2012-04-20 revision 35410) [x86_64-darwin11.3.0]

[briantobin (Brian Tobin)]

I'm seeing an important difference in the output of string.unpack between two patchlevels of ruby 1.9.3. The output of patchlevel 194 is expected, and the output of 286 is unexpected.

--- Console 1 ---

RUBY_VERSION
=> "1.9.3"
RUBY_PATCHLEVEL
=> 194
"foo =3D =".unpack("M*")
=> ["foo = "]

--- Console 2 ---

RUBY_VERSION
=> "1.9.3"
RUBY_PATCHLEVEL
=> 286
"foo =3D =".unpack("M*")
=> ["foo = ="]

Was this a purposeful change? Our code is processing emails with the "quoted printable" encoding. After upgrading to patchlevel 286 we are getting trailing equals signs after decoding. I feel like this is a regression.

Thanks in advance!

Brian

[briantobin (Brian Tobin)]

After reading the changelog I see that this changed b/c of bug #5635 (http://bugs.ruby-lang.org/issues/5635).

I still believe that this might be a regression. Looking at our data, the last lines of quoted-printable strings produced by Microsoft products generally have an "=" and no newline.

Reading RFC2045 (http://www.ietf.org/rfc/rfc2045.txt), page 21, item #2, I see that this is invalid. But in practice I'm seeing that it's common.

[briantobin (Brian Tobin)]

Second update

I discovered that the Mail gem (https://github.com/mikel/mail) is incorrectly stripping the newline after the "=", which makes it illegal. I'll file a bug there.

[briantobin (Brian Tobin)]

FWIW, bug ticket there: https://github.com/mikel/mail/issues/440

[drbrain (Eric Hodel)]

• Status changed from Open to Third Party's Issue