[Mailman-Developers] Splitting long header lines and the "bug demonstration"

[Barry A. Warsaw]

I know what is causing

  http://mail.python.org/pipermail-21/mailman-developers/2002-June/012093.html

It's the email package, and specifically it's Generator's behavior
when splitting long lines, i.e. Generator._split_header().

I actually think that method is terminally broken because it has a
ascii bias.  I doubt it would even work for long lines of encoded
text.

The good news is that we have a perfectly fine line splitter that
understands encoded headers and does the RFC-correct thing.  It's
called Header.encode().

The bad news is that Header.encode() isn't parameter aware and making
_split_header() call it will cause some messages to not generate
idempotently.  If your code is expecting the splitting done by
_split_header(), your code will break.

Interestingly enough, only 4 unit tests fail when I make this change.

In particular, it means that a header like

X-Foobar-Spoink-Defrobnit: wasnipoop; giraffes="very-long-necked-animals"; spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"

will not get split like

X-Foobar-Spoink-Defrobnit: wasnipoop; giraffes="very-long-necked-animals";
	spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"

but instead will get split like

X-Foobar-Spoink-Defrobnit: wasnipoop; giraffes="very-long-necked-animals"; spo
 oge="yummy"; hippos="gargantuan"; marshmallows="gooey"

I can obviously fix the test cases, but will this break your code if I
make this change?  It should fix the nasty Mailman problem.  I'm
wondering if I need to make some accomodations for _split_header()'s
old behavior.

-Barry