Puzzling doubled-period bug
I've run into a puzzling problem that I can't yet reproduce: our Mailman installation sometimes doubles periods in the HTML portion of a message, sometimes breaking a link or image as a result. Does anyone recall seeing this before?
For example, here's the difference between the HTML before it's sent and after it's received:
-> diff old new 60c60 < ational_Conference.gif" alt=3D"2010 National Conference" width=3D"235" =
ational_Conference..gif" alt=3D"2010 National Conference" width=3D"235" = 105c105 < must-attend conference. </p>
must-attend conference.. </p>
The first hunk is a change to an IMG SRC, so it breaks the display. These are the only two changes in an HTML file that is hundreds of lines long. Not every '.' is doubled, and '.' isn't even special to quoted-printable encoding, so this is very puzzling. Does this description sound even vaguely familiar to anyone?
Searches of the Launchpad bug tracker (and bugs.python.org) didn't turn up anything apparently relevant. I'll be chasing this down, since it might be a bug in Mailman, our customization of Mailman, our spam filtering, or in Python 2.5.0's email package, binascii module, or somewhere like that. (For historical reasons, we're stuck with Python 2.5.0, so possibly this is fixed in 2.5.4).
--amk
A.M. Kuchling wrote:
I've run into a puzzling problem that I can't yet reproduce: our Mailman installation sometimes doubles periods in the HTML portion of a message, sometimes breaking a link or image as a result. Does anyone recall seeing this before?
I've never seen this that I recall.
Look at section 4.5.2 of RFC 2821 (or the same section 4.5.2 of RFC 5321). While this doesn't seem relevant to your example, it's the only thing I can think of that doubles periods.
Possibly, the lines in the message are long and some MTA is breaking them for transmission and a line just happens to get broken so it begins with a period which is doubled, but somehow the receiver doesn't see the line break and doesn't undouble the period.
The fact that this only occurs in HTML is additionally suggestive of long lines being involved.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
On Apr 12, 2010, at 01:50 PM, A.M. Kuchling wrote:
I've run into a puzzling problem that I can't yet reproduce: our Mailman installation sometimes doubles periods in the HTML portion of a message, sometimes breaking a link or image as a result. Does anyone recall seeing this before?
Like Mark, I've never heard of this before. -Barry
participants (3)
-
A.M. Kuchling -
Barry Warsaw -
Mark Sapiro