I noticed this problem started with the CVS Oort about 1-2 days ago.
I'm subscribed to a mailing list that sends out digests.  When I
receive a digest, I go into the nnml group that contains hit and do a
`gnus-summary-read-document' to expand the digest into its separate
messages.  Starting the other day, when I expand the digest, this is
what my summary buffer looks like:

 . [ 178: Aimee Mann listproc    ] <* mixed> MANNLIST digest 1773
 .   [  12: Aimee Mann listproc    ] <1 text>
 .   [ 156: Aimee Mann listproc    ] <2.* digest>
 .     [  93: Aimee Mann listproc    ] <2.1 text>
 .     [  34: Aimee Mann listproc    ] <2.2 text>
 .     [  22: Aimee Mann listproc    ] <2.3 text>

Notice that the subjects are incorrect (in fact, they look like MIME
types).  The article buffers for the individual messages are also
incorrect.  It seems like something in the parsing code has broken
recently.  I'm attaching one of the digests to this message.

-Emerick
-- 
-------------------------------------------------------------------------
Emerick Rogul             /\/     "how young are you, how old am i?
emerick@cs.bu.edu         /\/      let's count the rings around my eyes."
------------------------------------------------- 'i will dare', the mats