[PATCH] vm-mime.el:vm-mime-encode-words creates garbled X-VM-v5-Data

[Marcin Koziej]

VM 8.0.12
X-VM-v5-Data is encoded by vm-mime.el:vm-mime-encode-words,
unfortunately, it is not aware of quote marks, so when it encounters
a string:
"\"Wechikuł czasu"\" it encodes it as follows:
"=?utf-8?Q?\"Wehiku=C5=82 czasu\"?="
As You can notice, quote marks are inside, making this string impossible to
evaluate later.

My proposed fix is to change vm-vars.el:vm-mime-encode-headers-words-regexp so it doesn't span over " marks.
(see patch)
After this change, string is encoded as follows:

"\"=?utf-8?Q?Wehiku=C5=82?= czasu\""

[Uday Reddy (reddyuday)]

Ulrich, I have tried this patch, but I don't think it is doing what was claimed. Can you double check? -- Uday

[Ulrich Müller (ulm)]

RFC 2047 says in section 5:
+ An 'encoded-word' MUST NOT appear within a 'quoted-string'.
So I think this patch is wrong.

If I read that RFC correctly, the right encoding for the string
"\"Wehikuł czasu"\" is:
"=?utf-8?Q?=22Wehiku=C5=82_czasu=22?="

Looks like there's more to the problem than simply changing a regexp.
Probably we should use some functions from rfc822.el for proper
parsing (I've checked, this file existed in Emacs 18, so it should be
available everywhere).

[Uday Reddy (reddyuday)]

Actually, I put quotes around your (Ulrich's) name in utf-8 in one of the messages, saved it and loaded it back into VM. The VM-v5-Data header had the quote marks inside the utf-8 bits, and VM seems to have handled it without any problem. So, the problem reported here was probably a consequence of the other VM-v5-Data problems which are now fixed. So, I think we can ignore this bug report unless Marcin comes back with some new information. I am changing the status to "incomplete".