Re: [Nmh-workers] corruption when attaching all-null files

[Paul Fox]

david wrote:
 > Paul F. wrote:
 > 
 > > aside from the logic aspect:  i take it that text with a NUL must
 > > be treated differently than text with an 8th bit set.  i assume it
 > > must be encoded with q-p?  (or base64, but q-p would be sufficient?)
 > 
 > Right, NUL is binary, not 8-bit, data.  It can be encoded with
 > q-p, according to RFC 2045 §6.7.  If the text has a lot of binary

okay, good.  after getting application files with NULs to work (which
fixed all the problems with attaching one or more NUL-filled files),
i took a look at text files with embedded NULs.  the q-p encoder
uses fgets, so it didn't work so well, but that's fixed now too.

while i'm on a roll, anyone have another favorite place where fgets
should be replaced?

 > data, it wouldn't be very readable as q-p, so base64 might be a
 > better choice.

i don't see that kind of heuristic being made anywhere currently.  does
that sound right?

paul
----------------------
 paul fox, address@hidden (arlington, ma, where it's 56.8 degrees)