On Sun, 17 Jul 2005, Earl Hood wrote:
Comments/questions:
* The term character should not be used. Use octet. Character has
specific meaning for those that deal with character encodings (e.g.
UTF-8, ISO-2022-JP).
For example, replace "octet characters" with "octets".
Ok, I'll do so and replace with octet.
* Why is step (2) needed?
In case line is longer then 998 characters (which it really should not
be for text data). Splitting such lines at 998 octet is what some
programs might do.
* Why does (4) only apply to the beginning of the body part and not
the end? I know of real-world cases where dropping LWSP at the
end of the body part is needed.
Can I ask which these are?
I'm not aware of any concerns about the beginning of a body part.
Wen message passes through mail list sometimes extra CRLF in the
beginning is added.
It seems (4) could be stated more concisely as follows:
All LWSP at the beginning and end of the body part is dropped.
Yes, that should not not compromise meaning of the data and if its a
problem and LWSP at the end are removed by some programs doing data
transmission, this is easy fix.
BTW - there is at least one bigger problem with Content-Digest-EDigest draft
(somewhat related to above). For EDigest I forgot to add CRLF to separate
data for multiple parts when its all added together. If I remove CRLF at
the end of any data part, then possibly that one CRLF between parts would
have to be double CRLF.
--
William Leibzon
Elan Networks
william(_at_)elan(_dot_)net