Steve Dorner <sdorner(_at_)qualcomm(_dot_)com> writes:
I looked back at that discussion. Perhaps I'm missing part of it. The
last large chunk I find is this (there are a couple of minor changes made
after, but nothing very substantive):
The two messages in the archive immediately after that message are
important, they dink with the definition of body-part, encapsulation,
and delimiter in order to deal with the requirement in the text that
LWSP be ignored at the ends of delimiter lines.
The first message following had a comment by me that the text of the
specification had an implication that LWSP had to be ignored in
certain places, particularly at the ends of delimiter lines and that
this requirement was not reflected in the grammar. (The requirement
in the text has been in MIME since 1341.) I suggested some grammar
changes to deal with the requirement.
The second message following has a new grammar from Ned, incorporating
some of my ideas.
And that's the final statement I have on boundaries--that delimiters are
NOT allowed to have an arbitrary amount of whitespace between the boundary
and the CRLF.
They are allowed to have an arbitrary amount of whitespace per the
grammar in Message-Id: <01HIP67ESB8W9VUJQ4(_at_)INNOSOFT(_dot_)COM>. (That
grammar was modified later to change the use of "[linear-white-space]"
to "*LWSP-char")
[I don't know what "Again, LWSP..." means. The others all say "No LWSP".
Typo? SD]
That's also covered in the discussion.
The comments on various nonterminals in the previous multipart
grammars explicitly prohibited whitespace between certain tokens, as
if RFC 822's free insertion of linear-white-space in structured field
bodies would have otherwise permitted it there.
I pointed out that these nonterminals were *not* in the context of a
structured field body, so free insertion of linear-white-space was not
allowed. (Good thing too, RFC 822 comments and header continuation
lines would be a bitch to parse there.) The spec was clarified to
make this "free insertion not allowed in this context" interpretation
explicitly stated and to remove the misleading "no LWSP between"
directives in the nonterminals' comments.
--
_.John G. Myers Internet: jgm+(_at_)CMU(_dot_)EDU
LoseNet: ...!seismo!ihnp4!wiscvm.wisc.edu!give!up