[Top] [All Lists]

Re: Definition of 'Body Part' in MIME draft, part 1

1995-04-18 11:35:40
I know those definitions very well. They were indeed in RFC1521, and they
were already in RFC1341. At some time during the preparation of what
would become RFC1341, the terminology used was quite unclear, so I did
write and propose the four definitions (message, body part, entity and
body) to the working group, who accepted them. I then made a pass on the
document to straighten the terminology, and submitted the corrections to
Nathaniel who applied them to the text. So, please consider that I really
have a full understanding of those definitions and their implications, since
I wrote them in the first place. So I feel entitled to ask you to read this
message carefully.

It does not help your case to insult me. I read your original message quite
carefully. I read all such messages quite carefully.

The definition of 'body part' was introduced to have a term to designate
one part of a multipart body without confusion.

Disagree. A body part doesn't have to be part of a multipart. It can be either
a part of a multpart or it can be the content headers as well as the actual
contents of a message that contains a single part.

There are places were one
must speak about either a message or a body part: that's why the definition
of 'entity' has been introduced. An entity is either a message or a body part.

Sure, but this is quite different from a body part, and is used quite

Then, the definition of body makes reference to 'entity', so the term 'body',
without 'part', is used to designate the body of either a message or of a
body part. Those definitions were thus designed to have one word to designate
each kind of object without uncertainty. BTW, that's the purpose of 
in any discipline, and changing a definition after it has been used as a base
for development for five years is risky.

Please. I understand this quite well.

Read the definition again. It says a body part can be a *part*
of a multipart entity. Nowhere does it say its an entity.

The 'new' definition says that a body part is 'either a single part message or
one of the parts...'. That means that it can now also be a message.

No it can't! Once again you haven't read what the definition says! Read it
again! A single part message is not the same as a message in general!

One thing
wrong with this, is that the text speaks of what makes sense in the header
of a body part (only Content-xxx lines make sense), while in a message all
the header lines defined in RFC822 and successors do make sense.

OK. This is a valid point. I will try to clarify the definition so that it
makes it clear it isn't talking about the entire RFC822, but only the content
headers and actual content of a single part message.

This is also incorrect. The old definition specified that a "body part" was
always a piece of a large multipart structure. And this is wrong, plain and
simple, since MIME messages can and often do consist of a single "body part"
that is *not* enclosed inside of a multipart entity.

Of course, the body of many messages is of a 'discrete' type. This does not
make such messages 'body parts'.

I'm sorry, but that's exactly what it does.

What are they ? Messages, of course.

They are both messages and body parts.

means that they have a (message) header and a body. They may also be called
an entity, and everything that is written about entities applies to them.
But they cannot be called body parts.

I'm sorry, but they most certainly can be. Indeed, they must be, since we
establish rules for body parts that must apply to the single part message

Look at their header, for example.
You will find a 'From:' line, which is mandatory in messages, but does not
make sense in body part, etc.

So what? You are generalizing about messages, whereas the definition clearly
talks about a very specific subcase of the more general case.

There is really no reason to change the
definition of body part, which was designed to speak about one part of
a multipart, to now also designate some kind of message.

This is absolutely wrong. The previous definition tied body part to multipart
structures only. This is not acceptable.

Nothing is gained,
since a message can be called 'message' (or entity if the properties common
to messages and body parts are to be applied). And there is no place in
the specification were things must be said about messages with a discrete
type body that does not apply to a general message (that's one of the beauties
of MIME, that it is completely recursive, and that each occurence of a
'message' may be again a multipart). Much is lost, since with the
'new' definition one does not have any word anymore to designate a part
of a multipart body when properties unique to those body part must be

That's exactly the problem. The "rules" for parts of a multipart also have
to apply to messages that only contain a single part. Under the old definition
they did not. For the third time, this is not acceptable.

And every use of the term
"body part" corresponds to the new definition, not the old one, so

Really, no.

Really, yes. I checked the text on this before I said it.

 I did read the documents once again, and each and any occurence
of 'body part' is there to speak about properties unique to the parts of
a multipart body.

What can I say other than you are 100% incorrect here? I'm sorry to have to be
so blunt, but I don't know how else I can put it.

What is true (of course), is that the term 'body' (without
the 'part' qualifier) is used to speak about the body of either a message
or a body part.

There is no "of course" about it. The old definition was wrong, and caused
major inconsistencies and misinterpretations. The new definition is basically
correct. I concede the point about only the content headers being involved and
will take steps to correc it.

But that's what is has been designed for per the fourth

The problem is that we define rules for body parts that have to apply
to both the multipart and single part cases but don't apply to all entities.

the new definition is going to stay.

I urge you to reconsider this. If the explanations in this message are not
clear/convincing enough, I will produce examples from the texts to show
what I mean in a more illustrated way. Thank you for your kind attention.

I've reconsidered twice now, and I still disagree with you on almost all of
this. I will clarify the definition to make it clear that only the content and
content headers are being referred to in the case of a single part message.
That's as far as I'm going to go.


<Prev in Thread] Current Thread [Next in Thread>