Re: DIME Use Case?

On Sun, 14 Apr 2002 11:16:00 -0700, in soap you wrote:

>Simon Fell <[email protected]> writes:
>> On Sun, 14 Apr 2002 08:28:48 -0700, in soap you wrote:
>> >It would be nearly as straightforward to define and implement an
>> >analagous protocol with MIME replacing DIME. Again, what value is
>> >DIME adding to the picture?
>> Simplicity of implementation, performance.
>I don't find the performance claim very convincing at all.
>I'd be extremely surprised if the overhead of MIME parsing
>was substantial in any real system.
>
>> No one has defined a subset
>> or profile of MIME which gets anywhere near the simplicity of DIME.
>Perhaps, but it's certainly possible to do so. That said, I'm not sure
>that simplicity is the only virtue. MIME is much more powerful and
>flexible than DIME. In my experience, the approach of defining a
>simple protocol that only services some immediate need often creates
>massive headaches when it is subsequently discovered that there were
>other requirements that were not incorporated into the original
>design.
>
>> In particular boundary management in MIME is a PIA, AIUI the boundary
>> searching can be simplified by the use of a content-length header in
>> the part, but then what's the MIME equivalent of chunking ?
>So, you're saying that it's worth inventing an entirely new
>packaging protocol to avoid writing the relatively trivial
>amount of code required to handle MIME boundaries?

No, I'm saying its worth it, to remove the runtime penalties
associated with it. [both on the generation and parsing side]

Cheers
Simon