Re: MIME vs. DIME vs. ...

>> While creating a "safe" string can be done at some cost,
>I'm not sure what you mean by "some cost". We're talking units of time
>that are unmeasurably small here, on the order of microseconds.
>ARC4 (which is fine for this purpose) can do 45 MB/s on even quite
>slow machines.

I was merely making the relative comparison to using an integer
indicating the number of octets.

>Again, I'd be interested in seeing performance numbers. In my
>experience, this sort of low-level protocol parsing is rarely
>a bottleneck in real world systems.

It is not just parsing but figuring out what amount of memory to
allocate, estimate transmission timeouts etc. Having a readily available
length indicator up front is really nice to have for lots of reasons.
However, if your point is that this is not in it self sufficient to
invent another wrapper then I would tend to agree... but it isn't.

While MIME provides an open-ended extensibility model, a main point of
DIME is *not* to do that but to stay small. Instead, it allows one to
use whatever open-ended description framework one wishes as part of the
DIME payload, be it XML, or whatever.

Henrik