Re: Byte and char types

Hello.

On Paul W. Cowan wrote:
> Each message we pass contains multiple elements
> who's value is a single character or number less than 255. It may
> seem insignificant, but the extra overhead of 3 or 4 bytes for each
> int or binary format value becomes significant in volume.

Even without protocol extensions if many small integers are to be sent
in single message they can be well packed into a binary element.

However I think that for cases where even more compact representation
is important a good compromise would be adding a compact integer
format that looks like one used for UTF-8 encloding. See for example
http://en.wikipedia.org/wiki/UTF-8. But int his case we do not need
keep compatibility with ASCII and encoding can be simpler.

It could be something like
'i' nnnddddd dddddddd .... (up to 9 elements)
where 3 bits nnn represent number of bytes in the sequence - 1
and remaining bits ddd... contain bits of required integer.

For example: 0xfff => 00101111 11111111

One reasonable variation here - instead of 3 use 4 bits for length
representation thus covering even those cases that are beyond 'long'
datatype.

I think that extra complexity should be added with extra care. But
this single extension to the protocol can cover many cases and
provides reasonable (but not best of course) compression.

--
Petr Gladkikh