[Vortex] Is vortex assuming that msg numbers on channels start at 1 and increase by 1/

[Robert M. Münch]

Am 25.11.2008, 01:00 Uhr, schrieb Sam Roberts <vieuxtech at gmail.com>:

> We use a version of BEEPy that we rewrote so that it works. It's in
> python and works with the twisted protocol infrastructure.
>
> We use beep4j, its in java, and works with MINA.
>
> Both of those are event-driven rather than multi-threaded, btw.

Hi, cool, thanks for the info.

> I'm also experimenting with using beepcore-c by binding it into Lua.

What's your experience with it so far? Is it a stable lib or does it need a lot of work to make it useable?

> Its main difference from vortex is that its not multi-threaded, and
> that it has a kernel of code (the "core") that processes beep protocol
> data while making no networking calls, and has close to no library
> dependencies.

This sounds interesting. Following a clear and strong seperation make sense IMO.

> Besides interop, a focus of my experiment is performance. BEEP seems
> to significantly underperform compared to raw TCP, we need this fixed
> (we download large pdfs over BEEP, as well as sending small messages
> back and forth).

Do you have any benchmark numbers? I expect that BEEP is a bit slower than raw TCP because of all the housekeeping stuff. For me the tradeoff is how much speed I loose while gaining a lot of reliable infrastructure that I don't have to care about. And so far BEEP has really been a good choice.

-- 
Robert M. Münch
Mobile: +49 (177) 245 2802
http://www.robertmuench.de