"cmulcahy" <[EMAIL PROTECTED]> writes:

> Jason, I am thinking about hacking TMDA to function as a long-lived,
> multi-threaded daemon ( UID vpopmail ) listening on a unix-socket
> for calls from light-weight C-Clients that are themselves targets of
> dot-qmail execution. I have had great success with this sort of
> architecture in the past.  The intention is to (dramatically) lessen
> the server-load generated by constant invocation of the
> python-interpreter. Have you considered this approach yourself.

I agree that in busy environments, TMDA's current architecture might
not be able to scale.  When I initially wrote TMDA, I had no idea it
would be so widely adopted, so I didn't think "big" from the start.  I
think something like what you describe above will eventually be
necessary for TMDA if it's to be used in large multi-user
environments.

Have you looked at ReadyExec? (http://readyexec.sourceforge.net/).
Someone suggested awhile back that this might be an easier way to
achieve this goal than rearchitecting and rewriting large portions of
TMDA.

> What caveats or pitfalls do you see?

It's hard to guess what the full impact of such a change might be, but
I suspect the very dynamic nature of TMDA's configuration interface
would be affected.
_________________________________________________
tmda-workers mailing list ([EMAIL PROTECTED])
http://tmda.net/lists/listinfo/tmda-workers

Reply via email to