On Thu, Feb 14, 2008 at 03:00:41PM +1100, Morgan Weetman wrote:
> Hi Robert,
>
> I am a systems engineer from Australia and I have been using
> policyd-weight for several months on my internet facing mail relays, the
> performance of the servers is vastly improved.
>
> I read your announcement regarding stopping development and wanted
> to assist if possible, I have not read through the code to gain a full
> understanding but perl is my language of choice. Please let me know if /
> how I can help you out,
The major problem of policyd-weight is, that it is somewhat
non-deterministic - at least from "not-familiar-with-the-code"-point of
view.
Because of this it does also lack a good documentation which
expresses the constraints of each check and their different
results such, that it would make users able to find the right
knob to adjust -- easily, if ever.
However, if you want to take over development and maintainership
you are welcome to also run the policyd-weight.org web service.
Also you should sign on the mailing list and tell others
that you want to continue, and what your plans and _your_ future
philosophy with policyd-weight are.
_My_ current philosophy was:
- as less modules as possible
- low latency checks first
- as many reliable short circuit decissions as possible
Next plan for 0.1.15 was:
- make policyd-weight multi-protocol
multiplex capable (DNS/Policy Proto via the IO::Select module)
- provide parallel RBL lookups under some situations.
- configurable scoring of RBL returns (127.0.0.1, 127.0.0.2, etc)
- use client_name (at the right places) before asking DNS
If have also some patches laying around which I didn't have the time
to verify/adjust/correct/include yet.
--
Robert Felber (PGP: 896CF30B)
Munich, Germany
____________________________________________________________
Policyd-weight Mailinglist - http://www.policyd-weight.org/
