On Tuesday, August 15, 2006, 6:09:23 AM, Christian Larsen wrote:

Hi,

I've made a simple C commandline program to query multi.surbl.org

I'd be very happy to receive comments. The program is available from http://surblhost.sourceforge.net (source only).

Thanks!

Thanks much for writing this Christian! Hope you don't mind some feedback that applies equally to any application using SURBL data:

1. It should not check common, whitehat domains as that results in potentially very many unnecessary DNS queries. For example SpamAssassin excludes from checking commonly appearing domains like google.com, yahoo.com, w3.org, etc.:

http://spamassassin.apache.org/full/3.1.x/dist/rules/25_uribl.cf

This list should be ok to hardcode.

2. It should reduce country code TLDs to 2 or 3 levels depending on whether it's in this table:

http://spamcheck.freeapp.net/two-level-tlds

In practice this is not nearly as important as #1 since most spam domains are in .com, .net, .org, .info, .biz, etc. Also be sure not to do excessive web downloads of that file. It probably should be hard coded too.

3. The WS list is not strictly part of SpamAssassin, it's Bill Stearns'. The JP list is the work of both Joe Wein and Raymond at Prolocation; it's more than jwSpamSpy:

http://www.surbl.org/lists.html

These points are covered in the implementation guidelines:

http://www.surbl.org/implementation.html

Let us know if you're able to make some of these revisions, then we'll add links to your Sourceforge page from ours.

Cheers,

Jeff C. -- Don't harm innocent bystanders.