Hal Fulton wrote:
James Britt wrote:
I'm reluctant to experiment too much to soon, for fear of losing mail.
As am I.
Is this for a machine that does not currently have a mail server running? Or are you looking to pull mail down to your local machine, processes it, and then pass it to Mozilla?
Right, my local box doesn't have a mail server (at least not that I use).
My pop server is on my web provider's box.
I wonder if you can just write the mail to the mail files directly?
Or run a local "mail" proxy so that Mozilla thinks it is fetching mail, but is really just getting passed files from disk.
Whatever's easier.
If you have further thoughts, let me know...
I have full access to my mail server, and had two motivations o try to address spam n the server, rather than on the client. One was to reduce the amount of mail that gets downloaded, so that the good mail arrives faster, and Thunderbird (and hence my poor laptop) spends less time processing crap. This is more valuable when I am traveling and have a dubious Internet connection.
The other reason was that Thunderbird, nice as it is, has a stupid bug where it simply stops fetching mail for a given account if it finds a mail item with screwed up headers. And it may stop fetching mail and not even give a warning. 
So I set about getting SpamAsasin to play nice with QMail, which was not so hard, and wrote some scripts to inspect mail. The cron job script could likely be useful either on the server or on the client. It takes a set of mail names and passwords, and a set of spam words and phrases, and checks the mail for each account. If a mail item contains any of the spam phrases, I write the message to a side folder on the server and delete it from the mail spool. It never reaches my home machine. I periodically check that folder to see what is going in there, and delete the obvious crap (so far, all of it has been crap. No one I know is trying to give me a Rolex and Viagra and take me to a casino).
This is something you could just as easily do on the client. As for actually filtering spam, I run SpamAssassin on the client. I hope to move that to the server at some point though. Client-side Spamassasin (on Win2k at least) runs as a local mail proxy; you tell the mail client to log into SpamAssassin, and SA in turn goes and fetches the mail, examines it, and passes it through; the bad mail gets munged so that the client can better deal with it.
Offhand, I don't see this as being hard to emulate with Ruby (someone may have already done this). I *think* all the proxy needs to do is pass back a series of mail messages. I don't know if they should come back as one big string, or if there is some protocol for passing back a series single messages while indicating if there are more to come. Worth a look perhaps. RCF time.
I may try this myself, or first see what Gurgitate-mail does (thanks, Fred!)
James
