SpamAssassin with SAProxy

I’m happy to say that spam isn’t usually too much of a problem for me. I get about half a dozen messages per day on my main sunpig address. That’s a manageable volume. Most spam is very easy to spot based on its sender and subject line, and hitting the “delete” key a handful of time just isn’t that much effort.

Nonetheless, it’s annoying that I have to. I try to be careful where I pass out my email address, and even though the sunpig web site features email links (click on my name at the bottom of each post, or on the page banner), they are obfuscated by some custom javascript. I don’t run around signing up for random mailing lists, or buying stuff from companies without opt-in/opt-out mail policies (although they’re not always worth the bits they’re encoded in).

This is where SAProxy comes in. SAProxy is built on SpamAssassin, which is generally regarded as one of the best spam filters around. I had looked at SpamAssassin last year, but I’d been put off by the fact that it ran under Unix (or Linux). And even while I was running Linux on my desktop last year, it seemed confusing and difficult to get set up effectively.

Not any more. SAProxy rolls up the SpamAssassin engine in a nice little package for Windows that sits between your email program and your mail server. Normally your mail client (Outlook, Mozilla Mail, etc.) contacts your mail server directly to download its mail. SAProxy adds a proxy layer into the mix: your mailer speaks to the proxy, and the proxy contacts your mail server. When your mail is downloaded, the proxy runs all of it through the SpamAssassin filters. If it finds any spam, it adds the text “*****SPAM*****” to the subject line of the message. Then all you have to do is tell your mailer to automatically shunt all messages with this text into a junk folder. Every now and then you can quickly run through this folder to make sure that it hasn’t accidentally tagged any non-spam, and bulk delete the rest.

It’s very easy to set up, and it works with most modern mail programs under Windows. I installed it yesterday, and so far it has successfully caught all seven spam messages I received since then. It’s a small sample, but it looks very promising.

Ultimately, I would prefer a server-based solution, because the spam still does get downloaded onto my own computer. If the volume of spam were to increase, this could become a problem. But until then, it’s nice to have a clean inbox again.

Related entries:

Related Links

Mail delays

Blueyonder, our ISP, has been having some mail problems in the last few days:

“The current queue on the secondary MX (Mail eXchanger) for blueyonder.co.uk has grown to approximately 1.2 million e-mails that are queued for delivery to the primary server in Knowsley as well as the business domains that this server acts as a backup MX for.

“Of this 1.2 million we have identified that approximately 95% of this is spam due to a sustained attack against the mail platform over the last few days. Therefore this entire queue is going to be placed into quarantine and filtered into the mail platform in a controlled manner.”

I know that some people have seen their mail to me bounce back (including someone looking for more information about the Minidisc player I’m selling on eBay, causing them to retract their bid) so if you’ve sent me something important…you might want to try it again.

Alphasmart Dana UK edition

The Alphasmart Dana is now available in a UK edition (i.e. with a £ key, and the quote characters in the right place). If you’ve never seen the Dana before, Charlie Stross has a good first look at it here. It’s basically a lightweight laptop alternative that runs Palm OS. Tiny screen, full size keyboard. All gorgeous. I want one.

Petabyte hard drives

In the article Petabyte Disk Drives in Seven Years–What Does That Mean for You?, Dr. Joseph Mercola talks about the prospects of having hard drives big enough to fit an entire lifetime’s worth of documents, books, photos, music and video. (Via The Shifted Librarian)

Great idea, but after last year’s Dead Disk Incident I don’t even trust current hard disk technology to keep my personal data safe. What was it Mr Scott said on Star Trek III? “The more they overtake the plumbing, the easier it is to stop up the drain.” I worry that smaller, finer, nano-scale components means that there’s more chance of something breaking. If a petabyte disk can detect magnetic fields a thousand times weaker than current drive technology, does that mean that the disk is a thousand times more susceptible to magnetic interference or other disruptive effects? How long do the magnetic regions (i.e. your data) remain stable before they decay and blur into a uniform, blank slate?

I think it’s useless to harp on about all the wonders this kind of storage technology will bring, without also wondering about the dangers of committing our entire lives to purely digital media. The main danger lies on having so much valuable data (remember: we may be talking here about all of the photographs you’ve ever taken, all of the emails you’ve ever sent or received) in one place. Bad idea. Really bad. It’s not a question of whether your hard drive will fail, it’s a question of when.

If you have any kind of attachment to your data at all, you ought to have backups. But backing up a petabyte drive (1,000,000 GB) onto DVDs (4.5GB capacity, or even 27GB with future Blu-ray technology) is the equivalent of backing up a 50GB drive to floppy disks. Even moving it all over a Gigabit ethernet connection to a remote server would take about three months.

Personally, I’d much rather see some big advances in data reliability engineering, or in personal backup solutions. Entrusting your precious data to a single hard disk without backup is like driving a car without a seatbelt. Don’t risk what you can’t afford to lose.

Microsoft buys Virtual PC

When I saw the news that Microsoft had bought Virtual PC from Connectix (via Sean Campbell and Andrew Swigart), my first thought was, “Wow–that’s different”. My second thought was, “What’s going to happen to the Connectix staff?” My brother-in-law works on VPC at Connectix, you see. (Further reading indicates that a lot of them will become Microsoft employees, reporting in to Microsoft’s Mac Business Unit.)

My third thought was, “Why?” This is the hot topic of the moment. Myself, I think that this has a lot to do with Linux, and selling Microsoft server OSs into the enterprise. It allows MS to go relax their stance on Linux without necessarily endangering license sales. (Not that they will change their position–just that they now can.)

Here’s the way it works: an MS client is tempted by Linux. MS can now say, “Go ahead and run it as a virtual machine. You can use whatever Linux apps you want, but at the same time you keep all the benefits of having the main server itself running Windows.” Or, to a sales target they want to convert from Linux/Unix to Windows: “You don’t have to port all your mission-critical apps over to Windows in a big bang. Just put in the new servers with Windows running on them, and let the virtual machines carry on running them until the native Windows versions are ready.” Either way, they win, because they can treat Linux as a third-party application instead of an operating system.

I also reckon they’ll gradually ease VPC away from the (Windows) desktop, and sell it primarily as a server product. VPC for the Mac will probably remain as a desktop product. Why? Because whether you use a Mac or a PC, a home user is likely to want to buy only one license for Windows. If MS promote VPC for the Mac, they gain license sales. If they promote it for Windows, they’ll be encouraging people to re-use their existing Windows license keys, and that’s a big no-no.

Interesting move, that’s for sure.

Memory prices

I still haven’t upgraded my PC yet, but
looking at the way memory prices are going (see graph below–the vertical axis show the cost of 512MB in GBP), I don’t feel in too much of a hurry.

Drop in DDR memory prices between December 2002 and February 2003

I can get 1GB of memory now for the same price 512MB would have cost at the start of December. Nice.