Don't do it, Jon! Fighting a war on two or three fronts would be suicide.
This isn't a feature request in the same way that asking for a new language or a file browser is. Jon can't just do this then move on to another feature. It's a strategic choice to start producing several new products. There would be some serious, company-killing problems with doing it, and I'd like to convince people that they really don't want Jon to spend his time porting ST to other platforms.
As I said, this isn't a feature, but a new product. Every actual feature that gets added has to be developed, tested, and supported on every platform. This means a huge amount more development and testing time. Remember that any code that deals with files is now suspicious. 3D rendering, the heart of the Sublime Text experience, must now run on a new platform. Any python code has to be re-checked to see if it relies on Windows behaviour. Community plugins will fail on linux, or be developed just for it, creating a world of half-arsed addins.
Instead of hacking one path through the jungle, Jon must now hack two or three paths. Inevitably, he'll get less far down either. This will kill the forward motion of the product. Fewer features will be developed, since each one is more expensive. Language support, project support, documentation, package help, API tools -- they'll all advance more slowly.
Innovation also suffers. When you straddle several platforms, it means you can only include features that all of them can support. For example, Windows 7 has a new feature called Libraries, which might integrate very nicely with the idea of Sublime Text Projects. A Windows 7-aware version of Sublime can help me find my files more easily using the Libraries features of windows. But libraries aren't available in Linux, so If Jon tries to straddle the two worlds, he has to write something that fails to take advantage of the build-in capabilities of Windows.
And why should he support linux? So that a few potential customers, on a platform famous for being free, would get a slightly less sluggish experience if they buy. Worse, the likelihood is that someone who buys a linux licence will not buy a Windows licence, meaning that all the effort doesn't positively affect Jon's bottom line at all. There's worse than no money in it.
Lastly, Jon charges for yearly licences. Which means people choosing to repurchase do it if, and only if, this year's version has enough new features over last year's version. Dividing across platforms means that, when that year is up, the number and quality of features will be lower; when deciding whether to repurchase, windows users are more likely to choose not to renew.
I say all this because I want Sublime to survive, and I think the idea of a linux version is both perpetual -- people muse about it a regular basis -- and a really, really bad idea for ST.