10.5.3 time and I'm still running 10.4.11. That alone should tell you something about how I feel about Leopard.
The Snow Leopard approach, a OS-cycle dedicated to "just" stabilization and performance, is a welcome investment on the part of Apple.
It's interesting though. It seems that Apple is saying: we can invest an year just to digest and make what we have better, because there is no competition for the next 2 years at least. And they are probably right - Microsoft operating system department is so messed up right now (XP doesn't want to die, so now you can buy Vista and upgrade legally to XP).
From the desktop Snow Leopard page, the only thing that I'm interested on is OpenCL. I don't know how "open" it will be but the idea of giving an API to the GPU is a welcome addition to the set of APIs. Other operating systems already have them, so its nice to see it reach our OS.
Personally, I think that OpenCL is the only API I will really use to its full potential on my desktops or laptops. This of course, if someone ports the H.264 encoder that Handbrake uses to OpenCL. Using the GPU to speed up my DVD-to-Apple TV sessions would be a most welcome improvement.
On the server side of the equation, Slow Leopard brings an interesting feature: read/write support for ZFS volumes. Nothing is mentioned about booting Mac OS X from a ZFS volume, so I don't think you'll get that, but your big disks should be able to use a modern file system next year.
One thing I find strange: you'll have to wait for 10.6 to experience a Safari with SquirrelFish? Thats bizarre. I understand why Apple would like to upgrade the Safari browser with major releases only (back-porting to the previous release like it happen with Safari 3.1 and 10.4.11), but the Webkit development is (or seems to be) very very fast, so I would like to see a more aggressive schedule for Safari releases.
So, next year, Apple will digest the 10.4 and 10.5 new features, to prepare a massive 10.7 upgrade in 2010/11. Just in time for Windows 7.