Tuesday, August 08, 2006

Thinking out loud about Leopard

The Leopard announcements have been widely judged as disappointing. Pretty much every SteveNote is judged disappointing just afterwards, and then folks gradually absorb the information and recast it all as amazing later (I, for one, thought the iPod was an overpriced useless gizmo that only geeks would buy; if you find reportage at the time, you'll see I was not alone), forgetting that at the time they were disappointed.

I wasn't disappointed. I'm writing this little note to myself as a kind of open diary entry (hmm I guess that's what a blog is :-) ). Maybe I'll laugh at myself in six months' time.

First of all, the Mac Pros are the most droolworthy pieces of hardware Apple has ever released. Well, at least since the Quadra 700 or perhaps the Mach 5 boxes (anyone remember those? Photoshop launched instantly on those suckers...). Sure, I think the cases look ugly and are freaking huge, but you get a whole pile of grunt for not much money. And after reading HardMac I discover that the CPUs aren't soldered in, so you can buy a 2GHz box and swap in faster CPUs at your leisure (more useful to folks outside the US where Apple's prices are ridiculous). The only real complaint is that there's no midrange graphics card option (e.g. an X1600 or a GeForce 7800).

But on the OS front, Time Machine -- if it works -- is a reason to buy Leopard on its own. Backups are a huge deal. We're reaching the point at which the hard disks on which millions of people are storing their entire photo and/or music libraries are going to die of old age and you can only imagine the kinds of horror stories that will start circulating by word of mouth. Similarly, the "but I was just magically saved by Time Machine" stories could become a serious selling point. For serious users developers, Time Machine could make, at minimum, version control for small, single-developer projects completely irrelevant. Whether it will have the ability to roll back servers, etc., remains to be seen, but just this much is a serious win (for me, anyway).

I might note that Time Machine -- at least as shown -- directly supports everything at file level via Finder, but it appears to offer an API for application developers to allow application specific and/or finer grained support. This is what makes having something like Time Machine implemented as OS level so powerful.

The ability to turn any part of any web page into a Dashboard Widget is not only incredibly cool, but it has the potential to both change the way we browse the web AND make Dashboard actually useful and amazing rather than a use-three-times-and-forget piece of eye candy (as it is now). Dashcode sounds pretty compelling, but a lot of that will hinge on whether its JavaScript debugger is as good as advertised.

iChat's screen sharing features (mentioned on the website but ignored during the KeyNote) are pretty mind-blowing. First of all, having this functionality for free at OS level is pretty amazing (but will iChat be able to broadcast and/or connect to non-Mac clients?) and might make remote meetings and collaboration a lot more doable than is currently the case with half-assed tools such as GoToMyPC or WebEx. The big question for me is the extent to which this functionality may be available independently of iChat. E.g. can we log in to a Mac OS X box remotely with a GUI, or can we only screenshare via iChat?

Next, there's Spaces. I've used several virtual screen programs over the years and at some point, such as when I stopped using the computer it was installed on, just given up on them. This is because while they are great in theory, they all suck in practice. It looks to me like Spaces may address this suckage in a number of ways, not least of which is by making it an OS feature so I don't need to install it (or license it) on specific machines. But it also appears to be much more intelligently and simply designed than other virtual screen implementations, in part because it's implemented at OS level:

1. When you jump to an app (e.g. by clicking on it in the dock) you automatically go to its space.

2. There's a well-defined spatial relationship between the spaces which is consistently reinforced with animations and screen layouts. This is really important (and one of the reasons why many people think the OS X Finder sucks*).

3. I'm hoping that it will be well-integrated with Expose (i.e. that dropping into the Spaces overview is, in essence, what Expose will do now).

I'm not convinced that Spaces will be wonderful; but it has the potential to suck less than its predecessors.

* I think the OS X Finder sucks too, but not because it isn't spatial. I think the OS X Finder sucks because it is trying to do a really hard job (manage hundreds or thousands of files in a directory, etc.) with a really bad UI design (arrays of icons or lists of text labels) that dates back to when users only had 500 files on their hard disk.

That said, most of the other announcements are pretty ho-hum. 64-bit -- yeah, whatever. Being able to install 16GB of RAM is all the 64-bit support I really need for now. Core Animation -- cute but will it lead to anything useful or just more cute screensavers? Spotlight -- stealing some ideas from QuickSilver and adding remote searches; cool but hardly earthshaking. Mail 3 looks nice, but I don't currently use Mail 2 (I use gmail) -- although I must say when I do launch it to check my old mail (from before gmail) it does make me think I should go back :-).

Finally, the new voice synthesiser rocks. It's a minor thing but still, it has useful applications. I could actually imagine writing a little app to turn texts from Project Gutenberg into iPod tracks or CDs for road trips ... it sounds THAT good.