Can we make that a $29 bet? :)

Posted by MB on 10 June 2008 @ 9pm

I’m not anywhere near that confident. :)

Posted by Waldo Jaquith on 10 June 2008 @ 9pm

As a software engineer, I understand the desire and need to do some refactoring now and then, but my question is, is the harddrive and memory space you save going to be more than $29 worth of harddrive and RAM? If not, as a rational actor (and as we all know, consumers are rational actors), why would anyone buy this? If it is, what about once we account for the opportunity cost of the time it takes for me to purchase this software?

It’s entirely possible that Snow Leopard will easily beat all these metrics and deliver other usability improvements (responsiveness, bug fixes) that make it well worth it. Based on the information available though, when I read “This is why I love Apple,” I can’t help but wonder why, aside from the partisanship that goes along with a nerd’s choice of operating system.

Disclaimer: This critique does not apply if we’re talking about running the OS on the hardware it will be bundled on, only upgrades. I’m also not addressing the side of this for developers looking to use the new multicore support or “OpenCL,” ’cause honestly I don’t know much about what they’re doing.

Posted by Ben C. on 11 June 2008 @ 9am

I suspect that you and I approach software development very differently, Ben. I’m not an engineer. I approach software more like an artist, something that makes engineers cringe. :) One of the reasons why I have a Mac mini is because it’s a beautiful machine. A cost benefit calculation would probably demonstrate that it doesn’t make much sense to buy one for its compact size — my rent just isn’t that high. :) Likewise, the time and hard drive space saved by Snow Leopard may not be worth $29. But I love that Apple apparently thinks of software as I do — that it should be smaller and faster for the sake of being smaller and faster, because that is better. It shows a love of the craft.

There are, of course, practical considerations, though it’s hard for me to say whether those are post-hoc justifications or not. For instance, this will extend the life of machines. Folks with machines that are sluggish with Leopard may well find that they can hold off on buying a new computer for another couple of years, if the improvements in Snow Leopard are significant. That amounts to some very real savings for hundreds of thousands (millions?) of Apple computer owners.

Posted by Waldo Jaquith on 11 June 2008 @ 11am

I’ve seen speculation that Apple will regain disk space by stripping out all the PowerPC code. Any thoughts? I’ve no idea how long Apple has said they’ll support PowerPC hardware.

Posted by Jon on 11 June 2008 @ 1pm

Heh, I’m actually not much a fan of the “engineer” term myself, but it’s in my official job title currently, so it slips out sometimes.

I can really appreciate elegance in software, both when I’m writing it and when I’m using it. I have no problem with people who buy an Apple because they like the UI, nor with you buying a mini*. However, I’m a skeptic as always and while it may actually be worth the $29 in terms of extending the life of the computer (a claim that will be more measurable once we know the actual improvement made), it seems that its real appeal is in selling you the idea of elegance.

This is actually somewhat of a funny argument for me to be making, as I’m someone who makes his own perhaps irrational decision about computers in order to obtain “elegance” or the idea thereof: my desktop is a pretty well customized version of Debian Linux running Fluxbox. I also value elegance, but I think I have a different measuring stick than most Apple enthusiasts.

* It’s worth noting that, while general space in your house or apartment might not justify the cost, we American consumers generally behave like desk space is at a premium, by purchasing smaller computers and making the switch to LCDs while they were still significantly more expensive but thinner than the CRTs.

Posted by Ben C. on 11 June 2008 @ 2pm

Jon, you are correct. Apple’s whole Universal Binary solution to the Intel switch consists of just gluing together the two different versions (Intel and PowerPC) of an application in one file, and the computer just runs the one that it knows. (For the geeky, see the lipo man page: http://developer.apple.com/documentation/Darwin/Reference/ManPages/man1/lipo.1.html#//apple_ref/doc/man/1/lipo ) So, dropping PowerPC support will reduce the size of many files by half. This move can easily reduce the footprint of OS X by several gigabytes. As an example, I just used lipo to strip the PowerPC executable out of Address Book and it reduced it by 49.5%. Now, this will only affect executables, libraries, plug-ins and the like, and other components of software (like images, XML data and so on) will be unaffected, but that’s still a substantial chunk of the OS that will be affected.

There’s a lot I want to say in regard to Ben and Waldo’s discussion, but NDAs require careful choice of words regarding what I hypothesize versus what I’ve seen. Honestly, I think the disk space and memory gains are incidental to what they’re doing behind the scenes, and Ben is absolutely right to be skeptical of that alone justifying the cost. It’s simply something easy for them to point at this early stage and have your average user understand its benefit. I expect the real magic will be in performance. OpenCL is nice and will benefit a lot of multimedia functions, especially in regards to ever-cycle-hungry H.264 video, but I’m willing to bet the real magic will be in Grand Central. This will require some changes in the code of 3rd party apps to take advantage of it (as if that hasn’t already become an obnoxious fact of live in the world of Apple development), but once developers start to finally get their heads around parallel computing rather than serial, I think we’ll start to see big performance gains without any hardware changes whatsoever.

Posted by Will on 12 June 2008 @ 2am