Switching to Intel: developers speak out

by Giles Turnbull

Late last night (UK time) I set about contacting a number of OS X developers, asking them for their initial thoughts on the switch to Intel chips.

Because I wanted them to be honest and speak from the heart, I offered them the chance to be quoted anonymously if they wished. Not one who has responded so far has taken up that offer; to their credit, all the individuals who responded were happy to be quoted by name.

Broadly speaking, the response is positive, although several respondents had concerns about the future.

First up, here's what Eric Boehnisch-Volkmann of Devon Technologies had to say:

For us and our users it means basically -- nothing much. We will have to update a few parts of our applications that are optimized to take advantage of byte orders, but this will take no longer than a week or two when the new compilers work as promised. For most of our smaller applications such as EasyFind, it will most supposedly be just a click on a switch in Xcode.

[The switch to Intel is] a logical step when IBM is not a reliable partner any more. We think that the switch to x86 processor -- and so basically PC hardware -- is a great move towards competitive Macs but also brings extremely high risks. What happens if some hackers develop a patch to make Mac OS X run on standard PC hardware? Apple depends on its hardware sales and this could drop into nothing when people buy standard PCs to run Mac OS X.

Personally, we don't like the x86 architecture and believe that the PowerPC platform is/was the superior processor architecture. On the other hand, using standard hardware will bring a series of advantages: cheaper Macs, better compatibility to third party extensions from the PC market, Virtual PC running at full speed (important for all of us who have to work with one or two Windows-only applications) and a common PC platform for all operating systems, to name just a few. But, to be honest, we are completely unsure if this is a sunrise or sunset for Apple and the Mac platform. It could even be both.

David Watanabe, creator of NewsFire, said he's happy with Apple's decision, not least because his apps should be pretty simple to re-compile for Intel:

I'm entirely supportive of the move. I'm exclusively a Cocoa developer and have zero PowerPC dependent code, so all I need to do is recompile my apps. I imagine life will be harder for those who have legacy code to deal with, and I can understand how they might be upset by this. For me, though, anything that helps move the platform forward and expand the user base (which I presume is behind this strategic shift) is great news for me. I'm delighted with the news.

Here's what Brent Simmons of Ranchero said:

My first thought that was that this is a Very Good Thing for selling
Macs. When the average person compares an Intel machine to a Mac, he or
she notes that the Intel machine runs at 3 Ghz and the Mac runs at 2
Ghz. So the "faster" machines wins. Never mind all the marketing about
how it's like comparing apples and oranges, never mind the explanations
about how the PowerPC is a superior processor and so clock speed doesn't
matter -- that stuff doesn't really penetrate.

My second thought was to be glad I'm a Cocoa programmer. Doing a
universal binary that works on Intel shouldn't be much trouble. Sure, it
adds a little to testing, but it's probably not as big an issue as
supporting multiple versions of OS X (which we do).

My third thought is that I have some questions. Could I buy a Dell and
run OS X on it? Could I buy a Mac and dual-boot with Windows?

My final (for now) thought is this: I'm glad I don't have to hear
Mac-on-Intel rumors ever again!

Gus Mueller, developer of VooDooPad at Flying Meat made a good point; that Apple needs to communicate with customers and developers better and clearer than ever before:

My very first reaction was: oh crap. But after watchings some demos, I think it will be ok, and probably good in the long run. They are also a lot further along with it than I figured they would be. Quicktime, the Finder -- everything in the the keynote today just worked. That's amazing to me. Rosetta will be a godsend as well.

The only thing that worries me is that people/customers will get scared. Apple has to do some marketing magic here, and get their reps out there showing people these new boxes with Intel inside. If everyone could see what I've seen today, with how easy it is to port cocoa applications, I think everything will be ok.

Nobody I've really talked to seems freaked out - which is a good thing. But most everyone I know uses Cocoa, and nobody uses Metrowerks.

Steven Frank, of Panic, was pleased to see a way out of the IBM "dead-end":

I'm not really worried about it from a technical standpoint. We've lived through the 68K to PowerPC transition, and the Mac OS 9 to Mac OS X transition. We've already migrated all our apps to Xcode, and all but one to Cocoa, so I don't expect a lot of work to be required on our part. It should be substantially easier than the OS 9 to OS X transition.

It does feel like we (the Mac community) were at a bit of a dead-end with the IBM chips. The G5 gave performance a nice boost, but then things started lagging again. My gut feeling is this is a plan that's good for the long term, although it's going to cause a little pain up front. I think that's as true for this transition as the last two.

If the net result is it moves more Macintoshes, that's good for our business, and I'm all for it. I don't really care what's in the box as long as it runs Mac OS X well, and I think for the most part that will be the opinion of the casual end-user as well.

Personally, my top two concerns at this time are:

  1. Is this announcement going to create an "Osborne effect" where Apple sales halt in anticipation of the new Intel-based Macs? Even if it does happen, Apple has a lot of cash, so I expect they should be able to ride it out.

  2. How does Apple prevent Mac OS X from being hacked into running on budget PCs? Do they intend to try? Apple is primarily in the business of moving hardware, so this seems to me like the biggest threat to their bottom-line, yet also seems inevitable.

OmniGroup CEO Ken Case said the company's history of developing for various processor types would come in useful:

We're looking forward to the Intel switch: we already have experience with developing our applications for multiple CPU architectures (from our experience on the NeXT platform, where we simultaneously supported NeXT, Intel, HP, and SPARC processors) and it adds the benefit of more hardware options with only a little additional mental discipline.

By the way, our open source application frameworks still include support for the Intel processor, and may work without any changes at all.

The last word (for now, because this is by no means the last we shall hear on this subject) goes to Rich Siegel, founder of BareBones:

For developers who, like us, have been following Apple's recommended best practices, we expect this transition to be a smooth one. If you're building using Xcode and writing software using supported and documented APIs, we expect that the conversion and continued development of software to run on Intel-powered Macs to be a straightforward task.

By using Apple's tool chain and adhering to the recommended best practices, we don't anticipate any serious difficulty. We have every confidence that Apple will provide what we need, including answers to any technical questions that may arise, in order to ensure the smoothest possible transition.

Apple's efforts are clearly intended to provide customers with the best personal computing experience, and from the business viewpoint they are clearly focused on growing the Mac market. As a longtime Mac-only developer, we consider this a very good thing, and we certainly don't feel let down by this transition.

It's natural to be cautious on the cusp of a transition as big as this one, but we've successfully led the way in making big transitions in the past (such as 680x0 to PowerPC, and of course Mac OS 9 to Mac OS X), so there's not much in the way of angst.

These are just the responses I've had so far; I hope more developers will be in touch soon with their thoughts.

And you have any thoughts you'd like to add, you can add them here.


2005-06-07 03:56:12
You should have asked Metroworks-using coders
Your post puts a nice sheen on the news, only because you interviewed people who are using xcode, which accounts for roughly only half of the apps out there. But xcode is a bit of a confusing and unwieldy beast and programmers often found they got more out of Metroworks products, especially games programmers.

Also, go ask some people who programmed for altivec what they think of this -- they have to recode for a new processor which won't run their apps as fast (and since Rosetta doesn't 'do' Altivec-enhanced apps, those apps won't work at all under emulation.

2005-06-07 04:02:00
Goodby, Mac
As a user who was intending to buy an iMac, and as a programmer who knew why the PPC is a better chip than anything Intel has managed so far, why would I now bother? Macs will now use the same ancient, slow, hot processor with the same lousy multi-processor performance. Given that a Windows PC will run all the classic Mac apps (InDesign, Photoshop, etc.) under Windows, and will no longer have the disadvantage of being hot and noisy while retaining the advantage of having far more software available in general, and probably costing less, what is my motivation now?

Jobs has sunk the Mac by not understanding why people who did use Macs used them and not understanding that people use Windows even though it sucks because they fundamentally don't care about the GUI of the OS. If they did care they would have switched to Macs years ago. Now there is literally no reason for such people to change.

And that's without asking the other obvious question: why Intel and not AMD, who are at least making some of the improvements to x86 that Intel has failed to deliver for more than twenty years?

Mad or stupid? I can't decide which.

2005-06-07 04:49:48
You should have asked Metroworks-using coders
I've sent out the same email to all sorts of developers, including several Metroworks/Altivec folks. I'm still collating responses. Watch this space.
2005-06-07 05:08:30
A Mac will still be a Mac... (3rd try)
[Lesson: don't try to clean your keyboard in the middle of a post. Sorry.]

An x86-based Mac will still be very much a Mac, I'm guessing: Open Firmware, FireWire Target Disk Mode, that wonderfully flexible multiple volume mounting and booting, etc. I suspect that once Wintel users realize what their architecture could have supported, but didn't, all along, and that cool features and stability aren't dependent on a "different" architecture, they'll be much less satisfied with the Wintel offerings.

Also, count on Apple to manage this transition well. It's much less of a disruptive shift than the 68000-to-PowerPC switchover. I once worked on Lotus 1-2-3 for Mac, which was built as 68K binaries for Systems 6 & 7. That code base (originally ported from MS-DOS) was already retired and wholly unsuitable for porting to PowerPC, yet, through Apple's compatibility efforts, it has continued to run flawlessly all the way through OS X 10.3.9 (so far) and dual-G5 hardware.

2005-06-07 05:14:11
Goodby, Mac
As somebody who has recently bought a powerbook (after 15 years of sticking with MS), I completely disagree with "Now there is literally no reason for such people to change".

I _didn't_ buy it for the CPU architecture (even though I am a programmer) - I bought it for the OS, the UI, and the software on it. I absolutely _love_ the consistency of the applications, the absolute god-like genius of Garageband, the integration between the apps. I bought it for the experience, for the productivity, the creativity, the flow - it just _feels_ nicer. I really don't care what's inside.

2005-06-07 05:21:55
You should have asked Metroworks-using coders
Still, your post should have been other than the rosy scenarios predicted by those using xcode. Metroworks-coded apps have to be significantly, tediously overhauled. And as for Rosetta, Apple says:

"Rosetta does not run the following:
- Applications built for Mac OS 8 or 9
- Code written specifically for AltiVec
- Code that inserts preferences in the System Preferences pane
- Applications that require a G4 or G5 processor
- Applications that depend on one or more kernel extensions
- Kernel extensions
- Bundled Java applications or Java applications with JNI libraries
that can’t be translated"

2005-06-07 05:59:11
Shrink wrap it. And make it 64 bit native.
I saw where Apple said they wouldn't allow OS X to run on non-apple hardware. I think they are fooling themselves if they think they can keep people from hacking it, given that Darwin (the core of the OS) is open source and already runs on many commercial PCs. Sure the installer won't work out of the box, but it won't be difficult to create one that will. The internet is full of people with a lot of time on their hands.

Why not just shrink wrap it for an approved set of PC hardware and sell it retail? Let the dishonest steal it and get hooked. That's how crack dealers (and Microsoft) get their start. Heck, the increase in OS sales might offset the loss of hardware sales from people waiting for intel boxen. With all the negative vibe toward MS these days they stand a good chance of taking over before Longhorn is out of the gate(s).

On another note, Everything I see seems to indicate that the current port is to the old 32 bit x86 mode. Seems like they should be targeting the nice, risc like, 64 bit mode. After all, they will be replacing 64 bit ppc with 64 bit intel eventually and why create another architecture to deal with. 64 bit intel/AMD isn't the same as 64 bit PPC. On the PPC the 64 bit architecture was a strict superset of 32 bit. On the intel/AMD architectures the 64 bit mode is a completely separate mode than 32 bit. They will have yet another awkward moment if they don't do this now...

2005-06-07 07:41:52
Shrink wrap it. And make it 64 bit native.
I agree that switching to 32bit intel (read pentium) is a stupid idea.
2005-06-07 08:17:54
A Mac will still be a Mac... (3rd try)
The Universal Binaries PDF (available from the Apple site) says "Macintosh computers using Intel microprocessors do not use Open Firmware", although it does go on to say that much of the information available through the IO registry is still available.

Similarly, it cautions that disk partitioning for "a Macintosh using an Intel microprocessor" (snappy, that) will differ.

Having said that, I'm not sure if most people will really care about that sort of low-level stuff if the rest of the Macintosh experience stays as seamless on Intel as it's generally been on PPC.

2005-06-07 09:01:31
A Mac will still be a Mac... (3rd try)
For the record, Apple is saying that their Intel-based machines will not have Open Firmware. Maybe they'll use something like an ARC PROM, or maybe they'll use some other more traditional BIOS PROM.
2005-06-07 10:57:03
A Mac will still be a Mac... (3rd try)
What makes you think Wintel users would realize anything?
2005-06-07 15:34:44
Goodby, Mac
Given that a Windows PC will run all the classic Mac apps (InDesign, Photoshop, etc.) under Windows

I think you need to check a few facts.

2005-06-07 15:39:26
Shrink wrap it. And make it 64 bit native.
Why not just shrink wrap it for an approved set of PC hardware and sell it retail? Let the dishonest steal it and get hooked.

The dishonest will steal it and run it on unapproved hardware anyway. Apple is still a hardware company, I don’t see how allowing OS X to run on hardware they made no profit from would benefit them.

2005-06-07 19:09:32
Look and Feel
I too was hoping that if Apple went x86 they would do so with AMD. That is not the case however ... and since Apple has always been knocked by the pundits for going with "that other" chipset the same would have happened with AMD.
The Cube was a flop, it was recently mentioned in an article I read but I thought to myself that the iMac Luxo lamp did quite well, and now the Mac Mini, so far so good. Is that not a cube? It just happens to be at a better price point.
So, the switch to Intel. OS X will still behave like OS X for the majority of things that people use it for. As X on Intel grows it will gain the equivalent of those things it might not support at first. I bought a Mac because I didn't like Windows. I didn't like Windows because I didn't like Windows. Not because I didn't like Intel.
With cheaper, and cooler running, processors think what apple could do with the iMac or Mac Mini. Cheaper processor means you can put resources into the other parts to keep up with the game, and stay ahead of it. You'll buy Apple because they will still be the first to support kick butt stuff like Bluetooth "the faster" or ultra gigabyte ethernet (or whatever is next).
I conclude with the knowledge that the majority of people don't know what goes on under the hood, just like with cars. If you push on the gas and it goes then they are all set. Do you know how many people can't function an iPod? It's the world's easiest to use yet people still unplug it when it says "DO NOT DISCONNECT." So, don't disconnect with Apple because it's Intel inside.
Oh and lets say that after IBM pulls off Cell, or something else, that Apple can't switch back? Universal binaries will mean that if IBM or PPC derivative does come back you'll theoretically be able to use those again. Whatever gives Apple the edge, right? As someone mentioned Darwin is on many platforms, as Unix is on many platforms. Doesn't it make sense to have a mainstream OS that will do the same?

P.S. Check out www.tomshardware.com for an article on the performance of a Pentium M adapted to a P4 Mboard and how it ran cooler and kicked not only P4 but also Athlon 64 FX in majority of tasks.

2005-06-08 03:52:26
Shrink wrap it. And make it 64 bit native.
Sure some people will hack it and get it to run on Dells (or whatever). But those people are definitely in a very small minority and do not represent the market in general. Do you honestly expect big corporations to buy a job lot of Dells then run hacked versions of OS X on them against the terms of the license and run the risk of being sued into the ground?

No, nor do I.

Do you expect Apple to want to spend development dollars on having to support every piece-of-junk mobo under the sun just so that they can lose hardware sales and hand them to someone else?

No, nor do I.

Do you want OS X to become far more flakey as a result of running on said piece-of-junk mobos?

No, nor do I.

It makes absolutely no sense for Apple to give this sort of thing their blessing.

Let's face it, those that intend to hack OS X to run on some cheapo build-it-yourself box are more likely to pirate the OS and any software running on it. Why should Apple make life easier for them?

2005-06-08 06:12:40
People who hack OS X onto a non-Apple PC
will get no support from Apple nor from their PC maker. For the vast majority of people who read this story until they reached this point, that is no big deal. But for the vast majority of people in the real world, it is a HUGE deal -- in fact, support is critical.

A few geeks are going to take the Mt. Everest because-it-is-there approach and shoehorn OS X onto a Dell, but the overwhelming majority of people who want to run the Mac OS with buy a Mac.

2005-06-08 14:40:16
Shrink wrap it. And make it 64 bit native.
That supposes that the OS and the computer are two different things. As computers become more appliance-like the line between OS and computer becomes blurred. Maybe you could put a Honda engine in a Toyota, but why? Only a small group of people will be interested.