[Big Mac]

People have been predicting an easy, rather transparent transition to Apple's Intel based computers, mostly due to Rosetta backward compatibility. But what will happen to all of the software that simply will not run at acceptable speeds through Rosetta? Any software compiled with G4 or G5 flags will not run. And only the less taxing software will perform tolerably. What will the novice owner of more demanding Mac software do when Rosetta won't run his applications (or only run them poorly)?

1. Should we assume that Rosetta will come with a set list of applications that will refuse to run and instead prompt the user to search out an Intel binary?

2. How do you think software distrubtors will treat people who expect to get Intel binaries of their Mac software? Do any of you believe the companies are going to provide free or low-cost Intel binaries for those who already own the Mac versions?

3. Slightly off-topic: How will universal binaries work out for applications that are very large?
a. Isn't it more plausible to assume that vendors of large applications will create separate Apple Mac and Apple Intel PC discs, and, if so, won't that be an nightmare for the neophyte?
b. Or is it even more likely that those large applications will, after a few releases, drop support for the Mac (PPC) entirely, and there will only be Intel binaries available?

[TETENAL]

Originally Posted by Big Mac

Any software compiled with G4 or G5 flags will not run.

That's not true. Software that requires AltiVec will not run, but 99.999% of the software does not and runs fine on G3 Macs. The exception are a few games maybe, but those you likely wouldn't want to play in a few years anyway.

Originally Posted by Big Mac

3. Slightly off-topic: How will universal binaries work out for applications that are very large?
a. Isn't it more plausible to assume that vendors of large applications will create separate Apple Mac and Apple Intel PC discs, and, if so, won't that be an nightmare for the neophyte?

The binary code of a (large even) application is not very much. We are speaking about a few MB. That will easily fit on a CD. Maybe some moron company will create two versions, but I doubt it. Companies that create nightmares for their customers don't deserve any sales.

Originally Posted by Big Mac

Or is it even more likely that those large applications will, after a few releases, drop support for the Mac (PPC) entirely, and there will only be Intel binaries available?

Intel Macs are Macs too. When only versions for Intel Macs will become available that doesn't mean a company "dropped support for the Mac". Eventually support for older machines will be dropped like it always has, but that's many years into the future.

[Big Mac]

Originally Posted by TETENAL

That's not true. Software that requires AltiVec will not run, but 99.999% of the software does not and runs fine on G3 Macs. The exception are a few games maybe, but those you likely wouldn't want to play in a few years anyway.

From universal_binary.pdf:
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
The way I read that that fourth bullet point, not only Altivec binaries but also any G4 or G5 optimized applications will not run.

The binary code of a (large even) application is not very much. We are speaking about a few MB. That will easily fit on a CD. Maybe some moron company will create two versions, but I doubt it. Companies that create nightmares for their customers don't deserve any sales.

So you're saying the amount of PPC binary within a PPC application is small? That does not make a lot of sense to me, but for Apple's sake I hope you're right.

[RevEvs]

Originally Posted by Big Mac

So you're saying the amount of PPC binary within a PPC application is small? That does not make a lot of sense to me, but for Apple's sake I hope you're right.

What hes saying is the PPC bit, the binary, is not that big. its the various resources and nibs etc that take up most room. so adding a intel binary does not add much to an app, so no need for 2 separate versions, you just have a slightly larger app by a few meg.
Both the PPC and Intel binary will share the nibs and other resources.

[Thinine]

Originally Posted by Big Mac

From universal_binary.pdf:
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

So? The only software that requires a G4 or G5 is stuff with AltiVec code. A G3 can run executables compiled with G4 optimizations just fine. In fact, G4 optimized is Xcode's default setting. So Rosetta will only choke on software that has no scalar fall back, which little software does.

[ghporter]

As RevEvs said, the actual executable binary code in an application-particularly an app that's been packaged for installation-is not huge compared to the "resources" such as icons, mappings for menus, etc. Further, considering that an app designed for distribution is usually compressed in some way, you can easily fit a lot of stuff on a single CD. Take a look at Office 2004 for the Mac; it has a bunch of rather large programs, but it's all on one CD. I believe that there will be NO problem in distributing dual-native applications on a single CD.

[Catfish_Man]

Originally Posted by Thinine

So? The only software that requires a G4 or G5 is stuff with AltiVec code. A G3 can run executables compiled with G4 optimizations just fine. In fact, G4 optimized is Xcode's default setting. So Rosetta will only choke on software that has no scalar fall back, which little software does.

Not quite. The G5, for example, has a hardware sqrt instruction. Any software compiled with the -mcpu flag set to G4 (or 7400, 7450, G5, or 970) will not run; Software compiled with the more common -mtune flag still will though. In general I've found that most software is tuned for the G4, but will run on generic PPCs.
<edit>
For purposes of comparison, the Adium binary is about 1MB, and the Adium Crash Reporter binary is around 60k
</edit>

[mitchell_pgh]

Big Mac, you are over thinking the issue.

Adding universal binaries won't add much space to a program 15% at the absolute most. A large majority of programs will run under rosetta. the ones that don't will either need to be updated or disappear.

[Superchicken]

Essentially Quark will die... and life will move on.

[Big Mac]

Okay, I appreciate the responses, but what about my second question:

2. How do you think software distributors will treat people who expect to get Intel binaries of their Mac software? Do any of you believe the companies are going to provide free or low-cost Intel binaries for those who already own the Mac versions?

[Chuckit]

For the versions that are Intel-compatible, I don't imagine there will be a need to provide Intel binaries for those who already own the "Mac" (you mean "PowerPC") versions, because they will be distributed together.

[Big Mac]

Originally Posted by Chuckit

For the versions that are Intel-compatible, I don't imagine there will be a need to provide Intel binaries for those who already own the "Mac" (you mean "PowerPC") versions, because they will be distributed together.

No, what I am asking about is software than you can buy today, next week and six months from now (if not later for certain titles) that will not have Intel binaries. The thing that made me pose the question is that I just picked up WoW yesterday (great game, btw), and it occurred to me that people will have (and will continue to buy) a number of Mac programs that will not run well on their Mactels.

[eevyl]

For the case you mentioned, WoW, I pretty much think Blizzard will continue to update it for PowerPC/Intel for a long time.

Heck, they still update Starcraft for OS X

[TETENAL]

Originally Posted by Big Mac

for those who already own the Mac versions?

The Intel Macs are Macs too.

[Big Mac]

Originally Posted by eevyl

For the case you mentioned, WoW, I pretty much think Blizzard will continue to update it for PowerPC/Intel for a long time.

Heck, they still update Starcraft for OS X

Oh yeah, I meant to add I was not saying anything in particular about Blizzard; I was just using WoW as an example because we all know games are not going to run in Rosetta. I know Blizzard will be on Mactel essentially from day one, but the general question I'm asking is, will their Intel binary be free for those who already have the Mac (PPC) version?

[Gavin]

The smart companies will have a downloadable upgrade that will get you the universal version. You already own the serial number. It doesn't really cost them anything, especially if they are already making point upgrades available for download.

I recall many companies had fat binaries for ppc/68k on their websites, although in some cases it took a year or so after the ppc came out. On the other hand there were those that forced you to buy / wait for the next upgrade - generated lots of bad blood.

We're gonna see a mixed bag.

[ghporter]

Originally Posted by Gavin

We're gonna see a mixed bag.

You can etch this in stone! Software houses are not well known for platform flexibility, so some will have even more of the problems they had in the OS 9 to OS X transition. Others will have seen the writing on the wall and have been working on expanding their compiler-base to allow for multiple platforms from the same (basically) code base. Still others will go with Mac/Intel compatible binaries for their newer products and let the older ones languish. It will be up to the market (that's US) to let these companies know what we want. Vocally, I hope!

[Tyler McAdams]

My problem lies in unsupported apps... one of the apps I depend on, blacklight.app is not longer supported and I can not contact the maker of the app. i'd hate to have to reverse engineer his work to recomplie for the Intel chip... but I don't see much of a choice now. Luckily it's a pretty simple app:

[eevyl]

Originally Posted by Big Mac

Oh yeah, I meant to add I was not saying anything in particular about Blizzard; I was just using WoW as an example because we all know games are not going to run in Rosetta. I know Blizzard will be on Mactel essentially from day one, but the general question I'm asking is, will their Intel binary be free for those who already have the Mac (PPC) version?

I don't see why not, as I said they continue to update Starcraft for OS X without update costs, they did a OS X installer even.

I'm pretty sure Blizzard is not going to charge for that update.

[Ludovic Hirlimann]

Originally Posted by Big Mac

1. Should we assume that Rosetta will come with a set list of applications that will refuse to run and instead prompt the user to search out an Intel binary?

No that would be too mluch work and trouble for Apple.

2. How do you think software distrubtors will treat people who expect to get Intel binaries of their Mac software? Do any of you believe the companies are going to provide free or low-cost Intel binaries for those who already own the Mac versions?

Depends on how those compagny do the transition themselves to fat binaries. If they release a new version with support for the x86 based mac , then those compagny will charge for it.

3. Slightly off-topic: How will universal binaries work out for applications that are very large?

A mac application is a Special folder which contains sub folders it looks like this
Supporting intel mac is just adding a sub folder containing the x86 executable. All the rest (images, nib files etc) canbe and will be reuse.

a. Isn't it more plausible to assume that vendors of large applications will create separate Apple Mac and Apple Intel PC discs, and, if so, won't that be an nightmare for the neophyte?

It would be. I think Apple will push developers not to split based on the architecture.

b. Or is it even more likely that those large applications will, after a few releases, drop support for the Mac (PPC) entirely, and there will only be Intel binaries available?

[/quote]
Todays (and wihout any doubts tomorow's) mac come with a DVD player so publisher will slowly switch to giving software on DVD instead of CDs.

[Chuckit]

Originally Posted by Tyler McAdams

My problem lies in unsupported apps... one of the apps I depend on, blacklight.app is not longer supported and I can not contact the maker of the app. i'd hate to have to reverse engineer his work to recomplie for the Intel chip... but I don't see much of a choice now. Luckily it's a pretty simple app:

There is no reason why Black Light would not be supported by Rosetta even if there weren't a new version. The criteria for working in Rosetta are so broad that very, very few apps are liable to break. I could probably name all the major ones (I can think of about five) that will need updating to run at all, and those all look pretty safe.

Incidentally, the author's blog seems to have been updated in the past month, so it doesn't look like he's dropped off the face of the earth or anything.

Then again, since the core functionality of Black Light has been rolled into the system, I don't really see the point in worrying.

[analogika]

Originally Posted by Big Mac

those who already own the Mac versions?

You meant to say "IBM versions".

Macintosh is 680x0. This unholy alliance with IBM is sure to be the downfall of Apple as ALL software breaks and the faithful leave the platform in droves to switch to the ethically superior Windows platform.

Don't pull a mAxximo, BM. It makes you look just as silly.

[analogika]

btw, is there any reason to assume that PPC software will run less well within Rosetta (provided it doesn't break)?

I mean, yes, it's emulation, but the Intel Mac systems, by the time they hit the market, will be so much faster than what we have now - will it really make a difference for 95% of applications?

Just musing.

[wulf]

I imagine it will be not dissimilar to the switch from 68k to PPC. Some things will run well, some badly, a few not at all. But most will run "acceptably" well, the speed improvements of the new platform minimising the hurt of running under emulation.

Of course those using Photoshop, FCP, Logic etc., in production environments will want their Pro apps native before they switch. I imagine all those major ones will be ready in time (bets are off wrt Quark), and a lot of the "lesser" apps too.

For pros, the switch from OS 9 to OS X was a far harder transition than 68k to PPC, and PPC to Intel will not be nearly as painful either.

[Maflynn]

I think people are over estimating the speed gains and under estimating the overhead for Rosetta.
The intel Macs are going to run the apps slower then the real Macs. That's just a fact, software emulation always runs slower then hardware.
A bad example but an example nonetheless is VPC; look at how slow that puppy is. While I won't say that rosetta will definitly have VPC speeds I'd be hesitant to say games will run fine.

The bottom line is at this early stage we really don't know how apps are going to run, will there be compatability issues (I'm going out on a limb and say there will be) and what CPUs apple will use for their intel macs.

Regards
Mike

[wulf]

I'm assuming that everyone understands that apps requiring high-performance will never be "acceptable" in speed under emulation. That includes games along with any media editing apps, particularly pro ones.

(From my point of view things like Photoshop are less of an issue unless you are really doing hardcore resizing and filter crunching all day. It will take a speed hit, sure, but that doesn't stop you actually using it.)

[ShotgunEd]

Hasn't it been established that Rosetta doesn't actually run apps under *emulation*, and thats why the speed hit won't be as noticable?

I can't remember the technical term, I'm sure someone can.

[Chuckit]

It is basically just a dynamic recompilation layer. It's not a full-fledged machine emulator, as far as I can tell. Whether or not it's comparable to "a PowerPC Mac" would surely depend on which PowerPC Mac you're thinking of. I wouldn't expect it to be anywhere near a G5, but it may match up well enough to, say, the old 800 MHz G4s I see a lot of people using.

[mitchell_pgh]

I find it interesting how people are trying to make this a bigger issue than it is.

Unsupported Apps:
An unsupported app is an unsupported app. While we all may have one, this is no reason to bitch at Apple for switching to x86. Eventually OS X would no longer support the app.

Speed Issues:
We have no idea how fast the Mac/Intel systems will run PPC code. Early reports are good, but who knows. Apple has another 6-12 months of tweaking before we need to worry about that. I'm guessing that games/high end programs will be the first to get universal binaries (beyond the ones that simply recompiled with little trouble).

On a side note, we may see BIG speed improvements from games as parts written for Windows in machine code will work under Mac/Intel.

[andgarden]

"Eventually OS X would no longer support the app."

I think this is a key point. Tiger broke lots of apps, so did Panther, Jaguar, etc. etc. Intel switch or not, the probability that the version of an app you own today would work in, say, five years, on a then current machine is variable. Doesn't anyone remember the switch to System 7??

[nonhuman]

Originally Posted by andgarden

"Eventually OS X would no longer support the app."

I think this is a key point. Tiger broke lots of apps, so did Panther, Jaguar, etc. etc. Intel switch or not, the probability that the version of an app you own today would work in, say, five years, on a then current machine is variable. Doesn't anyone remember the switch to System 7??

I don't think most of these people are old enough... Heh, I remember thinking the System 7 trash can was so cool and 3D!

[andgarden]

Originally Posted by nonhuman

I don't think most of these people are old enough... Heh, I remember thinking the System 7 trash can was so cool and 3D!

I was almost 6 yrs old, and seem to remember dismissing my father from the room to install it on our pizza box LC with a 12 inch color screen .

[Maflynn]

Originally Posted by ShotgunEd

Hasn't it been established that Rosetta doesn't actually run apps under *emulation*, and thats why the speed hit won't be as noticable?

I can't remember the technical term, I'm sure someone can.

Sounds like interpreted vs. compiled code. If that's the case then there's still overhead and still a performance hit. Remember interpreted basic vs. compiled basic, the compiled flavor was always faster.

Like another poster said, we really don't know how good/bad the performance is going to be but I'm of an opinion that it will not be as fast as running it natively and some apps will suffer more then others - Perhaps to a degree that it will be unusable. Then you though in the known and unknown compatibility issues and you have a recipe for a disaster. I'm not saying it will be but apple has a tricky road ahead to be sure.

Mike

[nonhuman]

Originally Posted by andgarden

I was almost 6 yrs old, and seem to remember dismissing my father from the room to install it on our pizza box LC with a 12 inch color screen .

Heh, I had the 13".

[nonhuman]

Originally Posted by Maflynn

Sounds like interpreted vs. compiled code. If that's the case then there's still overhead and still a performance hit. Remember interpreted basic vs. compiled basic, the compiled flavor was always faster.

It's vaguely similar to interpreted vs. compiled code, but not really. It's recompiling the code on the fly. That means it can look ahead and work to compile code further down the execution path so that it's ready when needed. I don't know exactly how it will work, but it may recompile the entire thing so that all the code is ready from the get go. If that's the case there would be an initial slow down, but once all the code was recompiled it would basically work as a native app (which makes you wonder why it wouldn't be able to basically create an x86 binary that could be used in the future, but I'm sure there's more to it than that).

As far as the speed difference between compiled and interpreted code, that's less of a factor when using modern languages with modern interpreters. Python is an interpreted language (although it can also be 'compiled' in a way) and is still quite fast and used for a number of applications. Java is also, in a way, an interpreted language as it's not compiled into machine code, but rather into Java byte-code which is then executed on a virtual machine. Modern Java VMs and compilers have made Java plenty fast for most purposes, in fact at some things Java is faster than C++.

So while there will be a speed difference between PPC and x86 binaries, it's not really comparable to the one you see between interpreted and compiled code. Sadly I don't know enough about the technical details of Rosetta to point to any specific differences, but based on what I do know about the general idea of what it does, I think it has the potential to be quite fast.

[Maflynn]

Originally Posted by nonhuman

Modern Java VMs and compilers have made Java plenty fast for most purposes, in fact at some things Java is faster than C++.

I knew Java was interpreted and to be honest I never thought of it as being fast. I've programmed in many languages (I've been programming for over 20 years now) and I've always seen compiled code execute faster then interpreted code. Your right at the effectieness of the interpreter and the speed of modern cpus has mitigated this somewhat.

To be perfectly honest I really have no idea at the speed hit were going to see when we move to the intel macs hopefully Apple has written some pretty tight translation routines to minimize the hit.

I'm anxious to see the first intel mac which should hopefully see some time next year

[Big Mac]

I'm reviving this discussion to post a pertinent article from MacAddict (September '05, p. 13, emphasis mine):

The Road to Intel Gaming
While Intel-based Macs are still a ways off, the transition has already begun for a few game developers. Epic Games' master programmer Ryan Gordon sequestered himself in his lab and ported playable (but slow) versions of Unreal Tournament 2004, GameHouse's Feeding Frenzy, and GarageGames' Torque Game Engine. Gordon used Duke Nukem 3D to test Apple's Rosetta PowerPC-application translator on an Apple-supplied Intel-powered Mac and said the game runs about as fast as it does on his PowerBook.

Blizzard's Mac programmers tested World of Warcraft with Rosetta and reported slow performance and minor rendering glitches. When they recompiled WoW to run natively on the Intel Mac, it ran into a bug a startup. It's been a bug hunt ever since.-Matt Osborn

Based on this article, I infer the following:
1. The notion that many applications will require only simple recompiles is likely a pipe dream;
2. A simple binary recompile will produce perfectly fine Intel binaries in certain cases, slow Intel binaries in some cases; in other cases there will be significant problems;
3. Developers who are faced with having to do a lot of recoding to optimize/fix their applications for both Apple Macs and Apple PCs will drop Apple Mac support as early as possible;

Ergo, It appears to me that if Apple's Intel switch is successful, a substantial portion of OS X applications will henceforth be developed for the Mactels only. Apple can barely support its own weight with third party developers as it is; if Apple's PCs bifurcate the platform as this news suggests they will, many are in for a rude awakening.

Note: I don't mind if you disagree with me, but consider the reality of the situation before you flame away. As I have said before, I would like to see Apple's switch succeed - yet logic dictates that Apple is placing its computer resources on various precarious grounds. Third party providers have enough trouble keeping up with minor updates to their Mac code bases. (Heck, even Apple has failed on numerous occasions to make important OS applications run properly on all supported Macs across the board.) This transition will not be effortless for third parties, and I frankly cannot envision a scenario in which Macs and Apple PCs are supported by them with any semblance of parity.

[Millennium]

Originally Posted by Big Mac

1. Should we assume that Rosetta will come with a set list of applications that will refuse to run and instead prompt the user to search out an Intel binary?

Yes.

2. How do you think software distrubtors will treat people who expect to get Intel binaries of their Mac software? Do any of you believe the companies are going to provide free or low-cost Intel binaries for those who already own the Mac versions?

Most will probably not go native until the next paid upgrade.

3. Slightly off-topic: How will universal binaries work out for applications that are very large?

Not too badly, actually. Unlike the old FAT Binaries from the 68K/PPC switch, these actually hold two separate copies of the app. Only the one which needs to run will run. End result: they'll take up lots of extra disk space, but not much more RAM.

a. Isn't it more plausible to assume that vendors of large applications will create separate Apple Mac and Apple Intel PC discs, and, if so, won't that be an nightmare for the neophyte?

Yes, and yes. That's what happened last time, and mark my words, it will happen again.

b. Or is it even more likely that those large applications will, after a few releases, drop support for the Mac (PPC) entirely, and there will only be Intel binaries available?

Again, yes. Most likely there will be one more release which supports PPC -perhaps two for a couple of apps- and then PPC support will be dropped.

I have nothing to back up my claims, except that this is exactly what happened last time Apple switched architectures. Apple tried to keep it from happenning, but it happened anyway, and it will happen again.

[Thinine]

Originally Posted by Big Mac

Based on this article, I infer the following:
1. The notion that many applications will require only simple recompiles is likely a pipe dream;
2. A simple binary recompile will produce slow Intel binaries in some cases, while in other cases there will be significant problems;

Ergo, It appears to me that if Apple's Intel switch is successful, a substantial portion of OS X applications will henceforth be developed for the Mactels only. Apple can barely support its own weight with third party developers as it is; if Apple's PCs bifurcate the platform as this news suggests they will, many are in for a rude awakening.

Note: I don't mind if you disagree with me, but consider the reality of the situation before you flame away. As I have said before, I would like to see Apple's switch succeed - yet logic dictates that Apple placing its computer resources on various precarious grounds. Third party providers have enough trouble keeping up with minor updates to their Mac code bases. (Heck, even Apple has failed on numerous occasions to make important OS applications run properly on all supported Macs across the board.) This transition will not be effortless for third parties, and I frankly cannot envision a scenario in which Macs and Apple PCs are supported by them with any semblance of parity.

Cocoa apps require just a simple recompile, Carbon apps not much more than that, but others are a crap shoot and Apple was clear on that at the beginning. WoW, as a likely pure C++ app, will of course require more work. So this doesn't really change anything. And the recompile isn't likely to produce slow Intel binaries unless there are significant Mac only optimizations (AltiVec, PPC assembly, etc.) or algorithmic optimizations that don't translate well. But that's why there's Shark, so devs can find those issues.

And you're completely wrong about the 'substantial portion' part. All Cocoa apps will be both, and even most games will be hybrids, though perhaps less focus on AltiVec/PPC optimization.

[saddino]

Originally Posted by Big Mac

Based on this article, I infer the following....

Based on one article, you infer that a substantial number of applications will not be produced as Universal Binaries?

Ok, based on the 20 or so comments by Mac developers here, I suppose I could infer that the Intel transition will be an absolute piece of cake, given comments such as:

“Porting Delicious Library to Mac OS X on Intel processors was literally a single click for Delicious Monster. A minute later we had an application that worked perfectly on both PowerPC and Intel machines, with no changes to our source code. As long as Apple picks the fastest chips, they could be tortilla chips for all we care.”

Having gone through the 68k->PPC and Classic->Carbon transitions, and after skimming the Universal Binary Programming Guidelines, I'm not sure I see why you're so concerned.

IMHO, this is a non-issue.

[Chuckit]

Originally Posted by Big Mac

I'm reviving this discussion to post a pertinent article from MacAddict (September '05, p. 13, emphasis mine):


Based on this article, I infer the following:
1. The notion that many applications will require only simple recompiles is likely a pipe dream;
2. A simple binary recompile will produce slow Intel binaries in some cases, while in other cases there will be significant problems;

World of Warcraft is probably not a good choice from which to infer things about what "many applications" will require. Games like that are are often among the most heavily optimized and system-dependent apps out there. I do think Apple's PR stuff is overly optimistic, but I'm not sure this is any more realistic.

[__^^__]

Originally Posted by Thinine

Cocoa apps require just a simple recompile, Carbon apps not much more than that, but others are a crap shoot and Apple was clear on that at the beginning. WoW, as a likely pure C++ app, will of course require more work. So this doesn't really change anything. And the recompile isn't likely to produce slow Intel binaries unless there are significant Mac only optimizations (AltiVec, PPC assembly, etc.) or algorithmic optimizations that don't translate well. But that's why there's Shark, so devs can find those issues.

And you're completely wrong about the 'substantial portion' part. All Cocoa apps will be both, and even most games will be hybrids, though perhaps less focus on AltiVec/PPC optimization.

The programs may only require a simple recompile but what about reading data that was created on PPC. There is still that endian problem.

[Chuckit]

Originally Posted by __^^__

The programs may only require a simple recompile but what about reading data that was created on PPC. There is still that endian problem.

Computers have been reading data written by systems with other processors for years. How is this a problem?

[Big Mac]

Certainly Apple wishes to portray the situation in the best possible light, but I'm not talking corporate spin here. The developer reactions Apple chooses to highlight on that page saddino cites are based primarly on either a) obscure and presumably small software, b) Cocoa applications (which don't exactly constitute a large portion of the software base) or c) are forward thinking, feel good statements that do not mention specific projects. Third parties often treat the Mac as a second-class citizen, and Apple's switch will only complicate matters for them. Think about your third party devices and try to imagine the type of support you may get in the event that Apple's PCs bifurcate the platform. It's hard enough to get a simple printer driver upgrade in a timely manner after a major revision of the OS. For me it is especially troubling that Blizzard has reportedly been having difficulties, since there should not be a whole lot of Mac proprietary elements within World of Warcraft. Blizzard's the greatest third party game developer, and even those fine folks are having a tough time. Granted, games are a special category - nonetheless it's time to get realistic about the implications of Apple's switch.

I am pessimistic that Apple can pull off yet another expensive transition. IMO, the reason why the Mac survived the PPC transition (other than Metrowerks) was that the Mac was still leaps and bounds ahead of the competition in usability and secondly that the PPC had so much in its favor. And the reason why the Mac survived the OS X transition was that the classic Mac OS was decaying, so Mac users were thirsting for a modern OS for their Macs. But the Intel defection? It comes on the heels of the OS X transition - in fact, as SJ pointed out the Mac had just gotten strong momentum going, post-transition. Now yet another transition is required - this time it seems rather unnecessary. And as we see, the Intel switch will definitely be disruptive - for some more than others yes, but certainly disruptive. Those who only use the basic Apple software bundle will notice the fewest issues. The truth is, many outfits will have to decide whether Apple support is worth the trouble, and if so what kind of Apple support? In the end we're faced with yet another costly and arguably needless transition. If Mactel is very successful, Mac support will be phased out hastily by many third parties that can not justify parallel development for one minority platform; If, OTOH, Mactel is unsuccessful then Apple will be guilty of risking the platform for nothing.

[__^^__]

I am pessimistic that Apple can pull off yet another expensive transition.

I am little more optimistic about the transition. There is nothing that can beat Job's RDF.

Now yet another transition is required - this time it seems rather unnecessary. And as we see, the Intel switch will definitely be disruptive - for some more than others yes, but certainly disruptive. .

I agree. What pisses me off most about the switch to Intel is the feeling that we are starting from ground zero, again. The other transitions were justified. Maybe this is Job's marketing plan. Major transitions every few years.

[__^^__]

Originally Posted by Chuckit

Computers have been reading data written by systems with other processors for years. How is this a problem?

Programmers. My experience has shown that programmers make alot of assumptions when dealing with software that must run on multiple platforms with different cpu architectures. Simple things like assuming that an integer is a certain size, that long ints are the same size as ints and endianess of data created on one architecture that needs to be read on another. This shouldn't be much of a problem with software that already has a windows version(quicken ms office etc) as the developers probably understand these issues. I think we will see problems with all of those nifty OS X centric programs.