Apple pushes devs to deliver 64-bit support with new Snow Leopard beta - Page 2

I'm gonna have to side with melgross here. Calling 32/64 binaries Universal Binaries would bring too much confusion.

People will download the Universal Binary thinking they can run it on their PPC Mac.

Sorry but m2002brian and shadow but calling 32/64 binaries UB simply won't work unless it also contains the PPC binary considering people have grown to associate the UB meaning to a PPC/x86 fat binary. You guys can keep calling 32/64 binaries UB but you'd be doing yourself and everyone around you a great disfavor. And if developed apps and called your 32/64 app a Universal Binary, you'd be getting sued by customers still on PPC Macs that bought the app on the premise that it would work on their PPC Mac.

UB means PPC/x86 fat binary. Period. Any other definition given to UBs would be misleading.

http://www.apple.com/universal/

Of course, this doesn't mean you can't have UBs that contain 32/64 x86 support...but it must also contain the PPC binary for it to be called a UB.

The universal binary mechanism is used to provide 32-bit and 64-bit versions of both PowerPC and x86 code. Intel-only applications still use it for 32-bit and 64-bit versions.

It's nothing special. Even a single-architecture binary is still a universal binary (marketing bumf for a fat binary) that only has one architecture included.

At least two people in this thread of <100 messages so far disagree with you, so how much did you bet again?

Nope. Sorry. At least 3 people in this thread are wrong...including you. You can't call it a UB unless it can run on x86 *AND* PPC. Apple rules, not mine.

If there was a bet going on with melgross, all 3 of you lost the bet.

Unfortunately, you assumed that UB is a marketing term for fat binary. This is only true in the sense that it contains more than one binary. But Apple clearly markets the Universal symbol as a guarantee that an app will run on PPC and x86. Changing the definition at this point would be total nonsense. You can't have UB term mean different things at this point. UB would lose all meaning to the customer because they would still have to inquire if the UB means that the app will work on their PPC machine or if it means that it's a 32/64 x86 fat binary.

Baby steps, young Padawan, baby steps. Let us review and then make a prediction.

Tiger (two builds, four architectures, one kernel)

• PowerPC - 32-bit with "Classic" environment
• PowerPC - 64-bit with "Classic" environment
• Intel - 32-bit, no "Classic" environment
• Intel - 64-bit, no "Classic" environment
• 32-bit kernel

Leopard (one build, four architectures, one kernel)

• PowerPC - 32-bit
• PowerPC - 64-bit
• Intel - 32-bit
• Intel - 64-bit
• 32-bit kernel
• "Classic" environment dropped

Snow Leopard (one build, two architectures, two kernels)

• Intel - 32-bit
• Intel - 64-bit
• 64-bit kernel (default)
• 32-bit kernel if hardware doesn't have 64-bit kernel extensions (based on my understanding)

OS X 10.7 (one build, one architecture, one kernel)

• Intel - 64-bit
• 64-bit kernel

Apple has enough on their plate to not worry about dropping 32-bit Intel support. This will actually be better for the user because not all hardware will have 64-bit drivers when Snow Leopard is released. Current hardware will most likely have 64-bit drivers, but drivers for "old" hardware might take awhile or might not get written at all because the manufacturers might use this as a way to get people to buy new equipment.

So is there an easy way to see what kexts you have on your machine right now?

That's why God invented pronouns and common sense synonyms like "the company", etc. Trust me as an avid reader and writer on things Macintosh, this is the only place I see "Cupertino based _____" unless it's in a general newspaper or something else outside the industry.

i was WRONG, was thinking of fat binary. Sorry dude

This made my day, absolutely hilarious. Apple isn't the one rushing out an OS...

I'm with ya. That's one thing I love about Vista, resolution independence for higher density monitors. I don't have to squint as much or pull the monitor closer to my aging eyeballs in Vista. \

I doubt that very much. People already have it set in their minds as to what it is.

For all the time we've been discussing the 32/64 bit question this is the FIRST time anyone ever referred to it as a Universal Binary.

I'm not talking about what's technically correct about the term, just what people expectations have been in using it.

Let's see here, you've posted a few dozen times, and have been here for about a year. How many times you lurked during that time, I don't know.

I've been here a lot longer, and as you can see, by my number of posts, I've been a bit more involved in discussions.

I'm going by that strong involvement to say what I've seen over the time that Apple had invented the term "Universal Binary". I've seen the way it's been used, and the way it's not been used.

I'm going by that.

It's not been used before to distinguish between 32 and 64 bit binaries for Intel.

It's been used to distinguish between binaries for PPC and Intel.

That's the way people, except for a small few obviously, think about it.

Even developers here have discussed it that way.

All I'm saying now, is that to change this is going to confuse a lot of people.

Well, this is the very first time anyone EVER called 32/64 bit Intel code a Universal Binary.

Like it or not, its nonsensical. It WILL confuse a lot of people.

And if you go to the page Kim linked to, you will see why.

Wow! Two whole people disagree with me on this. Impressive!

Much like we call our current broadband equipment "modems" despite the fact that they do not Modulate demodulate anything it's easier for consumers to understand.

Stay way from Universal Binary for 32-64-bit situation and the confusion can probably be ameliorated somewhat.

That's all I'm saying.

I also don't like calling our gateways modems, but the name has stuck, incorrectly.

At least here, we're NOT trying to get people to say something improperly, just the opposite.

And why do we call goods transported by land a "shipment", yet goods transported by sea "cargo"?

That's actually funny!

It's ok, we'll keep it your way, wouldn't want anyone to get confused. God forbid any consumer would ever educate themselves on what they're spending they're money on. You know, instead of just assuming, because "that's how it was. Or you know the kind that buy things based solely on the emotions it sparks. Like those people still buying American cars cause they're American. Really though you're right UB does mean what I thought it didn't. For someone who says so much, you should read the thread. I said I was WRONG, and I was thinking of fat binary. Is that hard to understand. Sorry DUDE

Actually I'm on here everyday. Hate Tecfud, ok not hate, but do be so blind to one's own ignorance must be nice. Love almost all the conversation. Lots of smart, loyal fans.

x86 is a single Architecture with 32 bit and 64 bit instruction sets.

This is one architecture. Now if OS X were on IA-64 then yes, IA-32 and IA-64 would be two unique architectures.


http://en.wikipedia.org/wiki/X86-64


PPC, Sparc and HP PA-RISC were three separate Architectures. With x86 NeXTSTEP/Openstep was Quad Fat.

No reason to get upset! This is a technical discussion, not an emotional one.

If you confused one thing for another, that's fine. We all do it one time or another.

I'm just pointing out that when a term is used one way by almost all people, and the term is correct, to apply it to something that's different, even if it might technically correct, will confuse a lot of those people. We've already had other people agree on that. Since Apple always refers to it as PPC/Intel code, I don't see why we shouldn't continue to limit it to that, and refer 32/64 bit code that way.

Even Wikipedia refers to it that way, though the 32/64 bit code has been done for a while.

http://en.wikipedia.org/wiki/Universal_binary

I just like to keep things consistent.

In Apple developer docs you could see references to a 2-way universal binary or a 4-way universal binary. These corresponds to the number of architectures used. If the binary is described as 2-way, you would expect 32 bit Intel/PPC, but I am not sure the PPC part is a requirement.

Tthere is a simple shell tool which reports what the binary is. I tested a small 32/64 bit intel binary and that's what it says:




Apparently, Apple's utility refers to the 32/64 bit intel binary as a Universal binary. I agree that this might be confusing for the end user, though, so Apple may choose to market this feature using different name, just as they did not reuse the "Fat binary" term.

I agree. Apple understands about user confusion very well. I'm just asking that we maintain their long time public use of the term.

The Mach-O has two strings. They are hard coded that proclaim two distinct architectures. They aren't two distinct architectures. Intel decides what is or is not a different architecture on it's platform. This string value was added to make a distinction so people would have a transition from 32bit to 64bit down the line. This allows Apple to proclaim an change in architecture without getting into the weeds of explaining just what is x84_32 vs. x86_64.

And why do we call the place we park a driveway and the place we drive a parkway?

You are confused here by the ambiguous use of the term architecture. Core 2 Duo and i7, aka Nehalem, are called different architectures sometimes, but the term microarchitecture should be used here. i386 and x86_64 are different architectures because the compiler needs to generate different code for them. x86_64 can execute i386 code in a separate process because it is backwards compatible but the binary code for x86_64 and i386 is different, the size of the operands, instruction alignment etc. are different. Intel can not "decide" to call x86_64 and i386 the same architecture.

It does seem that there is a shortcoming in our lexicon for referring to these binaries.

Depending on who you talk to "universal binary" means "runs on any mac" or "runs on PPC or intel". There's the official Apple stance and then various interpretations in actual usage. There's the literal interpretation from a layman's or developer's perspective. Then there's the people who try to reason through how the term would be most useful. Then there is the by-the-book (apple's definition) perspective as well.

I don't claim to know what the solution is, but it is obvious that Apple will need to be pro-active to avert confusion.

If we had a time machine, we could go back in time and not use the term "universal" in the first place. Going forward, such a term almost always necessitates the literal and official definitions being in contradiction. This same exact problem exists in more fields than just computing. Various fields have things like "universal wrenches" or "universal adapters"... that aren't universal. I think it must all be a conspiracy by the people who make bulleted lists for websites. These naming scheme flaws are a boon for the specification list on just about any website, where it is necessary to explain in detail just how "universal" "universal" is.

If "pro" is the opposite of "con", is Progress the opposite of Congress?

And that's why I think we should stick to what most everyone thinks it means due to Apple's promotion of it going back to 2006, which is PPC/Intel code.

When a couple of programmers, or would be programmers, get upset about an exact definition, an issue that was NEVER before raised here that I can remember, I think that we can safely ignore it.

"Everyone thinks it means" is probably a bit too strong of phrasing. If anything, I'd say pretty much nobody knows what a binary is, let alone a universal binary.

It isn't just programmers that interpret "universal" to mean "universal". Hell, a lot of mac users don't even know what PPC is. When they see universal binary, they vaguely recall that it runs on new and old macs.

But I suppose since most people don't know what a binary is, then they'd have to look it up anyway. Which I'll admit supports the general gist of your stance.

The only thing clear in my opinion is that universal binary wasn't the best term ever coined by apple. Why didn't they just say something more like "PPC/Intel Program"? Granted, most people would probably have to google that term too because they owned a "G3", "G4", "G5"... not a pee pee cee.

It doesn't matter if they don't know what it means. Like you said, they'll Google "Universal Binary" and they'll find Apple's definition. And they'll understand the term from then on. Nowhere does Apple say that UB means 32/64 binaries. In fact, and although I haven't checked, I'd be pleasantly surprised if there was a UB definition anywhere on the internet that mentions 64 bit support.

Having to google does matter to some people. As amazing as it might sound, plenty of people won't do so. Most people are around a lot of computer terms they've never googled. Some people will guess something different based upon "universal". I present this thread as evidence.

In my opinion, Apple did a lot better with their "fat binary" term used two transitions ago. I'm curious to see what they land on this time around.

The point is that it doesn't matter if people know what it means. UB means that it runs on PPC and x86. And even if most people don't understand what it means, it doesn't mean that it should be misused.

I don't see anyone trying to change definitions of words in the english language because they're too lazy to look it up. Why should it be different for "UB"?

It certainly does matter if apple is able to successfully communicate which programs are compatible with which of their computers. Transitions like this are largely viewed as a success or failure based on whether they go smoothly for users. Users understanding these terms is a key part of the transition going smoothly.

As for your second paragraph, languages drift over time. Sometimes the changes are an improvement, sometimes not, and sometimes neither. Saying how words _should_ be used is reasonable. Languages also splinter or are used inconsistently between regions. Are you really denying this happens?

How words _should_ be used, and how they're _actually_ used, are both worthwhile topics. Ignoring either is burying one's head in the sand.

Sorry, Kim, but look back at my post above and see what Apple's own utility reports for a 32/64 bit Intel only binary.
Also take a look at this link or type man arch in the terminal (you need to have Developer tools installed I guess).

It definitely states that 32 bit Intel and 64 Intel are different architectures:


It does not explicitly state that 32/64 bit Intel is a universal binary but this is implied from the description. Then again, as I mentioned before, Apple is likely to use different term to avoid confusion, but your comments regarding "anyone trying to change definitions of words in the english language" is stupid. It is not an english word to begin with (there are two words) but a marketing/technical term. What the term means is a matter of convention and this particular term is perfectly acceptable for developers (hence the terms 2-way universal binary, 4-way universal binary) but might be confusing as a marketing term. Note that (assuming SL does not run on PPC Macs) there is some room for consideration reusing the term for 32/64 bit apps that require Snow Leopard, but I don't think it is likely.

Nope, it's not implied.


Why would it be stupid? Apple decided that Universal Binary meant a PPC/x86 fat binary and people like yourself and others here are saying that it can be something else. The fact is that no it can't.


So are you saying marketing terms can't have definitions? Or are you saying that marketing terms can have definitions modified by anyone? I was using words in the english language as the perfect example since words have definitions and so do marketing terms, scientific lingo, technologic lingo, etc.


Come on, dude. You know it would be confusing. That's what melgross and I have been saying in every single post here. Universal Binary means PPC/x86. End of story. Since you're so found of finding implied meanings, you could imply that the Yin Yang represents a balance between PPC and x86. If a 4-way UB was to be marketed the logo would probably have to change to reflect 4 architectures...PPC-32, PPC-64, x86-32, x86-64.


Or Apple could just be absolutely explicit about it and put a fuckin' "64-bit" logo for developers to use for their product. No need to create a 32-bit logo or to specify that the binary is 32/64 considering that in 99.9% of cases the app will be compiled for x86-32 (and if it has the UB logo, PPC-32) which every Intel Mac can run.

Apple created the term Universal Binary for the PPC -> x86 transition. This shouldn't change as it would cause confusion for the people that actually knew and cared about what the term and logo meant.

If an app runs on PPC and x86 Macs then it should sport the UB logo...if it's also compiled for x86-64 then a shiny "64-bit" logo should be put on the box. That would certainly remove *any* confusion about whether the app is 64-bit or not.

That is currently the official Apple definition. But that is far from the end of the story. This thread alone should be evidence of that.

I doubt anyone could point out instances of anyone thinking anything other about a "UB" than PPC/Intel (or, as you say, old/new) compatibility here over the past few years, than a couple of posts from "programmers".

Apple has taught people to think PPC/Intel when thinking UB.

Yes, it's true that there are Mac owners who didn't think about it all all, because they are so very untechnical. But what does that have to do with it? People who DO think of a UB have a way of understanding what it is, and we're speaking to them.

Did Apple make a mistake in their naming scheme? I don't think so.

We have sets, and subsets. The set would be UB = PPC/Intel. The subsets would be PPC 32/64 and Intel 32/64.

Apple talks up the set, not the subsets. It's valid, and for most people enough to know.

Since it's even easier to say 32/64 for Intel code (just as we used to do for PPC 32/64 bit coding), I can't see why a couple of people are so upset about it.

If we refer to it as a UB, then I guarantee that many people are going to ask what we're talking about, because they will be confused, as has already happened here.

Then, we'll have to go off to explain what it means in reference to Intel code as well.

I'd be willing to bet, that with few people bothering to do the right thing when getting into a thread, and reading back to know what's going on, that we'll have to explain this every 20 posts or so. We've all experienced this happening.

It's not worth it. Let's keep UB for PPC/Intel as Apple has done, and use 32/64 bit for Intel.

It isn't incorrect.

As I said, people will come here, get confused, and, if they bother at all, ask us.
And what will we say? Exactly what we're saying now.

Over and over and over and over...

Don't you get it?

It doesn't matter!

You're all hung up on a definition that doesn't really exist, but that you think should exist.

None of this matters for the purpose we're talking about HERE.

And that means for the daily discussion in which many people who don't know about the intricacies of the definitions, and don't really care.

All they want, is an easy to understand, and use, definition that they've been using for years because it's the one we've all used for years, because it's the one APPLE has used for years.

Get over this "it's buried deep down on the developer website" stuff, and come out into the light.

It is the end of the story. But a couple of you have decided to attempt to prolong it with an epilog.

Nice try Melgross, but it takes two to have this conversation. There is no need to personalize the argument.

It is clear what you're stating. There really isn't a need to keep posting the definition of UB. What I've been trying to point out is that not everyone in the world will be aware of the meaning of UB, no matter how many times you post it to this thread.

Obviously you don't believe that there are people who misinterpret the phrase because of the words that comprise it. About this I disagree.

My criticism has been leveled squarely at using the term "universal" as a marketing phrase or consumer oriented specification. Sooner or later the "universal whatever" is no longer "universal".

(Perhaps you've lumped my point in with those from other posters and are arguing against everyone simultaneously. )