Re: I really do like OS X but . . .

foo@xxxxxxx wrote: 
On Sun, 01 Jan 2006 18:07:15 -0500, TheLetterK
<theletterk@xxxxxxxxxxxxxxxxxx> wrote:


foo@xxxxxxx wrote:

On Sun, 01 Jan 2006 16:13:23 -0500, TheLetterK
<theletterk@xxxxxxxxxxxxxxxxxx> wrote:



foo@xxxxxxx wrote:


On Sat, 31 Dec 2005 19:06:23 -0500, TheLetterK
<theletterk@xxxxxxxxxxxxxxxxxx> wrote:




foo@xxxxxxx wrote:



On Sat, 31 Dec 2005 17:46:00 -0500, TheLetterK
<theletterk@xxxxxxxxxxxxxxxxxx> wrote:





George, we've been over this time and time again. XP (and Vista) have
_nothing_ to do with DOS. It's an emulation environment, just like
OS9 (Classic) is an emulation environment in OS X. Just as OS X's
origins are FAR from OS 9, so are XP's origins FAR from DOS. You see
a command line window and you think that's DOS. You're wrong. PLEASE
stop endlessly repeating this.

I don't think he was talking about that. Or, at least, I hope he wasn't. I think he was pointing out that Microsoft has to ensure backwards compatibility with the older MS-DOS based versions of Windows, which introduces a number of problems.


Huh?

I'm not terribly surprised that you don't consider problems not of a technical nature.


Can you be more specific?

Most backwards compatibility problems come from third party developers being slow to adopt new technologies, not problems with the technologies themselves. So, while your latest version of the OS might have astounding new features, they aren't often used because third parties either had to implement those features on their own years ago, or they just don't want to go there yet. That's not even considering the fact that most people don't upgrade software that works well enough--so lots of people (and companies, for that matter) are still using old versions of the software, that don't take advantage of many of the very nice features and APIs included in 2000 and XP. In other words, there is still lots of code on Windows that was originally written for Windows 98, that people try to run on Windows XP, and that causes problems. And there's not really anything Microsoft can do about it.


And what, if anything, does that have to do with George's comment?

The legacy code in question, is coming from the MS-DOS days. 


XP has been here for 5 years now and has hundreds of millions of
users. Tiger has been here for about a year now and has a few million
users.  Who do you think is going to get better, higher quality
coding?  (Please note your arguments about x86 tuning while you're at
it.)

Yes, with lots of work, the limitations of Win32 can be somewhat overcome. That still has nothing to do with my point.


Sorry; I don't buy your argument, and I also believe George meant
exactly what he wrote (and what I interpreted) re:DOS.



I can agree with this - it's even worse on the Mac, with Apple
changing APIs every 12 months.



The NT VDM is well-documented and well-known, and I doubt MS has
anyone working on it anymore; why would they?





Like their reliance on Win32 instead of .NET...


Huh? Win32 is just an environment, one of several. Don't like it?
Put in another one.

Win32 is also the name for a set of APIs. The only real competition on Windows right now, would be from .NET or Java. Still, Win32 is, by far, the most commonly used.


So target for it if you want to.

That's my point--they're legacy APIs, with lots of trash left over from the 'MS-DOS days'. .NET needs to replace it, badly.


Code for whatever you want. 



http://mipagina.cantv.net/jjaguilerap/w2k_arq.html


I was speaking of the APIs, not the kernel and overall system architecture. Win32 is a throwback to the MS-DOS shell days, and should disappear.


Why?

Because it uses lots of legacy code, and is the origin for lots of the problems people see on Windows today? 


What specific problems do I see, and what does this have to do with
George's comment?

Instability, strange UI quirks, lack of integration in some applications, overly redundant design, etc. 


I can't agree with your argument.  Win32 is not a 'throwback to MSDOS
shell days', and the rest of your argument is similarly nonsensical.

Yes, yes it is. I imagine that if Microsoft could push a button and make Win32 go away, they would have wanted to do so last week. If Microsoft could drop Win32, the Windows userspace would suck much less than it presently does.

Nevertheless, George doesn't know a thing about any of this, so I have
a tough time believing this is what he had in mind.


That's what I thought he was saying, your mileage obviously varies.



Certainly, Microsoft's insistence that each new edition of Windows maintain a general compatibility with pre-existing software must compromise what they are able to do with new versions of the OS, 


I don't know about that.  What issues are you talking about?






and most assuredly handcuff the development team and limit new features. 


I don't see XP as feature-crippled at all.  What's the issue you're
talking about?

XP is the very definition of feature crippled.


How so?

Microsoft took an excellent kernel like NT, and married it with absolute *** for a userspace.


This is getting tiresome. *How so*?

Let's see... The primary API on the system was a piece of *** 10 years ago, and is now a badly aging piece of ***. The system's GUI is a mess of self-conflicting paradigms and 'fancy features'. The CLI is essentially worthless. It's really not a capable system out of the box, as it is missing reasonable applications for even basic functionality. There's not really much in the way of development tools, and scripting is simply a joke without at least python. Need I go on? 


Most people disagree with you, and it's the most common development
environment in use today.

It doesn't even ship with a compiler, and your going to try to tell me how great it is for development? 


If you need a compiler, you buy one.  By your horrible argument, the
PowerMacs don't ship with a monitor, so thus they're terrible for
anything requiring a monitor?  The PowerMac doesn't ship with
Photoshop; is it horrible at everything using Photoshop?  Don't be
silly.

I guess Im' just so used to having a compiler as part of the default install, that I consider such a philosophy to be simply unacceptable. After all, *almost every other modern platform out there ships with an ANSI C compiler as part of the default installation*. Would it really have been that difficult for Microsoft to include something as basic as DJGPP and emacs with Windows?



If you're accustomised to Unix though, it's
different.


Yes, yes I am.



This is likely why each new version of Windows ends up being just a mild facelift with no real improvement in operability or usability.


And this from the Mac guy? 10.2 to 10.3 to 10.4 - talk about minor,
incremental changes for $130 each!

You do realize that Microsoft also charges for major point updates? Unless XP was free for Windows 2000 users? If so, I would like a refund...

Windows 2000 == NT 5.0
Windows XP == NT 5.1
Windows Server 2003 == NT 5.3

Would it help if Apple had completely removed version numbers from their marketing campaigns, like Microsoft does?


$130 for this year's OS. Every year. Only Apple.

And until recently it was '$300' for Microsoft every 2-3 years. I 


1.  Upgrades are available for far less than that.  Heck, new
*machines* with the OS are available for less than that!

They weren't until recently.


It's been a few years...longer if you include AOL rebates. :)

New machines for $299 didn't start appearing until late 2004/early 2005. They've always been cheap, but the prices have been dropping. 


Machines were free or nearly free several years ago via AOL rebates; I
seem to recall that starting around 1999.

Deferring the cost to AOL, does not make the cost disappear. If anything, it was more expensive to go the AOL route...



2.  Software still works on Windows 2000; it isn't as if people are
forced to upgrade.  The need to upgrade on the PC isn't nearly what it
is on the Mac due to the more mature Windows APIs in use.

I'm still using Panther, so it appears as if I've managed to avoid said update. But, of course, Mac users have no choice but to upgrade...


Of course. :)



definitely prefer Apple's release model, even if it does cost more. There's nothing I hate more than having to wait 4 years for an OS update.


Why?

Because I'm a home user and a tinkerer. I can't stand some company going on about amazing new features and such, while not releasing the product for years on end. At least on GNU/Linux, I get such new technologies tricked down and I can try them in beta form without having to sign an NDA or buy an MSDN subscription.


Fair enough. 



As a developer it's vastly easier to support a simpler and more
mature release schedule, and I have to believe it's easier from a
support perspective too (both on Apple's part and on IT's part).

Nonsense. Though I admit, it might well be easier for Microsoft given their pathetic handling of packages...


Not nonsense at all. You're essentially saying it's easier to support
multiple releases with multiple different technologies in them that
act in multiple different ways. That's absolute nonsense.

No, I'm saying it's easier to require specific versions of the libraries you depend on. E.G, if you are building for GTK 2.8, require GTK 2.8. That way, you don't have to stall development of *everything* to simplify development for some developers. Let package managers handle installation of the required libs.




You make the mistake of most Windroids. You assume that because Macs have lower CPU clock speeds that the machines themselves are somehow slower, 


They are.  Particularly once you include $ in the equation.  C'mon
George, this is an old one too.






or that if they are, that the speed difference is enough to make any difference in productivity. 


It is.  Else why buy new Macs?






These are all strawman arguments. There is no real evidence that Macs are slower accross the board than are PCs, 


Oh please.  Obviously the evidence of 'benchmarks' is too much for
you.

Benchmarks tend to be mixed, and I think there are causes for benchmark differences that most people don't really consider. Like poorly optimized code.


So clearly every application is poorly written.

Every application? No. But a large majority of them, particularly professional applications that receive a great deal of effort at being optimized for specific hardware and Windows.




Or else, as is vastly more likely, the CPU's slower.

I don't think there's any real reason to suspect that the G5 is substantially slower, considering the fact that many highly optimized applications *do* perform similarly to their Wintel counterparts.


And some don't. And generally speaking, more don't than do.

Which is probably because... more time is spent optimizing the x86 version. A lot more time. For an example, see gcc--it's been optimized for x86 for the last 20-something years, but the PPC port didn't even support altivec until relatively recently. It even makes up part of the SPEC benchmark, which G5s do fairly well at despite the poor optimizations. 


You compare optimization for a general CPU architecture (x86) vs.
optimizations for a single CPU?  Get real.

I was using it as an example of how far behind the PPC port was. If your even going to try to claim the PPC port of gcc is anywhere near as good as the x86 version, I have some beautiful land in central Florida you might be interested in... 


If Apple's compiler of choice has a problem, take it up with Apple;
don't blame me.


The real question is how
quickly after introduction of a CPU does that CPU's specific
functionality and features get supported by gcc.

In the case of the PPC port, that takes a long time. IIRC, there's little G5-specific improvements in GCC right now. Certainly nothing like what has been done for the Athlon 64s and P4s. 


A shame.  Makes you wonder why Apple chose it, if it's so horribly
bad.

Historical reasons, and they had no free alternative until the release of the G5. Knowing what we know now, it probably had something to do with having the x86 version of OS X in the back...

 Did Metroworks(?) tank, or were they just unwilling to code a
compiler for OS X?

Couldn't make any money at it. Xcode/GCC may have been poor on PPC, but it was good enough for lots of the folks Metrowerks used to sell Codewarrior to. And there's always the option of using IBM's compiler...



If it's slow,
Apple/Motorola/IBM would be to blame, if gcc support is a critical
feature to you.   Now that OS X is Unix based obviously it's more
important than it was when Symantec / Codewarrior did the compilers.



and there is certainly no evidence that Macs are less productive. But there is plenty of evidence that Macs are the MORE productive of the two platforms. 


Not in a long, long time have I seen Gartner and similar for modern
Mac systems.






That's not surprising, but what is surprising is that you feel the need to "go back to Windows" to get your work done. I do all my work on the Mac. I produce documents with strong graphic content on my Mac on a daily basis, yet my entire corporate clientele is Windows based. They never know the difference. The idea that "business" MUST be carried out on PCs is largely a myth. 


...because all businesses work just like you do, right?






I think that you'll find that Intel Macs won't be significantly faster than PPC Macs. 


I hope you are completely wrong on this.  Just speaking of the new
Pentium-M designs alone (Yonah?), I believe this is another one of a
seemingly endless string of stupid and wrong comments from you.

I have no doubt they will be faster for many uses. Though I don't know if moving the Xserves over is a good idea.



Why?

The only reason anyone even looks at the Xserves, are because they use G5s (which are very well suited to high performance clustering on the 'cheap'). Without that advantage, no one would give them the time of day. 


A particular label on the CPU isn't an advantage in and of itself.
What advantage does that G5 give you?

It has extraordinarily good FP performance and accuracy, at a price most major universities and medium-sized businesses could afford. For clusters, this is a very good combination. 


Fair enough, but I suspect for most universities and businesses that
the AMD designs are going to be more common.

They don't seem to be.


What is the most common, iyho? 

Probably Xeon clusters running GNU/Linux.



How is it high performance
clustering on the cheap?


See above.



Do you do a lot of said clustering, and what
do you do with it? 

Me? No. I'm just pointing out why the G5 is a good processor for clustering.



The G5 isn't particularly faster than a Intel chip, but it's not slower either. 


This is funny, George.  Really.






Again, you are making assumptions based on no evidence. For instance, where did you get the idea that Intel-based Macs will be any cheaper than PPC based Macs? Just because there exist ultra cheap Windows boxes made in China with virtual slave labor that are often sold by big PC manufacturers like Dell as loss-leaders for practically nothing doesn't mean that Apple will go the same route. 


You're now suggesting Apple doesn't outsource to China, to Quanta and
Asustek, the same guys that make most of the PC hardware?  What
fantasy world DO you live in, George?

Indeed, they do. However, the Intel Macs will probably not drop in price. If anything, they'll be more expensive.


Why?

I don't see why the price would drop, when Apple switches to a manufacturer that charges even more. Apple has no bargaining room with Intel, like they had with IBM and Freescale.


You mean like saying they'll go to AMD?

And if they went with AMD, would Intel be able to sell the processors they had? Yes. With IBM and Freescale, they *have* no other substantial market for desktop processors. Apple was the only real buyer. 


And IBM/Freescale didn't make substantial money out of Apple - both
made more of their CPU money elsewhere.

Yes, but that doesn't mean they would want to let those processors sit in a warehouse with no one to buy them. Even if it was only a small portion of their profits, it was still profit.


Then you'd think they'd invest in better compiler technology.

IBM has a better compiler available. Apple still uses GCC--and because of this, lots of third party developers do too.

Hmm...
guess it just wasn't worth the money to them. Anyway, another reason
to move to Intel... 


Yes, Apple's still a bit
player with Intel, but at least now they have the flexibility to
easily use AMD designs if they want.

They probably signed an exclusivity contract with Intel (like Dell did), for price benefits on both processors and mobo chipsets...


Maybe. Maybe not. Do you know?

It seems to be Intel's standard operating procedure. IIRC, I read something about Apple and Intel signing a deal which assured that Apple would buy processors from Intel until some specified date in the future. A quick search brings up lots of other BS relating to the Apple-Intel deal, and I don't want to sift through it right now.



LOOK at a G5 Tower. Take the sides off of it and peer inside. Now, do the same for $299 Dell special and you'll see the difference. I'm not saying that everyone values the difference between the build quality of a $300 Dell and a $2000 Mac tower, but, you see, APPLE DOES. And as long as that's the market that they feel comfortable participating in, that won't change. And there is nothing to indicate that Macs won't always be more expensive, often much more expensive, than at least SOME similar Windows PCs, processor costs notwithstanding.


One is silver, the other black. And they both sit below a desk
identically.

Surely even you realize that there are other differences between the case of your average PC and average PowerMac G5. Apple spends a lot more on cases than most OEMs.


Their loss.





As for the PPC becoming obsolete; if it happens at all, it will be after the end of this decade and by then we'll all be ready for a new computer. But ponder this. It will also, likely be the end of the decade before Mac applications support and device driver support has migrated away from the PPC processor to the Intel processor, so the PPC Macs will remain the most viable products in the line until at least that time. Remember, we Mac users have been down this road before (68K to PPC) and we know how it unfolds. We know from experience that being an early adopter is not the smartest move in such cases. If you simply must have the earliest Intel Macs, be prepared to run PPC applications in emulation and to have virtually NO driver support for many printers, scanners, and other hardware for a couple of years at least.


What's new?

The lack of support.


Again, what's new?

Again, the lack of support.


Not new.

Yes, it is new.


You're kidding me.

Not really. When something says it's Mac compatible, it's usually Mac compatible. This might not be the case with the Mactels.



My next Mac will not be an Intel Mac for the reasons stated above, but likely the one AFTER the next one will be Intel based. I'm not planning on going Intel until the next decade.


You'll, as usual, be the laggard, in this case technically. 





--
"There is nothing I understand." - ***
.