[Ti] Apple's True Market Share!

[Paul Russell]

>On Wednesday, December 11, 2002, at 03:44  PM, Paul Russell wrote:
>
>>It's not quite that simple - I can a few issues:
>>
>>(i) Endianness. PowerPC is big endian, x86 is little endian - lots 
>>of additional code will be needed for byte swapping file-based 
>>data, etc
>
>You mean like every vendor that ships dual platform software has 
>already been doing for years? Adobe, Microsoft, etc etc? The 
>Application developers do not have to concern themselves with 
>endian-ness unless they are doing some intensive assembler. Cocoa 
>and the MacOS API takes care of all this and removes it as a concern.

If you design a cross platfom app in the first place with endianness 
in mind then yes, it's a no brainer. But for exisiting Mac-only apps 
it will be a non-trivial amount of work to deal with endian issues in 
file I/O and elsewhere.

>>
>>(ii) AltiVec. I wouldn't mind betting that Quartz and other high 
>>performance chunks o' code are optimised for AltiVec, in which case 
>>they'll run like molasses on x86 hardware.
>
>Well you don't know that do you? Molasses I think would be pretty 
>strong considering that x86 will be pushing 4 GHz by the time OS X 
>is ready for public release. And if QuartzEx was written properly 
>from the start -- as I'm sure it was -- then the processor specific 
>enhancements will be easy to port to another architecture.

I do know that certain parts of Mac OS are already optimised for 
AltiVec - I've stepped into the code when debugging. I can't say any 
more without breaking NDA's.

>>
>>(iii) Carbon. Most apps are still only Carbon at best, not Cocoa. 
>>It will be years before this balance shifts in Cocoa's favour. 
>>Carbon on x86 would be a major enigneering challenge.
>
>I said this already. However Cocoa is the future and most high 
>profile OS X apps, Photoshop 7 being the major exception so far, are 
>or will be Cocoa based applications. Office v.X, the OS X port of 
>Quark, The next rev of Final Cut Pro are all native Cocoa apps.
>
>Like I said, this is at MINIMUM a year or so away, but it will happen.
>

Given that developers have had the Carbon API's for what, 3 years, 
and some of them have /still/ not managed to even Carbonise their 
apps, let alone re-write them for Cocoa, I think it will be a very 
long time before developers would catch up with such a major change. 
Many would probably just not bother - if they've got to do a complete 
re-write then they might as well just re-write for Windows. This was 
why Carbon was so crucial in preventing a mass exodus from Mac to 
Windows when OS X was introduced. Unless Carbon can be ported to x86 
then the same logic applies.

Maybe in 3 - 5 years, but not in the next year or two.

Paul

--