jpom said:
I have a thought that has been kinda nagging at me about this, maybe somewhere can help me out. Do you think that companies will be more inclined to create software for the mac now?? I mean in my warped mind I see it as being easier to take a program that runs on windows and just "port" it to a Mac running on the x86 arch rather then having to "port" it too a completely different arch. (ppc). Am I on the right track or have i completely missed the ball??
Thanks,
James
Not at all. The system API call's are still different. So you have missed the ball.
NetRyder said:
You're going to be waiting for at least another year.
Just like many others out there that feel they would get a bad deal if they buy a PowerPC G5 iMac at the moment.
jpom said:
Yeah that's what I kinda figured. I know it's not an easy thing to do, otherwise like you said there would be program ported to linux. I could see companies possibly putting the time and effort into porting program to OS X if apple would allow users to install it on "beige box" machines or whatnot, But since Apple is going to only allow OS X on their hardware I don't see it happening anytime soon.
Guess we'll see what happens, should be interesting no matter what.
Thanks,
James
Apple is going to allow Windows to be hacked to run on their box, but they won't support it. It is not locked down to only accept Mac OS X.
The other way around, they are not going to put any special protection in (Who cares, it's unsupported!) to stop people from trying to install it on standard beige box PC's.
NetRyder said:
There are two things here that I'm really interested in finding out more about:
1) As I mentioned earlier, the method used to prevent OS X from being installed on any x86 system. Purely because I want to see it running natively beside XP on my machine at some point, even if it's unofficial and unsupported.
2) Rosetta and the concept of Universal/Fat binaries. This is freaking cool stuff from a technical standpoint!
For others who were wondering how Rosetta (realtime binary conversion) would affect the performance of apps that rely on stuff like AltiVec, the answer is in the universal binary programming guidelines.
Essentially means that any apps that fall under one of those categories above will need to be modified and recompiled to run natively on x86.
1. I doubt there will be anything stopping it, all that will need to be done is someway for it to recognize the hardware, and or a custom kernel. Someone will do it.
2. Rosetta is pretty neat, however any translation thing is bound to have a performance hit. I am expecting this to work the same way that FreeBSD has the Linux ABI on top of it's kernel. It can run Linux binaries on FreeBSD without much trouble at all, as everything for Linux gets auto translated into FreeBSD's system calls, and back.
Geffy said:
From the keynote it sounds like the Universal Binaries are simply the app compiled in both formats which is easy to do, then when the .app is run a loader interrogates the system to find out the type of arch then triggers the right binary from there. If you want to get an idea of how Rosetta works then take a look at the FreeBSD Linux ABI system. Its similar, granted not dealing with an arch difference but the basic theory is the same I think.
I am wondering how much space will go into this. If it causes bigger binaries, or if they are doing it the smart way, and compiling as much as possible for both the same, and only the parts that are different are put in the .app double.
Grandmaster said:
I know...why can't they bring it out sooner? :lick:
"It's a two year transition"
LordOfLA said:
Netryder, I think you can bet on OS X being cracked to run on standard PC kit eventually, but it'll have to wait for someone to buy an Intel mac so we're looking at 1.5-2 years minimum.
Not really, I am getting the dev kit at school, and i am gonna try to install it on a beige box standard intel pentium as soon as possible
At least to try.
NetRyder said:
Yeah, unless the build from the dev kits leaks. Those should be out later this month.
They probably will. Someone will accidently misplace the install DVD, and or one will get swiped from the press where the DVD's are made.
Xie said:
I think opening up OSX to every "beige box" will decrease it's stability. I think thats what of MS's biggest issues with Windows OS's. They are expected to run on such a wide array of systems.
This is indeed the reason that i think Apple wants to keep it to their own hardware/boxen.
NetRyder said:
If I remember correctly, NeXT had a similar system of universal/fat binaries. Seems reasonable to assume the idea arose from there, just like many UI aspects of OS X first appeared in NeXTSTEP. After all, it was Steve's baby.
That is where the fat binaries are coming from indeed. Also, they were used in the beginning of some Unix systems to run on x86, sun used them for a while as well.
As for Rosetta, I'm pretty sure it's very heavily based on QuickTransit. You read the post about that on my blog, didn't you? Apparently, there were collaboration and licensing agreements between Apple and Transitive. I might see if they have some technical whitepapers on Transitive's site. On-the-fly binary conversion from one architecture to another without a significant performance penalty is really intriguing.
Not really that intriguing. It is just the difference as to how the CPU stores it's integers, the translation will take some performance, but not something you will notice.
Also can't forget the Pentium-M line of CPU's. Those little things are gems. Solid performance and low power usage make them pretty much ideal for use in the next generation iBooks, Powerbooks and small form factor desktops like the Mac Mini. The dual-core 'Yonah' chips are really going to be something. 2MB of L2 cache, two CPU cores, and still 8 hours of battery life. None of AMD's or IBM's offerings could hold a candle to one of these things. AMD is firing on all cylinders in the desktop CPU space, but Intel is clearly kicking butt with their mobile processors.
Indeed. I would love for my iBook to last just 2 - 4 more hours. Currently it lasts approx 6 hours, 10 hours would be awesome!
Examination of error reports have shown that the largest percentage of instability issues and STOP errors caused in Windows XP are due to shoddy drivers. You have a point there. But as long as you use decent hardware that ships with solid drivers, you really shouldn't be having problems with stability on a XP system. The Windows 9x line admittedly had stability issues but it was nailed down pretty well in Win2k, and even more so with XP.
It is that fact that Apple does not want to run on other boxen, as they can probably already see the problem with the drivers. Unless Apple defines the interface to the devices, and the manufactures follow them so that it will work with Apple's Mac OS X, we are always going to have bad drivers.
<Machine with hardware> <OS> <Driver>
The driver talks to the hardware with whatever it wants, and hooks into the OS with specified hooks.
What really should happen is that the OS says "To be used for this, you need to have these hooks defined in your hardware, where we can send generic commands to make it do stuff, and it does it, we don't care how you do it underneath"
<machine with hardware> <os>
Then a driver suddenly is not a necessity but rather just an extra add-on to unlock special features or access lower firmware status. For instance:
You have a NIC that dies.
Take old NIC out
put new NIC in
boot up, and you are set
or with Windows currently:
Take old NIC out
put new NIC in
Pray you have drivers, if not roll over to another PC to go download them
stick CD in, install drivers.
reboot and you are set.