You miss the point of my post. The filesystem doesn't care how the endians are defined--the rest of the computer does. ...
You're using the word 'filesystem' in a nonstandard way. I use the word to mean the layout of data on a piece of media (within a partition). It contains inode numbers and things like that which are definitely endian-specific. The filesystem is a data structure, not a piece of software or hardware. For example, when formatting a disk we specify we want a FAT32, HFS+, NTFS or ISO-9660 filesystem. That filesystem is the data structure on the media; it has nothing to do with the machine that created it; the media can be moved from one processor/OS to another. The filesystem does not 'write' or do anything else.
Misconceptions, idiocies and other flaws in the post:
Why would they? How could they? FAT32 is patented; Microsoft has allowed third-party implementations so far, but I'm sure they won't be so tolerant when a direct competitor uses the file system as the OS's default. Also, FAT32 is about as modern and feature-rich as anything that we had 10 years ago. Its only two uses these days are compatibility -- pretty much any system on the planet can handle it -- and small media, e.g. floppy disks.
I beleive FAT is not patented, the us patent office threw i out IIRC. My memory is far from perfect though...
The filesystem doesn't care how the endians are defined--the rest of the computer does.
It doesnt matter, what you are saying is irrelevant to the problem.
Eg: there is a field in the filesystem which specifies how big a file is, in bytes. To allow for big files the field is 32 bits longs. Those 32 bits can be written in a number of ways, and in this example they are written as a standard long integer in the cpus native format.
Now, if you take the disk from a big-endian machine, and plug it into a little endian machine, it wont work. The little endian machine will read the field in its own endianness, and it will be corrupt.
There are a number of ways to approach this.
- It is possible to use a standard endianness, and convert the data when you load it. This is what I think MacOS will do with HFS+ disks.
- The fields could be designed to be endian free, just like the files stored on the disk ( actually, they are a repeat of the same endian issue ).
- Seperate disk formats can be used for different endianness, which is what BeOS does. It includes drivers for big and little endian disks, and uses the appropriate one.
The filesystem is not the data, it is how the data is organized and physically moved on/off the media via the HW driver and OS code interactions. That organization and movement scheme is endian independent. Period. The data stored on the media, and it's endianness, is irrelevant to the operation or design of the filesystem.
As noted, until you get the terminology right it's hopeless.
Does anybody know if apple will be using the NTFS file system on the new Intel Macs? I can see this as a major problem for those hoping to run windows on thier mac.
Also, if Apple does do NTFS, will Mac OS X read HFS+ still (my guess is yes). And, if they go NTFS, does that mean that people using PPC macs will be able to read and write NTFS?
Thanks for the help.
You don't know much about filesystem design I see.
First, NeXTSTEP used its own modified UFS filesystem with built-in support for FAT12, 16, Apple HFS, Minix, so on and so forth.
Apple can expand or decrease the types of filesystems it wants to support, while the Hardware is booted into OS X for Intel.
The Bootloader Apple develops can either play nicely, like GRUB (Grand Unified Boot Loader -- GNU Foundation project) or it can be a totally prick like Windows XP Bootloader that loves to install over bootloaders and wipe out access to separate disk partitions.
Whether or not the hardware will recognize other operating systems will be up to folks in various OS Communities to develop, test and certify support to format and install their OS of choice onto a separate boot partition.
With regards to VMware, Wine and other Software Emulators, one will be able to run Windows, via SoftPC inside of OS X, on Intel. It will be emulated but I'm sure the old, "I have invested in so much Windows Software" gripe will still be prevalent even after OS X for Intel is released.
If you want to know more on how operating systems are bootstrapped go:
Whatever OS is shipped with Intel Macs, it will still support HFS+. Look at FreeBSD for a clue - are they using FATxx or NTFS? Hell no. Why would Mac OS X? OpenDarwin runs on Intel based systems and runs on HFS and with Darwin 8.0.1 it runs on HFS+ including journaled volumes. Please!
Hell, since Dominic Giampolo (sp?) has come on board and implemented journaling in Mac OS X maybe they will switch to a BFS file system.
Comments
Originally posted by Hiro
You miss the point of my post. The filesystem doesn't care how the endians are defined--the rest of the computer does. ...
You're using the word 'filesystem' in a nonstandard way. I use the word to mean the layout of data on a piece of media (within a partition). It contains inode numbers and things like that which are definitely endian-specific. The filesystem is a data structure, not a piece of software or hardware. For example, when formatting a disk we specify we want a FAT32, HFS+, NTFS or ISO-9660 filesystem. That filesystem is the data structure on the media; it has nothing to do with the machine that created it; the media can be moved from one processor/OS to another. The filesystem does not 'write' or do anything else.
Originally posted by Chucker
Misconceptions, idiocies and other flaws in the post:
Why would they? How could they? FAT32 is patented; Microsoft has allowed third-party implementations so far, but I'm sure they won't be so tolerant when a direct competitor uses the file system as the OS's default. Also, FAT32 is about as modern and feature-rich as anything that we had 10 years ago. Its only two uses these days are compatibility -- pretty much any system on the planet can handle it -- and small media, e.g. floppy disks.
I beleive FAT is not patented, the us patent office threw i out IIRC. My memory is far from perfect though...
Originally posted by cubist
You're using the word 'filesystem' in a nonstandard way.
Well technically he is, but it's very much "standard slang" to use "filesystem" and "filesystem driver" interchangably.
Originally posted by Hiro
The filesystem doesn't care how the endians are defined--the rest of the computer does.
It doesnt matter, what you are saying is irrelevant to the problem.
Eg: there is a field in the filesystem which specifies how big a file is, in bytes. To allow for big files the field is 32 bits longs. Those 32 bits can be written in a number of ways, and in this example they are written as a standard long integer in the cpus native format.
Now, if you take the disk from a big-endian machine, and plug it into a little endian machine, it wont work. The little endian machine will read the field in its own endianness, and it will be corrupt.
There are a number of ways to approach this.
- It is possible to use a standard endianness, and convert the data when you load it. This is what I think MacOS will do with HFS+ disks.
- The fields could be designed to be endian free, just like the files stored on the disk ( actually, they are a repeat of the same endian issue ).
- Seperate disk formats can be used for different endianness, which is what BeOS does. It includes drivers for big and little endian disks, and uses the appropriate one.
Its really not a big deal.
As noted, until you get the terminology right it's hopeless.
Originally posted by Jwink3101
Does anybody know if apple will be using the NTFS file system on the new Intel Macs? I can see this as a major problem for those hoping to run windows on thier mac.
Also, if Apple does do NTFS, will Mac OS X read HFS+ still (my guess is yes). And, if they go NTFS, does that mean that people using PPC macs will be able to read and write NTFS?
Thanks for the help.
You don't know much about filesystem design I see.
First, NeXTSTEP used its own modified UFS filesystem with built-in support for FAT12, 16, Apple HFS, Minix, so on and so forth.
Apple can expand or decrease the types of filesystems it wants to support, while the Hardware is booted into OS X for Intel.
The Bootloader Apple develops can either play nicely, like GRUB (Grand Unified Boot Loader -- GNU Foundation project) or it can be a totally prick like Windows XP Bootloader that loves to install over bootloaders and wipe out access to separate disk partitions.
Whether or not the hardware will recognize other operating systems will be up to folks in various OS Communities to develop, test and certify support to format and install their OS of choice onto a separate boot partition.
With regards to VMware, Wine and other Software Emulators, one will be able to run Windows, via SoftPC inside of OS X, on Intel. It will be emulated but I'm sure the old, "I have invested in so much Windows Software" gripe will still be prevalent even after OS X for Intel is released.
If you want to know more on how operating systems are bootstrapped go:
http://www.gnu.org/software/grub/
Hell, since Dominic Giampolo (sp?) has come on board and implemented journaling in Mac OS X maybe they will switch to a BFS file system.
Originally posted by piwozniak
the sky is falling, abandon the ship, now now!!!!
lol Calm down people. Wait for Apple to say what it will do. Then go from there. Have a nice day.