Leopard gets dose of Solaris ZFS
Apple u-turn on denial
Apple has apparently admitted its next operating system will utilize Sun Microsystems’ ZFS encrypted file system, contradicting earlier denials.
In a display of corporate left-hand/right-hand syndrome, Apple has reportedly confirmed comments made by Sun’s chief executive Jonathan Schwartz last week that Sun’s 128-bit ZFS for Solaris will appear in the Leopard Edition of OS X, due in October.
The only question is how far, and whether ZFS will be offered on a limited basis compared to Apple’s weaker HFS+.
Apple was expected to unveil the news at the company’s World Wide Developer Conference (WWDC) this week in San Francisco, California, until senior director of product marketing for Mac OS Brian Croll reportedly said “ZFS is not happening” in Leopard.
For a moment it looked like Sun’s chief executive had succumbed to one of his trademark unscripted, and factually dubious, news announcements.
The kind of announcement that last year saw Oracle deny Schwartz’s claims that the database giant was adopting Sun’s open source NetBeans development framework.
Apple has now, though, reportedly (see comments) backed up Schwartz by admitting ZFS would be available in Leopard, only in a limited form and not as Leopard’s default file system. The confession came after astute InformationWeek readers pointed out pre-release editions of Leopard are using ZFS.
Unfortunately, despite multiple requests for confirmation and clarification for Reg readers, Apple had not - at the time of writing - budged from its communications policy of not engaging with The Register. [Will someone in China please include a carrier pigeon with the next crate of iPods heading stateside? - Ed]
The only question remaining is why Apple chose not to pull the lever on the ZFS news. Early indications are developers will welcome ZFS over Apple's file system, as this would have provide the ability to securely handle large volumes of data using a system built for the robust, and increasingly popular, Solaris 10 Unix operating system. ®
Why do I need to reformat to make Case-Sensitive?
The ONLY difference between a volume/partition that is formatted as CS or Non-CS is a flag in the directory. Since a Non-CS partition can NOT have file names that would be duplicates if rendered as all lower case, all that SHOULD be needed to go from Non-CS to CS is to flip the flag and update an in-memory tables that show the CS/Non-CS status. An attempt to convert a CS partition to Non-CS is harder and would actually need for the partition to be unmounted (to prevent adds/deletes) while the directory is scanned for duplicate Non-CS file names before the flag is reset or an error message was issued.
lower case madness
or is it fred, do you think the world should be flat too?
maybe we will rewrite the bible and the constitution to accommodate windows and dos users.
i did it all in lower case so it is easier to understand :)
A little more data: Apple now says that ZFS will appear in Leopard but it will be read-only, at least for the moment.
And as for case-sensitivity... yuch. The single dumbest idea I've heard in a long time. The only reason to have case-sensitive filenames is so that you can have 'Readme.txt' 'README.TXT' 'readme.txt' and 'ReAdMe.TxT' in the same directory. Which is an abomination.
Computers are supposed to do things for the convenience of their operators, not the other way around. All of the positive aspects to case-sensitive filesystems are for the computer or the programmer, not for the operator.
Reminds me of a Linux program I once worked on. I was talking to the original author and he said 'Okay, take a look for that in the something.h header file' (I can't remember the real name). So I typed 'vi something.h' (in the project/headers directory) and it opened a file. I said 'I don't see that definition in there,' and he said, 'Oh, it's probably in the capital-something.h'. I said 'What?' and he said 'Try SOMETHING.h, all caps'. Yup, it was in there. He had decided to split the rather long 'something.h' into two files and claimed that this was easier to remember even than 'something1.h' and 'something2.h'.
Oh, and you should have seen the variable names.
I quit the team shortly thereafter.