The Register USA www.theregus.com is reporting today that Dominic Giampaolo, who wrote the file system for BEOS was hired last week by apple to work as a file system engineer. Does this mean that apple might be working on a big surprise? Maybe for the update after 10.2?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Journaling File System?
- Thread starter wrylachlan
- Start date
- Sort by reaction score
Here's the link. Good article.
http://www.theregus.com/content/4/24485.html
"while Dominic, we are delighted to learn, has subsequently joined Apple as a file system engineer. He started last week."
http://www.theregus.com/content/4/24485.html
"while Dominic, we are delighted to learn, has subsequently joined Apple as a file system engineer. He started last week."
Cool technology.
If Apple can overhaul their filesystem, that would be incredible!! Also, they have to make it work right. Unlike the way M$ doing it....
With Dominic at Apple this really increases our chance of a evolutionary file system happening.
Anyone out there ever used BeOS and tried out queries on the filesystem??
Matthew
If Apple can overhaul their filesystem, that would be incredible!! Also, they have to make it work right. Unlike the way M$ doing it....
With Dominic at Apple this really increases our chance of a evolutionary file system happening.
Anyone out there ever used BeOS and tried out queries on the filesystem??
Matthew
Be had a majpr advantage...
With Be they had *zero* legacy to support. Everything they did was new in the context of a computer OS. The engineers did come from UNIX and Mac/Windows and knew how they worked (or didn't) and so they set out to do something COMPLETELY different.
Apple cannot do this because they have a decade of legacy to support and millions of existing customers with tens of millions of disks all in legacy Mac formats (or DOS or UNIX or whatever).
This is primarily why Rhapsody was an utter failure; throwing out ALL legacy for the sake of the future appealed to no one outside of Amelio/Hancock who actually had to make apps for and use the Mac computer.
Mac OS X definately with its substantially extensible inner core has a better advantage to move past its legacy baggage than Windoze. The file system is abstracted and layered all to hell to read Mac disks and whatnot already in Mac OS X. Journaling as default shouldn't be too far off esp. for a more robust server-capable OS that Mac OS X begs to be.
Mac OS X 10.x is just the transitional phase, the real advanced stuff has yetto come. I'd rather Apple be on time than 10 years ahead of everyone for the upteenth time technology-wise. Newton anyone?
With Be they had *zero* legacy to support. Everything they did was new in the context of a computer OS. The engineers did come from UNIX and Mac/Windows and knew how they worked (or didn't) and so they set out to do something COMPLETELY different.
Apple cannot do this because they have a decade of legacy to support and millions of existing customers with tens of millions of disks all in legacy Mac formats (or DOS or UNIX or whatever).
This is primarily why Rhapsody was an utter failure; throwing out ALL legacy for the sake of the future appealed to no one outside of Amelio/Hancock who actually had to make apps for and use the Mac computer.
Mac OS X definately with its substantially extensible inner core has a better advantage to move past its legacy baggage than Windoze. The file system is abstracted and layered all to hell to read Mac disks and whatnot already in Mac OS X. Journaling as default shouldn't be too far off esp. for a more robust server-capable OS that Mac OS X begs to be.
Mac OS X 10.x is just the transitional phase, the real advanced stuff has yetto come. I'd rather Apple be on time than 10 years ahead of everyone for the upteenth time technology-wise. Newton anyone?
I can't see Apple just making the switch right over into a journaling file system given the difficulties that this would impose (transition). But if I had to pick one person on the earth to go work at Apple's file systems group, it would easily be this man, creator of perhaps the most advanced filesystem ever implemented. Whatever he can lend, however subtly it will affect the OS, it will be a positive thing.
blakespot
blakespot
Re: Be had a majpr advantage...
Okay, if the "problem" is legacy systems, the support for that can simply be programs and drivers. Not direct OS. Drivers are used not by PowerPrint and Hayes. In theory UNIX is designed and set up to interface with EVERYTHING. In theory one could add a PCI card to any Mac and add parallel port, ISA, PS2, Serial, PCL, composite video.
Once the "resource" is available to chooser or a super chooser, then it should be accessable by any compatible application. The purpose of the driver is to make it as accessable to as many apps as possible.
Rocketman
Okay, if the "problem" is legacy systems, the support for that can simply be programs and drivers. Not direct OS. Drivers are used not by PowerPrint and Hayes. In theory UNIX is designed and set up to interface with EVERYTHING. In theory one could add a PCI card to any Mac and add parallel port, ISA, PS2, Serial, PCL, composite video.
Once the "resource" is available to chooser or a super chooser, then it should be accessable by any compatible application. The purpose of the driver is to make it as accessable to as many apps as possible.
Rocketman
This is a somewhat flawed generalization. It is true that UNIX is designed to be able to communicate with hardware, viewing it as a file. But just because UNIX can see it does NOT mean it can interact with it. In order for communication to take place, the software driver must be in place to setup rules and pipes for communication between the kernel and the hardware.
For example, if you add a network card to the system, it doesnt automatically work because UNIX can see it. No matter how many "compatible" apps there are for the network card, they would not be able to access its resources unless there was a piece of software telling it how to communicate and which protocols to use. The network card does not connect to the internet, the driver does. Hardware is hardware is a resource and cannot do anything without being told so.
Then the software asks the kernel for the internet, and the kernel, using the driver, sets up communication.
The reason all of this is relevant is the sheer amount of legacy hardware to be supported will be staggering at best.
Journalling?
Somebody mind summarizing or pointing us less informed folks a a link that offers the pros and cons of a journaling filing system.
Thanks,
Eirik
Somebody mind summarizing or pointing us less informed folks a a link that offers the pros and cons of a journaling filing system.
Thanks,
Eirik
Journalling.
A journalled filesystem is basically just a filesystem that somehow logs activity to the filesystem.
This is a very general statement though...it really implies many more useful features that can utilize the "log" of the filesystem. It is a very sophisticated topic and I'm pretty sure its still evolving.
Matthew
A journalled filesystem is basically just a filesystem that somehow logs activity to the filesystem.
This is a very general statement though...it really implies many more useful features that can utilize the "log" of the filesystem. It is a very sophisticated topic and I'm pretty sure its still evolving.
Matthew
Re: Journalling?
I can only point you in the general direction of some enlightening discussion of same which might impress you. There was an article written by someone (I think one of the authors of the BeOS Bible; Scot Hacker, Henry Bortman, Henry Bartman)discussing his move over to Mac OS X.
http://www.birdhouse.org/macos/beos_osx/
I have only seen beOS in action on TV... it was impressive, we probably would have been equally as happy if Gassee and BeOS was chosen instead of NextStep to become the template for OS X. Of course BeOS was hampered most by having no one develop for them besides themselves... no market momentum. And Jobs wouldn't have been along with his paternal instincts to save the company in other ways.
I can only point you in the general direction of some enlightening discussion of same which might impress you. There was an article written by someone (I think one of the authors of the BeOS Bible; Scot Hacker, Henry Bortman, Henry Bartman)discussing his move over to Mac OS X.
http://www.birdhouse.org/macos/beos_osx/
I have only seen beOS in action on TV... it was impressive, we probably would have been equally as happy if Gassee and BeOS was chosen instead of NextStep to become the template for OS X. Of course BeOS was hampered most by having no one develop for them besides themselves... no market momentum. And Jobs wouldn't have been along with his paternal instincts to save the company in other ways.
Re: Re: Journalling?
And Jobs wouldn't have been along for the ride, as you point out. Apple would not have benefitted w/ Gasee in a position of power at Apple.
blakespot
Be had some notable technologies, indeed. Excellent SMP, "real-time" OS spec, journaling file system, POSIX complaiance, etc. It was close to Unix, but it was not Unix. If Apple had gone w/ BeOS intead of NEXTSTEP, OS X would have been just another proprietary OS. It would _definitely_ not have been as wise a move as going with a solid, robust Unix-based OS such as NEXTSTEP.
And Jobs wouldn't have been along for the ride, as you point out. Apple would not have benefitted w/ Gasee in a position of power at Apple.
blakespot
POXIS Compliant but not quite Unix?
Please forgive my ignorance, but I am confused about BeOS being POSIX compliant. My understanding of this implication is that if said OS conforms to this set of standards such that any application/software made for Unix that is POSIX compliant would be readily portable/adaptable to another compliant Unix flavor with little if any modifications.
The BeOS refugee, Scott Hacker, seemed to imply if not state in his articles that the greatest advantage that NextOS had over BeOS for Apple to base a new OS on was that NextOS would come with Steve Jobs.
Until I read Scott Hacker's article, or perhaps from earlier in this thread, I didn't realize that BeOS was a Unix variant.
I haven't finished reading all of the articles that Scott Hacker wrote and referenced about journaling, application binding, and the object-oriented database nature of the OS/file system in Be. I'm particularly looking forward to an article that speaks of a recommendation for MacOS X with respect to enhancing MacOS X to incorporate some of these features.
I would be thrilled if all you Unix-wise folk would engage in a detailed discussion about this. I suspect there may be previous discussions on this subject in the Macrumors or MacSlash archives; I'll look there after I read these articles.
Also, I am wondering what Apple can do to further multithread MacOS X to improve its performance to rival that of BeOS. Wouldn't it be great in our bitch fights with the Windozers about performance to tout BeOS-like performance on MacOS X? What can Apple do, practically speaking, to improve MacOS X performance as did BeOS.
I'm thrilled that that author of the BeOS, can't recall his name, recently joined Apple's OS team. I hope this indicates that Apple will aggressively apply a lot of BeOS's optimization and architecture. BTW, I imagine only so much can be done due to legacy constraints.
Interesting stuff!!!
Eirik
Please forgive my ignorance, but I am confused about BeOS being POSIX compliant. My understanding of this implication is that if said OS conforms to this set of standards such that any application/software made for Unix that is POSIX compliant would be readily portable/adaptable to another compliant Unix flavor with little if any modifications.
The BeOS refugee, Scott Hacker, seemed to imply if not state in his articles that the greatest advantage that NextOS had over BeOS for Apple to base a new OS on was that NextOS would come with Steve Jobs.
Until I read Scott Hacker's article, or perhaps from earlier in this thread, I didn't realize that BeOS was a Unix variant.
I haven't finished reading all of the articles that Scott Hacker wrote and referenced about journaling, application binding, and the object-oriented database nature of the OS/file system in Be. I'm particularly looking forward to an article that speaks of a recommendation for MacOS X with respect to enhancing MacOS X to incorporate some of these features.
I would be thrilled if all you Unix-wise folk would engage in a detailed discussion about this. I suspect there may be previous discussions on this subject in the Macrumors or MacSlash archives; I'll look there after I read these articles.
Also, I am wondering what Apple can do to further multithread MacOS X to improve its performance to rival that of BeOS. Wouldn't it be great in our bitch fights with the Windozers about performance to tout BeOS-like performance on MacOS X? What can Apple do, practically speaking, to improve MacOS X performance as did BeOS.
I'm thrilled that that author of the BeOS, can't recall his name, recently joined Apple's OS team. I hope this indicates that Apple will aggressively apply a lot of BeOS's optimization and architecture. BTW, I imagine only so much can be done due to legacy constraints.
Interesting stuff!!!
Eirik
I think you only need look at the weak spots in BeOS, then think about what unix is REALLY about, and you will understand why, despite its strengths, it wasn't chosen.
Unix is about networking.
BeOs apparnently really didn't have that together at all, despite any posix compliance.
The people who DID court BeOS? Imbedded systems. Appliance OS. They only saw something small they could burn on a chip for a nickel and shove in a consumer electronics item.
Unix is about networking.
BeOs apparnently really didn't have that together at all, despite any posix compliance.
The people who DID court BeOS? Imbedded systems. Appliance OS. They only saw something small they could burn on a chip for a nickel and shove in a consumer electronics item.