[haiku-development] Re: Partition Table Woes (Was: Filesystem corruption and Alpha-1)
- From: André Braga <meianoite@xxxxxxxxx>
- To: "haiku-development@xxxxxxxxxxxxx" <haiku-development@xxxxxxxxxxxxx>
- Date: Thu, 2 Jul 2009 18:15:28 -0300
Em 02/07/2009, às 17:20, "François Revol" <revol@xxxxxxx> escreveu:
Well, x86 or any other, if you want to be able to read foreign media
(usb disks) you need to read those partitionning schemes anyway.
Not that I'm disputing your point, but most FAT-formatted USB keys are
not partitioned and contain a single volume that spans the whole
media. Just like we do with our raw images.
Anyway, that was not what I meant to say.
x86 just has a little more of them, but you'll want to use them from
ppc too.
Here's what I meant: we could target any number of hardware platforms,
but those usually have an OS already installed. Multiple, even. Often
they each have their idiosyncrasies with regards to partition schemes.
Some meant to overcome IBM-PC partitioning limitations. Some actually
"native" to that platform. Geek factor aside, we can't realistically
support them all, at least not on the default distribution.
Sometimes you can't assume a partitioning scheme existing at all. On
most embedded platforms one might not even exist. Still, it's
interesting to boot multiple OSs; take Rockbox for example. Or a
hacked Linux image for the Nokia Nxxx Internet devices. Or... And the
existing partitions might contain volumes formatted with a proprietary
(or encrypted) file system, so you wouldn't share data with them
anyway. Not natively at least.
Requiring a partitioning scheme to boot a disk that's shared with
another OS is actually *added* complexity. The ability to completely
define the boot volume knowing only its absolute offset and a couple
fields of its superblock *simplifies* a lot of things, amongst them
coexistence with "exotic" operating systems on "exotic" platforms.
Resilience to a malformed (or unknown) partition table is just a nice
consequence.
Cheers,
A.
Other related posts:
