Pixel wrote: > Ron Stodden <[EMAIL PROTECTED]> writes: > > Pascal Cavy wrote:
> > > I suspect diskdrake to have a bug in certain conditions. I have noticed > > > several mdk installations where fdisk or cfdisk complains about overlapping > > > partitions, or partitions not ending on cylinder boundary (for ex the first > > > primary ending on 788 for exemple and the extented partition starting at 788 > > > too !). It was the case on MDK90, I dont know if it's still true on MDK9.1B1 > > > ? > > All logical partitions must be in cylinder order in the various MBRs along the >extended chain > > for Windows. Partition Magic does this. > "must be" is truly wrong. There's no such things as a specification > for this. Not a specification, a tradition. DOS & DOS-heritage partitioning tools typically fail if they find the logical chain order doesn't match the physical placement order. > windows tools do create non-ordered logical-partitions linked list I've yet to see one. > > On the other hand, Linux utilities create MBR entries in order of partition > > creation time. NAICT, they, as well as most others, create MBR table entries in order of availability (FIFO). If previously created were primaries A, B & C, using table positions 1, 2, & 3, after which B was deleted, most partitioning tools will reuse table postition 2 when primary partition D is later created even though physically placed beyond C. > > This is true for fdisk and cfdisk, to my knowledge. Windows cannot handle > > this. > wrong. AFAIK windows doesn't (didn't?) like many things, esp. when > there is more than one primary partition (why??) M$ programmers with blinders that don't foresee *and* accomodate the possibility of more than one OS installed per device or system. > > Accordingly, Linux-created partitions cannot be guaranteed to be acceptable to > > Windows. Usually this happens only when a partitioning tool creates a partition that does not start on a cylinder boundary. I'm not aware of any non-Linux tools capable of such behavior. > you mean FAT partitions created under linux? Regardless whether any format at all. > AFAIK there's a pb regarding the boot code which is not written > correctly when windows is installing on a linux-pre-formatted > partition. > > The best solution I know is never to create partitions except with Partition Magic >(which > > does not support ext3, Reiserfs, etc.) or you must dedicate a hard drive to Windows > > partitions only if you need to double-boot with Windows. You could also choose to > > dedicate an entire machine to Windows only. The best solution short of separate disks is to use a tool that understands multiple formats. The more formats the better. Generally this means just about any tool that is *not* distributed with any OS, and includes Partition Magic among many others. Understanding formats is independent of partitioning, and really just a convenience to the user of the partitioning tool. A partition can be created without any real format, having the "format" merely designated in the table but not in the partition's data area. Once in the table, any smart formatter can change the table entry to match whatever format is actually applied to the partition's data area. > > Yes, you can also use cfdisk etc very carefully, making sure that all > > partitions are created and exist in start cylinder order. > wow, i wonder why you write "very carefully" since it's the default > behaviour, unless you mess around quite a lot with your partitions. Maybe by carefully he means avoiding Linux fdisk, which throws out table entries beyond hdx16. ;-) > as for me, i think diskdrake is powerful enough As pertains to its use during installation, I certainly don't. It needs to be smart enough to create HPFS fstab entries when type 07h partitions are in fact formatted HPFS, rather than useless NTFS fstab entries. -- "There's nothing so absurd that if you repeat it often enough, people will believe it." William James Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://members.ij.net/mrmazda/partitioningindex.html
