Re: [Bug-parted] Re: parted 1.3.x errata

[Andrew Clausen]

Damien Genet wrote:
> 
> Salut,
> 
> Andrew Clausen wrote:
> > > the func should, maybe, return true to PED_PARTITION_HIDDEN
> > > only for FAT, and false for PED_PARTITION_BOOT when
> > > partition is extended.
> >
> > Hmmm.  I guess I should do this :-/  It's ugly.  Added to TODO,
> > pending further sleepness nights.
> 
> I join a diff, code is untested (Gtk-Parted is still broken due to
> libparted 1.3.x port), modifications should also be applied
> to disk_pc98, but I don't know how much this architecture
> (is it the right word?) differs from a standard PC.

s/architecture/disk label/ ?  architectures is a general word, so it's
OK too, I guess.

Anyway, it doesn't need to be applied to pc98, because *all* partitions
on PC98 can be hidden.

> > > Is it possible to know what flags would be available before creating
> > > a partition?
> >
> > Interesting question.  Perhaps I should make it policy, that if you
> > pass NULL as the "part" argument to ped_partitition_is_flag_available(),
> > then it should return some reasonable values.  Added to TODO.
> 
> the problem is that if part is NULL, you can't know on what disk type
> you wan't to add the partition... and so you can't know
> what flags are available.
> maybe direct use of disk->type->ops->partition_is_flag_available()
> should be better, or this call could also be in a
> ped_disk_is_flag_available(PedDisk *disk, PedPartitionFlag flag)
> funtion.

Good point.

Hmmm.  This is complicated.

I've got another idea: when you're asking the user about adding a
partition, you actually ADD the partition (ped_disk_add_partition)
from the beginning.

When the user changes what fs type they want, you call
ped_partition_set_system() immediately.

Then, you can get flags, etc.

This is much cleaner, and was subconsciously what I was thinking
about for PedDisk, anyway.

Do you understand what I mean?  I should document this nicely,
somewhere.

Thanks!
Andrew Clausen