Bug#246841: qualified success on iBook G4

[Anton Zinoviev <anton@lml.bas.bg>]

On  3.V.2004 at 22:59 (+0300) Anton Zinoviev wrote:
> On  3.V.2004 at 02:32 (+0100) Colin Watson wrote:
> > 
> > What were you proposing to do with property="$method", if not pass
> > it to a valid_filesystems script? If it's supposed to be passed to
> > valid_filesystems, I'd like to reinstate
> > partman-newworld/valid_filesystems,
> 
> Yes, you are right.  The script partman-newworld/valid_filesystems
> should return hfs if and only if the property=newworld.

No, I wrote this too quickly.  At present there is no package
partman-hfs to take care for the creation of a hfs file system.

I see the following options:

=== 1. ===

partman-newworld doesn't allow a partition without existing hfs file
system to be used as newworld boot partition.  

In order to implement this a check should be added in
choose_method/newworld/choices.  No script in valid_filesystems is
necessary. 

=== 2. ===

partman-newworld takes care to create a hfs file system when
necessary.  This supposes that if the newworld boot partition already
contains hfs file system then the uses will never want to reformat it.

In order to implement this a script commit.d/format_newworld should be
added.  (You can look in partman-reiserfs for example of such a commit
script.)  No script in valid_filesystems is necessary.

=== 3. ===

A new package partman-hfs is created.   Its script
valid_filesystems/hfs understands the argument 'newworld'.

I don't know if it is allowed to install the boot loader in an already
mounted partition but if it isn't then the packages partman-hfs and
partman-newworld should take care of this.

Anton Zinoviev