Re: Questions, additions, and ideas.

[Sven Luther]

On Thu, Dec 30, 2004 at 01:10:26PM -0600, address@hidden wrote:
> > Notice that it won't work always, as seems the case on solaris partition
> > tables which need to be written for partitions to be created, or something
> > such. I have a debian bug report open about that, but i am not sure i fully
> > understand the implications of it.
> 
> libparted will still remain the exact same.  See I don't see that being a
> problem. Because, when users are making the changes they are only modifying 
> the
> struct in memory no actual data is being written therefore the only thing we
> need to do is check to make sure it is a sun disk in this case and then when 
> the
> user is ready and executes the write command we write everything to disk in 
> the
> proper order. 

Sure, that is what debian-installer's partman does. I still think a proper fix
in libparted itself would be better though.

> As far as the "undo" command working when the interface loads we load a disk
> struct and modify it.  See the partition table on disk has not changed so if 
> the
> user feels like they have made some kind of mistake and issues a "undo" then 
> we
> dump the current struct and reread from disk.(Because the disk has not be
> altered. If I am misunderstanding please correct me.

How do you handle stepwise undo ? you copy the in-memory structure for all
steps ? 

Friendly,

Sven Luther