Anton Zinoviev wrote: > "Abort" does the same as the <Back> button -- partman exits > immediately with error code. It is necessary for frontends without > back-capability. There are such frontends? Shouldn't that be fixed instead of working around it in says that are nonintuitive in properly working frontends? Granted, debconf (not cdebconf) has a couple of oddball frontends that do not support backing up, but it has the CAPB command not return backup if one of those frontends is being used. I'd be suprised if cdebconf doesn't do the same, but it would be easy to fix it if not. > Suppose that for some reason you want to see the > main-menu without any partitioning. Without "Abort" and "<Back>" this > is impossible. Consider this and then apply your patch. I'd run undo, and then continue, I suppose. It might be better to add a boolean question when the user commits the changes, that verifies that they want to destroy their data, with a strong warning and a default of no. > > * If I do choose a device, the first menu item is "Cancel this menu", which > > is redundant because there is a back button. The third item is something > > about dumping the partition info, which might be useful for debugging, > > but I did not understand. The second item is the only one I think anyone > > will want to select. > > I added the "Cancel" item only recently. I think it is safe default > because the second item is very dangerous. I think it would be better to replace this menu with a boolean question with a default of no. > The third item is intended > for bug reporters, I should change it to dump in syslog rather than in > /var/log/partition_dump. It might be good to integrate this with bugreporter-udeb instead. > > * If I create a new partition, its settings default to "do not use", not > > bootable, not mounted, no filesystem type. These are not good defaults and > > mean a lot of work to get a usable setup. > > It is not easy to find the good defaults. Surely we can find some of them. A default filesystem type, default the root partition to bootable on i386, and default the mount point to the next unused mount point in the list. > > * The mount options screen is excessively long, all these defintions should > > be in the select list after the option names. > > I will try this but I am not sure cdebconf will accept it. This > screen shows only once per partman-session. You can do it in debconf, it't just a matter of using a choices list like this: option - description, option - description And then parsing the options out of the results. > > * I cannot seem to figure out how to use partman to create extended > > partitions, at all. > > It creates one automatically together with the first logical partition. Why couldn't I create a 5th partition then? Should I try this again, I'm pretty sure it didn't work. > > * When it writes the partition data, I always get this error from partman: > > > > The kernel was unable to re-read the partition table on > > /dev/scsi/host1/bus0/target0/lun0/disk (Device or resource busy). > > > > That is the usb device I booted the installer from, and I did not tell > > partman to do anything with it (nor does it), so I do not know why it > > displays this message. > > Currently partman doesn't keep track which devices were changed nor > when they have been changed. It reads the partition tables of all of > them and then writes them back even if there have been no changes in > some of the partition tables. Urk. :-/ > > * After setting up 3 1 gb ext2 partitions (/, /usr, /home), and 128 mb swap, > > and letting partman format them, it blue-screened. ps shows > > finish.d/90aptinstall_basicfiles is running. The load average is 0.08, and > > this is a fast p4. I waited for 5 minutes, and nothing. I am able to > > reproduce this easily if there is something I can do to debug it. > > This script is recently added. Please change its number from 90 to > 70. The same change should be done with the script aptinstall_ext3 > from partman-ext3. Ok, I will give this a try. If you don't mind, I will individual bug reports on the problems I found, so we can keep track of their status. -- see shy jo
Attachment:
signature.asc
Description: Digital signature