Comment 2 for bug 719129

This behaviour is as designed.

BACKGROUND INFORMATION
One challenge with dmraid is that it is inconsistent in naming partition devices. Earlier versions never append a 'p' followed by the partition number. Later versions by default always append a 'p' followed by the partition number for partition device paths.

Unfortunately neither of these behaviours match with the commonly accepted partition naming practice:

a) If the device name ends with a number, then append a 'p' followed by the partition number.
b) If the device name does not end with a number, then append the partition number only.

The kpartx command does follow this commonly accepted naming standard, as does parted and udev as far as I can tell.

Earlier version of GParted used to rely on kpartx to create the proper device path names for partitions.

REASON FOR CURRENT BEHAVIOUR
Due to a number of bug reports from users not able to use gparted on dmraid devices when kpartx was not installed, I changed GParted to only use kpartx if it is available. To maintain compatibility with old versions of dmraid, I instruct new dmraid to never use 'p' between the device name and partition number. This permits gparted to work reliably with all versions of dmraid. Hence maximum compatibility is the reason for how GParted currently handles dmraid devices.

When GParted is used to make changes to dmraid partition tables, GParted will ensure that all affected /dev/mapper partition entries are deleted. GParted then creates the new partition entries without a 'p' between the device name and partition number. If kpartx is available, GParted will call kpartx to create device names that follow the commonly accepted partition naming practice.

Hopefully that explanation helps with understanding this design decision.