[parted-devel] Feature requests

[Justin Gombos]

I'd like to suggest some features for GNU Parted.  I noticed that
there are situations where a user is relatively blind when working
with parted as to possible ranges of values for partition boundaries.
For example, a user has a 500MB partition, and without knowing how
small the partition can be, they specify something like "resize 1 0MB
1MB".  Naturally parted determines a value that is more realistic than
1MB, and takes action without telling the user what it found.

In addition to having a disclosure of what actual boundary parted
uses, it would also be useful to inhibit parted from acting on the
information.  ie. it tells me it can only shrink the partition to
400MB, at which point I decide that's not good enough, and go delete
more data from the partition.  Or I might decide to factor in 25MB of
slack space, and shrink it to 425MB.

Those are just the more basic features I would like to see.  Another
feature that would be useful:

Ability to have parted shift the data to the front of a partition
(which is a precursor to shrinking), but without actually taking the
action to change the partition boundary.  This would be useful for
someone who wants to image their partition using dd without including
unused space.  They could use the boundary disclosure to specify a
size to dd that only includes meaningful content.

Currently there seems to be no GNU software that provides a practical
way to image drives without including unused space and forcing that
unused space into the restoration partition, and it seems the features
above would just be a baby step in making that happen.  Currently the
only technique to make a compact image the GNU way is to fill a
storage device with zeros, then take care to write the data
sequentially from the front of the partition, and then read the
partition backwards until a nonzero is encountered to determine the
size to pass to dd.  Or to use parted to shrink a partition that the
user doesn't want shrunk, image it, then expand it back, which implies
unnecessary risk.