On Sun, Jul 20, 2008 at 01:39:59PM +0200, Frans Pop wrote: > Consistency between D-I and commands given from shell is another factor to > take into account. Fact is that LVM by default uses binary-based units > while e.g. disk manufacturers and fdisk use SI units. Partman is > currently consistent with that. It may not be internally consistent, but > it may also be just what some users (e.g. experienced sysadmins) expect. LVM tools default that display sizes to binary units instead of SI units. So on a command line, IMHO, an experienced sysadmin would use other LVM tools to create PVs, VGs and LVs together with the ones that would display the current status. As most people rarely bother to specify "--units M", I don't really see that as an issue. > I'd much rather see an inventory of what partman does _as a whole_ and a > proposal/discussion how we can make it consistent and clear to the user > what units are used/expected/displayed for partman _as a whole_ than to > introduce such "random" changes based on a single aspect of the problem. partman reference seems to be given by longint2human() and human2longint(), defined in lib/base.sh. Which is why I decided to use them to as the basis for calculation in the proposed patch. Cheers, -- Jérémy Bobbio .''`. lunar@debian.org : :Ⓐ : # apt-get install anarchism `. `'` `-
Attachment:
signature.asc
Description: Digital signature