Bug#1111577: partman-partitioning: Missing information about default units (GB since trixie)

[Holger Wansing <hwansing@mailbox.org>]

Hi,

Pascal Hambourg <pascal@plouf.fr.eu.org> wrote (Wed, 20 Aug 2025 15:30:45 +0200):
> This is part of the issue. Supported and default size units are defined 
> in partman-base and independent message strings advertising them are 
> defined in partman-lvm and partman-partitioning templates. IMO this 
> design is sub-optimal and prone to inconsistencies:

Ok, now I see the whole picture.

> - partman-lvm template advertises supported units and the default unit.
> - partman-partitioning template only advertises the most common 
> supported units and not the default unit.
> - When support of binary units was added in partman-base, partman-lvm 
> and partman-partitioning templates had to be updated to reflect it.
> - Support of P(i)B and E(i)B was added to partman-base at the same time 
> but partman-lvm was not updated to reflect it.
> - When the default unit was changed to GB in partman-base, partman-lvm 
> template was not updated to reflect it.
> 
> I do not expect other changes in this area any time soon so my comment 
> is probably coming too late, but IMO a better design would have been to 
> define the message string advertising the supported and default units in 
> partman-base templates, and insert it with db_subst in message strings 
> of other partman-* packages which need to display them.

Since we need to make changes to the templates, we can also make other
changings on the design as well, if we agree that it's worth it.

If I look at the current situation:
We have partman-partitioning with its strings mentioning "New partition 
size", and we have partman-lvm with strings like "Logical volume size".
To form *ONE* message string which can be used for both packages, we would
need a cover term, which works for both, partitions and logical volumes.
But I think that would be difficult, or if we find one, it is not easily
understandable to the user?

So I guess we need to keep these two sorts of strings separately (with 
improvements on several strings), but having all strings together in one 
package?
Is this, what you think of?

I cannot completely overlook, if there are any drawbacks with this BTW, 
being the reason why it was made the way as it is now.

One point would be, that the preseed files which work for the users for many 
many years, do no longer work then (regarding partitioning) ...


Holger

-- 
Holger Wansing <hwansing@mailbox.org>
PGP-Fingerprint: 496A C6E8 1442 4B34 8508  3529 59F1 87CA 156E B076