Re: disklabel(8) allows addressing non-existent space

[Matthew Mondor <mm_lists%pulsar-zone.net@localhost>]

On Sat, 31 Dec 2011 23:47:00 +0100
Julian Djamil Fagir <gnrp%komkon2.de@localhost> wrote:

> Hi,
> 
> the following patches are my supposal for:

>  * Adding a batch mode (option -s)

I don't necessarily object, but does the -i interactive mode require a
tty?  Otherwise isn't it similar in functionality to -s ?  We also have
the -R option to apply a label from a file (I personally used the
latter often).

>  * Adding an option to set partition alignment rather than just cylinder or
>    sector rounding

I think that this is useful.

>  * Add option to specify percentages. This is somewhat unclear, as it will
>    currently take d_secperunit, which won't work with mbrs.
>    We could add some magic to check whether we have an mbr, but this won't be
>    very clean.

I doubt that I'd personally use this functionality, but I'm sure some
users will.  For instance I wouldn't necessarily make /, /usr, /tmp
or /var much larger than necessary because the disk is huge, and
wouldn't want them to be too small because a disk is small.  I instead
am pretty confident about the needed sizes, depending on the machine's
use.

> Please have a look at the code and comment.

I unfortunately couldn't fully review the diffs for quality yet.
Unless others comment about them I'll look at them more tomorrow
(unfortunately, I also expect some internet downtime soon for an ISP
switch, however, I'm actually surprised the link is still up).

Thanks for working on this.

Just as a note, since this domain interests you, disklabel can only
handle partition sizes restricted by 32-bit, while gpt(8) can handle
larger sizes and is likely to become more widely used in the future.  I
personally never used it, but it's also possible that you'd find that
gpt(8) also requires user-friendliness improvements...
-- 
Matt