svn commit: r333438 - head/sbin/geom/class/eli

[Ian Lepore]

On Thu, 2018-05-10 at 00:13 +0200, Mariusz Zaborski wrote:
> On Wed, May 09, 2018 at 03:05:43PM -0600, Ian Lepore wrote:
> > 
> > On Wed, 2018-05-09 at 20:51 +0000, Mariusz Zaborski wrote:
> > > 
> > > Author: oshogbo
> > > Date: Wed May  9 20:51:16 2018
> > > New Revision: 333438
> > > URL: https://svnweb.freebsd.org/changeset/base/333438
> > > 
> > > Log:
> > >   Change option dry-run from 'n' to 'C' in geli attach command.
> > >   
> > >   'n' is used in other commands to define the key index.
> > >   We should be consistent with that.
> > >   'C' option is used by patch(1) to perform dryrun so lets use that.
> > This seems like a dangerous change. Today I am begining to write new
> > product image creation scripts that involve geli, and I'm going to do
> > so under the assumption that the commands and arguments are not going
> > to suddenly change their meaning on some future OS update.
> > 
> > Hopefully the new meaning of -n implies an argument which is validated
> > as being either 0 or 1, so that any existing scripts that contain -n to
> > mean "dry run" will very likely fail the validation and not silently
> > misbehave.
> > 
> > At the very least, it seems like an entry in UPDATING is required.
> Oliver, Ian - thank you for noticing this.
> I wasn't sure if we need to add UPDATING or release notes because I added 'n'
> option couple weeks ago and wasn't integrated to any stable/release branch.
> Considering that should we still update UPDATING?

Oh, I didn't realize the -n was that new. In that case, I think this
change is probably safe and doesn't need an UPDATING entry. For release
notes, maybe just a note that the attach command now lets you select
the key index.

-- Ian