[pacman-dev] [PATCH] Add short options for mark as deps or explicit

[Dave Reisner]

On Thu, Mar 07, 2013 at 09:29:55AM -0600, Dan McGee wrote:
> On Thu, Mar 7, 2013 at 9:08 AM, Sébastien Luttringer <seblu at seblu.net> wrote:
> > On Thu, Mar 7, 2013 at 6:32 AM, Allan McRae <allan at archlinux.org> wrote:
> >> On 07/03/13 15:30, Andrew Gregory wrote:
> >>> On 03/07/13 at 02:51pm, Allan McRae wrote:
> >>>> On 07/03/13 06:31, Dan McGee wrote:
> >>>>> On Wed, Mar 6, 2013 at 2:19 PM, William Giokas <1007380 at gmail.com> wrote:
> >>>>>> On Tue, Mar 05, 2013 at 02:03:14AM +0100, Sébastien Luttringer wrote:
> >>>>>>> The main (only) purpose of -D is to be able to change packages installation
> >>>>>>> status (deps or explicit). Having a short form offer a similar experience that
> >>>>>>> other main pacman option (e.g. Su).
> >>>>>>>
> >>>>>>> Signed-off-by: Sébastien Luttringer <seblu at seblu.net>
> >>>>>>
> >>>>>> The --asdeps option for -S and -U does not have a shortopt. In my
> >>>>>> worthess opinion, this is a bad idea, as -d for those operations is
> >>>>>> --nodeps.
> >>>>>
> >>>>> This was my thought as well. If we are willing to use a shortopt, it
> >>>>> should apply to ALL top-level operations in the same fashion (or be
> >>>>> rejected completely), and not mislead. -Q/--query match this criteria,
> >>>>> but currently -d for -U/-S would be totally unexpected. So -1 from me.
> >>>>>
> >>>>> I have consciously made decisions over the past 3 years to not add new
> >>>>> shortopts unless they are universally applicable, so this would be a
> >>>>> step against that. If we were to do this, we would want to remove the
> >>>>> -d shortopt for --nodeps in the next release, and then add these in
> >>>>> the following release. However, this is cumbersome as `--nodeps
> >>>>> --nodeps` is really silly to type out as we allow this option to be
> >>>>> passed twice for even more dep-ignoring behavior.
> >>>>>
> >>>>
> >>>> I made the decision to take this based on:
> >>>>
> >>>> 1) it would be good to have a short options
> >>>> 2) the short letters made sense
> >>>> 3) the current usage of -d/-e in -Q is fairly similar
> >>>> 4) the current usage of -d in -S is an operation that is unrelated to -D
> >>>> so will not cause confusion.
> >>>>
> >>>>
> >>>> People manage to understand that -Sd is different from -Qd.  Why the
> >>>> need to enforce consistency when there is already none?
> >>>>
> >>>> Allan
> >>>
> >>> I think that the problem is not just that -d means different things for
> >>> different operations, but that --asdeps and --asexplicit shorten differently
> >>> based on the operation.  A user would likely see that --asdeps shortens to -d
> >>> with -D and assume it to do the same for -S because --asdeps is a valid option
> >>> there too.  A short option may mean different things for different operations,
> >>> but all operations that accept a particular long option should use the same
> >>> short option for it.
> >>>
> >>
> >> OK.  I separate out the --asdep for -S/-U and -D mentally because they
> >> are doing completely different things.  But I see the point.
> >>
> > ok, we can use -e to explicit deps, as it's not used on -S/-U/-D
> > (which have all --asexplicit).
> >
> > -d is used by --nodeps, do you have a suggestion for a short option
> > for --asdeps?
> 
> Not every operation deserves a shortopt; these are used so much less
> than other operations that I don't feel the loss of self-explanation
> is worth it.
> 
> -Dan
> 

+1. We've made a point of explcitly *removing* the shortopts for
infrequently used or dangerous options (-k no longer exists for
--dbonly, and -f for --force has gone away). I tend to think that
anything involving the -D operation can be destructive and infrequently
used. Let's not go in the opposite direction of this.