[darcs-users] [patch43] Add --dont-ignore-times counterpart to -... (and 9 more)

Florent Becker florent.becker at ens-lyon.org
Thu Nov 19 09:33:34 UTC 2009


Le jeudi 19 novembre 2009 à 13:22 +1100, Trent W. Buck a écrit :
> Ganesh Sittampalam wrote:
> > On Thu, 12 Nov 2009, Trent W. Buck wrote:
> > 
> > >Florent Becker <florent.becker at ens-lyon.org> writes:
> > >
> > >>The object of these patches is to add a --dont-frobnicate option for each
> > >>--frobnicate option. That way, you can put "command frobnicate" in your
> > >>prefs file, and still use --dont-frobnicate when you need the default
> > >>behaviour.
> > >
> > >I'm glad argument consistency is getting some attention[...]
> >
> > They look ok to me as a short-term improvement that may well be
> > obsoleted by a proper reorganisation of the command-line flags.
> 
> I agree with that analysis.
> 
> > Trent, are you happy for these patches to be applied?  The only
> > possible reservation I have is that they introduce yet more
> > inconsistent language into the way that command-line arguments are
> > negated, and might create more of a legacy to support in future.
> > Does anyone else have any opinions on that?
> 
> My only reservations are those you outline above.  The support burden
> is particularly significant if this goes into the 2.4 release, but the
> larger option overhaul (discussed in other threads) doesn't.

I agree with you both on that point, and in hindsight, I should have
made the name more coherent. I propose we decide now on a standard
negation. I suggest --no-foo (even when foo is a verb). In a very
short-term followup patch, I rename all the new flags from that bundle
to --no-whatever, and add --no-whatever aliases to all the other
negative flags. Then we can start thinking about doing something
smarter, which should come with a better abstaction for flags, and so
probably after 2.4. I'd still like to see that bundle applied soon,
because it's a dependency for more useful stuff like --reverse.

Florent




More information about the darcs-users mailing list