Re: [Nano-devel] should ^W be rebound to 'search_forward'?

[David Ramsey]

Benno Schulenberg:
> Thanks.

No problem.

> Yes.  I will include that change.

Thank you.

> Hmm.  And if I changed "different" to "other", would it be okay then?

Yes, that should be fine.

> Something else.  When ^Q becomes "Start a backward search", shouldn't
> ^W be changed to become "Start a forward search"?  Instead of just
> "Start a search"?  Because if not, and the user actually makes use of
> ^Q to do a backward search, then the next ^W will *also* do a backward
> search.  Which is awkward.  It would require a toggle of M-B to get a
> forward search again. But of course, for the users that are accustomed
> to the current behavior of ^W (its remembering the direction of the
> last search), changing ^W to search always forward will be annoying --
> it would require them to learn to use ^Q instead.

Other search interfaces I've seen (in web browsers, mostly) have
separate forward search and backward search buttons.  And your recent
work with Meta-Up and Meta-Down is in that same vein anyway.  Making ^W
go forward and ^Q go backward makes sense to me.

While making ^W search forward would cause some confusion for people
used to the backwards option, there's a way to mitigate the problem:
bump the major version to 2.9 and list the splitting for forward and
backward search buttons as one of the changes warranting the version
bump (and I would list saving executed command history as *another*
change warranting it, but that's a separate issue).

The rcfile option for backward searches would have to be made a no-op
and deprecated, but the workflow disruption should be minimal.

> Changing the behavior would make for a more consistent interface,
> which is good for new users.  And leaving the rebinding of ^W (to a
> more sensible function) up to the user will not upset old users.

Indeed.

> I don't know what to do, but tend toward rebinding ^W by default and
> incurring the wrath of the habitual backward searchers (those with
> 'set backwards' in their .nanorc file) -- which I hope will be very
> few.

I'd agree with doing it, subject to the above conditions.