Re: Regression affecting completion with NO_CASE_GLOB

[Bart Schaefer <schaefer@brasslantern.com>]

On Thu, Jun 2, 2022 at 5:28 PM John Heatherington
<jheatherington@gmail.com> wrote:
>
> After bisecting, I found a regression originating in the commit:
> d82604843bf2b743e04666d4644dd109831252f7
>
[...]
>
> The change was originally introduced to correct cfp_matcher_range,
> which doesn't call bld_line, so it seems like this could be fixed by
> putting the burden of adjusting the index on bld_line instead.

Indeed, it makes sense that if the semantics of an argument to
pattern_match_equivalence() has changed, then all callers need to
adapt to that.

There are two ways to go here:  Revert the pattern_match_equivalence()
change and alter cfp_matcher_range() to add one to the index that it
passes, or change the other two callers to decrement their indices by
one.  I tested the latter and it works as expected, but I think the
former might be the more sensible approach.