Re: [PATCH] diff-highlight: Fix broken multibyte string

[Jeff King <peff@peff.net>]

On Thu, Apr 02, 2015 at 06:59:50PM -0700, Kyle J. McKay wrote:

> It should work as well as the original did for any 1-byte encoding.  That
> is, if it's not valid UTF-8 it should pass through unchanged and any single
> byte encoding should just work.  But, as you point out, multibyte encodings
> other than UTF-8 won't work, but they should behave the same as they did
> before.

Yeah, sorry, I should have been more clear that I meant multibyte
encodings. UTF-8 is the only common multibyte encoding I run across, but
that's because Latin1 served most of my pre-UTF-8 needs.  I suspect
things are very different for people in Asia. I don't know how
badly they would want support for other encodings. I'm happy to go with
a UTF-8 solution for now, and see if anybody wants to expand it further
later.

> >I timed this one versus the existing diff-highlight. It's about 7%
> >slower.
> 
> I'd expect that, we're doing extra work we weren't doing before.

I was worried would be 200% or something. :)

> >Makes sense. I'm happy enough listing perl 5.8 as a dependency.
> 
> Maybe that should be added.  The rest of Git's perl code seems to have a
> 'use 5.008;' already, so I figured that was a reasonable dependency.  :)

I shouldn't have said "listing". I just meant "have" as a dependency. I
am also happy with adding "use 5.008", but I agree it's probably not
necessary at this point. It was released in 2002 (wow, has it really
been that long?).

-Peff