Re: 7.3 no longer using indexes for LIKE queries

On Wed, Dec 04, 2002 at 16:02:28 -0500,
  Matthew Gabeler-Lee <mgabelerlee@zycos.com> wrote:
> Can anyone explain the rationale behind that sort order?  I'm guessing it
> has something to do with getting sequences of words sorted 'right', but I
> fail to see how that is right.  I'd think when sorting it would make sense
> to have all sequences of words that start with the same word(s) together.
>
> $ echo -e 'bobbill\nbobrob\nbob bill\nbob robber' | LC_ALL=en_US sort
> bobbill
> bob bill    -\
> bobrob       |-- Seems these ought to be adjacent?
> bob robber  -/
>
> Perhaps I'm just old fashioned.  In any case, I don't define locale
> collation orders, so *shrug*.

I prefer en_US for sorting names. I like that last names like "des Jardins",
"desJardins" and "Des Jardins" will all show up next to each other, because
that is what I think humans will expect. (Note that I sort by first, middle
and last names separately.)