Re: sortsupport for text

On Jun20, 2012, at 19:38 , Peter Geoghegan wrote:
> On 20 June 2012 17:41, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> In any case, if you have to redefine the meaning of equality in order
>> to justify a performance patch, I'm prepared to walk away at the start.
>
> The advantage of my proposed implementation is precisely that I won't
> have to redefine the meaning of equality, and that only the text
> datatype will have to care about equivalency, so you can just skip
> over an explanation of equivalency for most audiences.

Ok, so what exactly is it that you're proposing then? AFAICS, you're
proposing to make the less-than operator rely solely on strcol(). If you
don't also redefine the quality operator to match, you'll end up with
cases where !(a < b) and !(b < a), yet also !(a == b). This breaks the
trichotomy law, no? Which in turns breaks merge-joins - I believe we'd
output the cartesian product {potty, potyty} x {potyty, potty} when
merge-joining {potty, potyty} with itself, unless the comparison operator
contains a tie-breaker. Which is obviously wrong unless you decree
that 'potty' = 'potyty'.

I do agree that there's a use-case for having a textual type which
treats equivalent strings as equal (and which should then also treat
equivalent Unicode representations of the same character as equal). But
it should be a separate type, not bolted onto the existing textual types.

best regards,
Florian Pflug