On Dec 6, 2010, at 11:29 AM, Peter Eisentraut wrote:
> This has been touch upon several times during the discussions on past
> patches.
>
> Essentially, the current patch only arranges that you can specify a sort
> order for data. The system always breaks ties using a binary
> comparison. This could conceivably be changed, but it's a separate
> problem. Some of the necessary investigation work has presumably
> already been done in the context of citext.
Okay, thanks, good to know.
Best,
David