Martijn van Oosterhout wrote:
> On Thu, Aug 14, 2008 at 05:57:26PM -0500, Decibel! wrote:
>
>> FWIW, there is desire to be able to re-order columns within real
>> tables, too. But before that can happen we need to divorce
>> presentation order from on-page order (which is actually desirable
>> for other reasons), but that's an awfully big task that no one's
>> taken on.
>>
>
> Actually, ISTR that someone posted a patch and it was rejected on the
> basis that it made the backend coding too confusing and would cause
> bugs to creep in (by using the wrong position during coding). I don't
> buy the argument though, since the end goal is to have logical position
> != physical position, so I don't see how bugs could survive very long.
>
>
>
NO, IIRC it was rejected because it didn't implement what we wanted,
namely Tom's three-number scheme (immutable id, plus mutable logical and
physical order).
cheers
andrew