On 10/01/14 12:55, Peter Geoghegan wrote:
> On Thu, Jan 9, 2014 at 3:41 PM, David Fetter <david@fetter.org> wrote:
>> We have dropped support, as you put it, for bigger and harder-hitting
>> mistakes than this. Anybody whose code has this kind of silliness in
>> it will be in other kinds of trouble, too.
> While the decision to make it possible to set the lower bound index
> value arbitrarily was made before I was active in the project, I
> imagine it went something like this:
>
> Person 1: We should make our arrays similar to those found in a
> certain proprietary system's standard procedural language - with one
> as a lower bound.
>
> Person 2: I don't like that, it should always be zero.
>
> Person 1: We can all be winners.
>
Starting arrays at zero makes the most sense, as then you can calculate
the displacement simply as (index) * (size of entry), and not have
subtract one from the index first. This would be my preference.
Both C & C++ use zero as the default, so Java naturally followed their
example.
But I wouldn't push it, as people who mainly know SQL are more used to
starting arrays at 1, and it is not worth the effort to change it as
this stage. At least we should be consistent how we treat arrays in SQL.
Cheers,
Gavin