Vlad Khorsun escreveu:

> Alex Peshkov :
>
>> On Tuesday 24 June 2008 21:11, sergey.mereutsa wrote:
>>
>>> Agree, if entrie server is written in C/C++, why you must use non-
>>> native for C indexes? Personally, I prefer zero for first array
>>> element.
>>>
>> I also prefer 0. But this is _SQL_ server, and we should better follow SQL
>> traditions here. In SQL all enumeration is done based 1.
>>
>
> This is not SQL-language interface. This is C++ interface. Any plugin\external
> engine will implement its own user-friendly wrapper on the top of this interface
>

The (generic) interface is expected to not need wrappers, at least for C++.

> to use with target programming environment. And in this wrapper they may use any
> number for low index bound.
>
> I don't know on which SQL traditions you reffers above. Am i SQL newby ? :) Why
> should we ugly our own interface because of some suspicious traditions ???

I remember to see one time in the standard something related to indexes
(from API POV) starting at 1. But can't found now. But for ORDER BY and
GROUP BY, they start at 1.

> I guess our own traditions must be closer for us (remember XSQLDA ? ;)

Completely different thing. Of course we should never give "normal
arrays" with dummy first element. :-)


Adriano