>the issue is not "to understand what ago VFP after that it has made it",
>but to preview what it will make.
>
>To remember all the rules that generate a CHAR or VARCHAR is useless.
I still don't see what the big deal is.
>I say that the documentation is corrected, and that the implementation is mistaken
> (I know also the reason, but he is along describing).
I dissagree with you and consider current implementation correct.
>
>You say to use CAST().
>Possible, but it has been implemented badly and it is much slow one.
>
>The much most effective one if logic overthrew ( as current documentation ):
>
>with OFF = > always CHAR
>with ON = > always VARCHAR
>and to leave CAST() in hand to the developer for the little cases to control.
>
See above
>However,
>I see that the Team very rarely places the problem
>to maximize the determinism of the result.
>
>
>
>>* EMPTY STRING IS MAPPED TO VARCHAR(1)
>>SELECT;
>> SUBSTR('ABC',3,1) ;
>>, SUBSTR('ABC',4,1) ;
>
>*** You cannot represent empty string in char field
>
>***** This is a contradiction because with OFF this return two CHAR(1) fields
>
>
>Of course,
>I did not wait for that this comes changed to me,
>but only send these messages in order to stimulate the thought of who
>I will have to make of the choices for the next one release.
Again, see above.
--sb--