im,

>>It would severly affect query performance. I wouldn't want that.
>
>I wonder why... comparing two fields (memo or not) has to be faster than first PADding those same fields before comparing them, no?

No, PADR() converts memo into character filds which cannot be longer than 254 bytes.

>>Why you want to keep VFP SQL Engine ANSI incompatible is beyond my uderstanding.
>
>I'm happy to have ANSI compatibility. I'm just not happy with:
>1) the allegation that it was NECESSARY to implement ANSI to FIX this problem;

I said that it was right way to fix the problemand and you didn't ofer any viable alternative.

>2) the choice to require a change (to SET ENGINEBEHAVIOR TO 70) when an old-styled SELECT that was otherwise satisfactory is encountered. As I've said a few times, a warning message in SYS(3054) output would have suited me just fine.

Here we go again. It was shown to you by different people that "old-styled SELECT" wasn't producing correct result but you don't want to see that.

As I see it, there's no reason to continue this discussion.