|
From: Mike N. <ta...@al...> - 2002-12-03 10:46:02
|
Dmitry Yemanov wrote: > Since FIRST/SKIP has been originally implemented, we still have some weird > things with this feature. I think it's obvious that FIRST/SKIP doesn't make > much sense (although can be used) without an ORDER BY clause. The only input I can give on this is: I believe a FIRST/SKIP on an unordered set would be completely useless. If I have an unordered set S, and subtract a set "first" M from S, and S-M could return completely different responses from time to time given the same set S, would it be useful for anything but stochastic theory research? In plain language; unless the set S is ordered, SKIPping first 'n' entries just makes no sense in something that's supposed to provide a predictable result. /Mike |