Subject RE: [ib-support] rowcount
Author Helen Borrie
At 03:33 PM 26/02/2003 -0500, you wrote:
> > It's actually
> > select first 10 skip 10 * from mytable
> > , or
> > select first 10 skip 10 nameofcol1, nameofcol2 from mytable
> > --
> > Sean Leyne
>Yeah, I know - but the first(x) skip(x) syntax works (1.5 beta 2), so I
>didn't question it, just addressed the lack of selected columns...
>Do you think acceptance of the parantheses is accidental?

For simple integer arguments the parentheses are optional. The m and n in
SELECT FIRST m SKIP n can be expressions. Brackets would usually be
obligatory for expressions. I documented this command in the releasenotes
*with* parentheses, because this was how it was given by the developer who
implemented it in the DSQL engine. At that time, it wasn't even documented
that optionally omitting the brackets was possible for simple arguments,
but it's a 'plus' that it is so.

There are plenty of other places in SQL where brackets are optional. I
don't know of any arguments or moves to either enforce or drop them, so
there is no reason to suppose that SELECT FIRST, which actually NEEDS them
under some conditions, would be so targeted.

>If so, he probably should not use them, since they might not
>work in the future.

Believe it or not, there are actually checks and balances to avoid
arbitrary changes happening that break existing databases/applications
(overt bugs apart, of course!)

>Minus a true up-to-date FB manual, one of the developers
>would have to address that, I guess.

Not sure what you think needs addressing.

As for a "minus a true up-to-date FB manual", I'm sure you meant to say
"minus a true up-to-date FREEBIE FB manual...", didn't you? Or perhaps you
really don't know about the Firebird manuals?

Nevertheless, the release notes are free... The real problem is getting
people to RTFM.