Answer inline:

> -----Original Message-----
> From: David Jencks [mailto:davidjencks@...]
> Sent: Wednesday, June 06, 2001 4:52 AM
>
> Hi,
>
> We've had a long discussion here, featuring:
>
> -limit m. n
>
> -bookmarks
>
> -rownums
>
> -Ann's arbitrarily complicated query to fetch the next records using a
> bookmark
>
> and we found that people want to do this with arbitrarily complex joins.
>
> My question is, is there any reason to think the rownum or complicated
> query solutions won't force the server to fetch all the rows
> satisfying the
> query into memory, then skip the first n, or rows not satisfying the
> "nested after" condition, and start returning more?
>
> In other words, don't these all boil down to asking the server to do all
> the extra work it would have to do with limit m, n anyway? The syntax for
> limit is certainly simpler and easier to use, doesn't return a fake column
> ( although that might be ok anyway), and might be easier for the engine to
> understand.
> ...

Yes! Completely agree!!!
That would be my favorite solution.

> I still think the best answer to this question is to figure out how to

make

> the user ask a question that will only give her 7 rows, perfectly

answering

> their needs, but this is not really a database question.

I wish I could ensure that trough an interface that the user would still
want
to use.
=:o)


Thanks,
Paulo Gaspar