Ole,

<< I would never beat a horse - dead or alive >>

There, see, you do have some redeeming qualities after all.

--
Tim Young
Elevate Software
www.elevatesoft.com

Ole,

<< I don't quite understand how changing it to refresh the views should take
any longer time than unprepare/prepare the cursor statement, which does the
same thing. >>

I thought the same thing, but I believe it is different because of the
nested nature of the view references.  I really have to tear this one apart
piece by piece and go through it in more detail before I can see anything
more.  I'm really not sure what is going on yet because I haven't figured
out yet what the Sudoko code is actually trying to do.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim,

<< I'm really not sure what is going on yet because I haven't figured out
yet what the Sudoko code is actually trying to do. >>

Oh, that one is easy - the Sudoko code is trying to solve a given Sudoko
puzzle

Ole Willy Tuv

<< There, see, you do have some redeeming qualities after all. >>

Being a non-native speaker, I've no idea what "redeeming" means, but I take
it as something positive

Ole Willy Tuv

Tim,

<< I thought the same thing, but I believe it is different because of the
nested nature of the view references. >>

I wouldn't think so. The views are equally nested whether they're
materialized from a close/open cursor or an unprepare/prepare cursor
statement.

Did you eliminate the possibility of an endless loop, like I suggested ?

Ole Willy Tuv

Tim,

<< Did you eliminate the possibility of an endless loop, like I suggested ?
>>

The reason why I'm asking this question is that the sp_Sudoko_EDB_debug4.sql
version, which does an uprepare/prepare is pretty fast with the 5 iterations
(and so is the limited loop version I posted with plain close/open cursor)
and all that is needed to get the correct result is 10 iterations.

Ole Willy Tuv

Ole,

<< I wouldn't think so. The views are equally nested whether they're
materialized from a close/open cursor or an unprepare/prepare cursor
statement. >>

Yes, but internally they aren't necessarily executed in the same fashion.
You have to think of things in terms of reference-counting and other
attributes that may cause a view to not be executed, but only served up
"as-is".

--
Tim Young
Elevate Software
www.elevatesoft.com