Re: *LANGIDSHR and SQE

[vhamberg@xxxxxxxxxxx]

I suggest that you do the shared weight indexes - CQE is not a bad thing - SQE 
was intended to speed up complex queries and it won't matter which engine is 
used for many kinds of queries - your mileage WILL vary.

The presence of a supporting index ALWAYS helps, whether SQE or CQE. Without 
the index, a table scan will take place, AND the translation will be done. You 
can, of course, always use the UPPER function - but the results are the same - 
use of a function slows things down. Better if that work is done in the index.

And read the docs on that QAQQINI setting to see what it relates to.

Eventually I believe IBM will bring everything under the new engine - it all 
takes time and resources. This is not an official statement, perhaps a wish, 
but it makes sense to this civilian.

HTH
Vern

-------------- Original message -------------- 
From: "eftimios pantzopoulos" <eftimios@xxxxxxxxxxx> 

> I came across an excellent tip in a Centerfield Technology Newsletter(Vol 2 
> Issue 2) concerning the use of *LANGIDSHR. I actually got it working and was 
> able to retrieve records via SQL with the same column value regardless of 
> case. This involved creating an index while the STRSQL task was set to 
> *LANGIDSHR. 
>
> I was researching this in order to try and overcome the problem of v5r4 
> sending the SQL to the CQE and I optimistically thought that this might 
> overcome that problem and thus despatch it to the SQE. Alas and alack! When 
> I looked at the Visual Explain, the 'Reason SQE was not used' was - 
> 'Translation Required'. 
>
> I presume this is something to do with the upper/lower case process in using 
> the index? 
>
> Is it worth pursuing this path? Or is there no way of overcoming the UPPER 
> barrier to the SQE? Something like the IGNORE_DERIVED_INDEX setting in 
> QAQQINI? 
>
> This is a continuation of a post I made a few weeks ago so I'm not hopeful 
> as *LANGIDSHR came up in that topic as well. 
>
> I tried telling my superiors that we could change every program which 
> populates the table to force an upper, however this is not really feasible 
> as it involves names which are used in communications. 
>
> Looks like we're stuck with the CQE which means there's no point in creating 
> indexes to assist the SQL as they'll be ignored. Does this mean that CQE 
> will always do a full table scan? Or should we retain the logical file in 
> the SQL as it's being despatched to the CQE anyway? 
>
> My head is spinning! 
>
> Thanks......... 
>
> _________________________________________________________________ 
> Advertisement: Silverchair: Young Modern ? listen to full tracks plus win 
> gig tickets! 
> http://ninemsn.com.au/share/redir/adTrack.asp?mode=click&clientID=772&referral=c
>  
> ompetitionpage&URL=http://music.ninemsn.com.au/silverchair