Roy,

<< With the clue that your code was okI found MY fault. Do you think if I
took this week back I could get a refund? >>

If you do get one, I want one also.  I'm just now starting to feel
up-to-speed.  I haven't been this "under the weather" in years, and it
sucks.

--
Tim Young
Elevate Software
www.elevatesoft.com

Roy,

<< Suggestions, comments, help etc >>

My guess is that the index is probably messed up with all of the playing
around that you've done.  Did you try dropping the full-text index and then
re-adding it with the latest word generator, filter, etc. ?

<< Final question I added a showmessage back into the wordgenerator for the
_readby column - why is is called 2 or 3 times when running the query? >>

Because it has to in order to evaluate the query properly.  Remember, cost
estimates need to be made, etc. and these things all require that the words
be parsed out so that they can be used.  These areas of the engine that need
this information are also in vastly different areas of the engine, so this
information cannot be cached.

--
Tim Young
Elevate Software
www.elevatesoft.com

Roy,

<< OK its the NULL handling - I think >>

Your listing of the row counts really don't help me much without being to
actually see the data to see if what is being returned is correct.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim

>My guess is that the index is probably messed up with all of the playing
>around that you've done. Did you try dropping the full-text index and then
>re-adding it with the latest word generator, filter, etc. ?

This column was never touched but it was on my mind to try that this morning anyway.

><< Final question I added a showmessage back into the wordgenerator for the
>_readby column - why is is called 2 or 3 times when running the query? >>
>
>Because it has to in order to evaluate the query properly. Remember, cost
>estimates need to be made, etc. and these things all require that the words
>be parsed out so that they can be used. These areas of the engine that need
>this information are also in vastly different areas of the engine, so this
>information cannot be cached.

Thanks. I wondered if it was like the DBGrid thing where events get called several times. I'm not to concerned since (I hope) it only relates to parsing the search terms and if anyone types in more than (say) a dozen words they deserve all the slowdown they get

Roy Lambert

Tim

>Your listing of the row counts really don't help me much without being to
>actually see the data to see if what is being returned is correct.

I agree they don't help to identify the error but in my eyes they prove something is wrong, even if its with me

How can this pair of queries NOT return the full set?

select * from emails where _message does not contain 'default' or _message is null
select * from emails where _message contains 'default'

I just tried

select * from emails where (_message does not contain 'default') or (_message is null) or (_message contains 'default')

and that does give the right number of rows but as separate statements its still wrong.

Then there's the 54 items where neither _fkCompanies nor _fkContacts get set in which _Message is neither null nor does the generated word list contain 'default'

I'd also like to know why adding the null test ramps the time taken from subsecond to 150 secs+ (sensitive result set in both cases)

I'll let you know what happens after dropping and recreating the index later on today.

Roy Lambert

Tim

I stripped out a load of fields and got it down to 24Mb so I've emailed it to you. I'm hoping its my problem cos they're usually easier and faster to fix than when they're yours <vbg>.

Roy Lambert

Roy,

<< I stripped out a load of fields and got it down to 24Mb so I've emailed
it to you. I'm hoping its my problem cos they're usually easier and faster
to fix than when they're yours <vbg>. >>

Both the performance issue and the issue with DOES NOT CONTAIN are fixed,
and the fixes will be in the next build 4.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim


><< I stripped out a load of fields and got it down to 24Mb so I've emailed
>it to you. I'm hoping its my problem cos they're usually easier and faster
>to fix than when they're yours <vbg>. >>
>
>Both the performance issue and the issue with DOES NOT CONTAIN are fixed,
>and the fixes will be in the next build 4.

OK so your's was "easy" to fix this time - I away b4 with anticipation.

Roy Lambert

There is definitely a problem with the keyboard on my new note book - "I away b4 with anticipation."

Roy Lambert

Roy,

<< There is definitely a problem with the keyboard on my new note book - "I
away b4 with anticipation." >>

I knew what you meant.

--
Tim Young
Elevate Software
www.elevatesoft.com