>>
>>select *, 'Missing' as cInfo from FirstResult where ID not in (select ID from mySecondResult) ;
>>union all ;
>>select *, 'Extra ' as cInfo from mySecondResult where ID not in (select ID from FirstResult)
>
>
>Excellent idea. :)
>
>I am starting to see that the query is not my problem. This little exercise is part of a Fox-enabled web site using Active Foxpro Pages. The query itself appears to only take a fraction of a second to run in fox, but the code that produces results rows for the html table is slow, now that I test them both.
>
>Assuming temp to be the result cursor from the previous query, does this look like an inefficient use of TEXTMERGE to produce my html result set rows? This produces one HTML table row per query result, the resulting text is then written out to the client with response.write()
>
>
>
>

>Function CreateTableRows
>	Local cText,i
>	i = 0
>	If !Used("temp")
>		Return []
>	Endif
>	cText = []
>	Select temp
>	Scan
>		i = i + 1
>		TEXT TO cText ADDITIVE TEXTMERGE NOSHOW
>  <tr bgcolor="#dcdcdc" class="QueryTable">
>    <td width="103" bgcolor="#dcdcdc" class="FieldHeadings QueryResults">
>      <input name="WorkCode<<i>>" readonly type="text" class="TableHeadings" id="WorkCode<<i>>" value="<<temp.wcode>>" size="15" onclick="MM_goToURL('parent','<<>>');return document.MM_returnValue" />
>    </a> </td>
>    <td width="360"><span class="QueryResults">
>      <input name="Borrowername<<i>>" readonly type="text" id="Borrower_Fname<<i>>" value="<<temp.borrowername>>" size="50"/>
>    </span></td>
>    <td bgcolor="#dcdcdc"><span class="QueryResults">
>      <input name="Loan<<i>>" readonly type="text" class="TableHeadings" id="Loan<<i>>"  onclick="MM_goToURL('parent','<<>>');return document.MM_returnValue" value="<<temp.display_loan>>" />
>    </span></td>
>    <td bgcolor="#dcdcdc"><span class="QueryResults">
>      <input name="Parcel<<i>>" readonly type="text" id="Parcel<<i>>" value="<<temp.acct>>" />
>    </span></td>
>  </tr>
>		ENDTEXT
>	Endscan
>	Return cText
>Endfunc
>
>

I would probably do it slightly different way, e.g. put

set textmerge on
set textmerge to memvar lcHTML noshow
scan
\ ....
endscan

In other words, you only set textmerge once before the scan. I'm not sure, how much would it speed up the process.