>>>>I have a table that contains a list of projects. Each project may be a parent, child, or grandchild project. One of the fields 'parentproject' identifies the relationships. If the field is populated with the ID of another project, then the referenced project is the current project's parent. I have a report with 2 groups to display these like this:
>>>>
>>>>Group1................Parent
>>>>Group2.....................Child
>>>>Details..........................Grandchild
>>>>
>>>>This report seems to work OK and is built on the following SELECT
>>>>

>>>>select r1.system, r1.rtp_id, trim(r1.st_name) + " / " + trim(r1.termini) + " / " + trim(r1.project) as st_name_a, r1.lead, r1.del1, ;
>>>>	r2.rtp_id, trim(r2.st_name) + " / " + trim(r2.termini)  + " / " + trim(r2.project) as st_name_b, r2.lead, r2.del1 as del2, ;
>>>>	r3.rtp_id, trim(r3.st_name) + " / " + trim(r3.termini)  + " / " + trim(r3.project) as st_name_c, r3.lead, r3.del1 as del3 ;
>>>>	from rtp_projects r1 ;
>>>>	left join rtp_projects r2 on r1.rtp_id = r2.parentproject ;
>>>>	left join rtp_projects r3 on r2.rtp_id = r3.parentproject ;
>>>>	where (r1.parentproject = 0 or isnull(r1.parentproject)) and ! isnull(r1.st_name) and ! isblank(r1.st_name);
>>>>	order by r1.system, r1.st_name, r2.st_name, r3.st_name	

>>>>
>>>>Now comes my problem, there is a second table that also contains a list of projects that may be a child to any of the projects in the first table. I would like to include these projects in the report as well, either as a child, grandchild, or great-grandchild. I'm not clear on how the select should be modified to do this, and the report also. Do I need 2 or more detail bands or additional groups, a select with a UNION or more JOINS, etc?
>>>
>>>>
>>>>Any pointers will be appreciated. Thanks.
>>>
>>>
>>>Because you speak about more than one Detail band I assume you use VFP9.
>>>
>>>You can use UNION of course, but you could also use so called derived table.
>>>But I would do this in two steps: First UNION both tables in ONE cursor and then using that cursor I'll do the main SELECT (that one you show us)
>>
>>Borislav -
>>Thanks for your suggestion. It is simpler and more elegant than the solution I had cobbled together. However upon trying it, it fails because the cursor formed by the UNION has no data for the keyid (rtp_id) for the records coming from the TIP table. See below
>>

>>select system, tip.tip_id, 000.00 as rtp_id, rtp_id as parentproject, trim(st_name) + " / " + trim(termini) + " / " + trim(project) as st_name, lead, .f. as del1 ;
>>from TIP ;
>>inner join tipdatabase!proposed on tip.tip_id = proposed.tip_id ;
>>where proposed.status = "Active" ;
>>union ;
>>select system, 000.00 as tip_id, rtp_id, parentproject, trim(st_name) + " / " + trim(termini) + " / " + trim(project) as st_name, lead, del1 ;
>>from rtp_projects ;
>>into cursor crsTemp
>>
>>select r1.system, r1.rtp_id, r1.st_name as st_name, r1.lead, r1.del1, ;
>>	r2.tip_id, r2.rtp_id, r2.st_name as st_name, r2.lead, r2.del1 as del2, ;
>>	r3.tip_id, r3.rtp_id, r3.st_name as st_name, r3.lead, r3.del1 as del3 ;
>>	from crsTemp r1 ;
>>	left join crsTemp r2 on r1.rtp_id = r2.parentproject ;
>>	left join crsTemp r3 on r2.rtp_id = r3.parentproject ;
>>	where (r1.parentproject = 0 or isnull(r1.parentproject)) and ! isnull(r1.st_name) ;
>>	and ! isblank(r1.st_name);
>>	order by r1.system, r1.st_name, r2.st_name, r3.st_name ;
>>	into cursor crsRTP

>>My solution of setting this value to 0 in the first SELECT creates duplicate keyids and spoils the second SELECT. I am thinking on a solution to this but so far I'm blank.
>>:)
>>
>Just do not use UNION but UNION ALL. That way all dupicates will be presented in the new cursor.

I think I wasn't clear. The problem is that there ARE duplicates in the result. All the records coming from the TIP table have 0.00 as their rtp_id which is supposed to be unique.