I have been doing some messing around with joins and have found that I can join files various ways and still seem to get the same result. Was wondering if anyone could tell me what they thought of the following options and maybe which is better to use or which to use for what cases....

Option 1. Tree join with all joins left-outer multiple. So file1 to file2, file1 to file3, file1 to file4........with this, if I sort by a field in file1 and file3 or just file3 I get the verb objects not in same sort path.

Option 2. Similar to option 1 but switch the join that would contain your sort bys to a single/unique instead of multiple. This eliminates the sort error. So file1 to file2, file1 to file3, file1 to file4.....sorted by fields in file3 but file1 to file3 would then be left-outer single. From what I can tell, WebFocus sees this as a single join even though I have 1 row in file1 and several rows in file3 since they have the same ID# we can use single and still get all the information/rows from file3.

Option 3. Linear join with left-outer and multiple. This puts everything in 1 straight path and WebFocus can logically get from one file to another. File1 to file2 to file3 to file4......Now it does not matter where your sort fields are located.

Option 4. Use a hold file.

Thanks!
malinda

This message has been edited. Last edited by: Kerry,

I may be mistaken, but I think that this has been talked about before in the forum.

I would also suggest that each join needs to be "crafted" for each situation and database.

Picking out one phrase in your post -

quote:

This eliminates the sort error.

and this would lead me to ask - Do you know why?

If the answer to my question of you is - No - then I would urge you to look at forecasting what your join would produce in a given scenario and then design your test data and prove that forecast.

As Waz states -

quote:

... that each join needs to be "crafted" for each situation and database.

T