Very slow, very short value lists

[WilliamCB]

Lots of people have reported slow value lists, and often the response here has been to say, "Your value lists are too long". Well, now I have a very slow value list that is very short, so I'm hoping someone has an idea I might try to solve this problem!

We index some records with a fully flexible thesaurus of about 4000 terms. So any term can be a child of any other term. In practice, there are a number of (mainly) distinct branches of the tree and to simplify selection for users, these are identified by a number field in the term record called "branch_id".

So, all I want is a value list of the all the terms in a given branch. To do this, I have created a global field "gValueList" with the value of the relevant branch_id and formed a relationship gValueList - branch_id.

I can then define a value list that has term_id as the first field, term_name as the second and is limited by the relationship. This selects the correct list of terms for the value list. But...

...it's INCREDIBLY slow. For a list with five terms, it takes about 10 seconds.

Any suggestions?

[IdealData]

Are the items in the value list the results of calculations?

Is the relationship sorted, particularly by multiple fields?

[WilliamCB]

No, there's no calculations involved. It's a simple serial number id and text name.

And the relationship is unsorted.

But when you define the value list, it has to be sorted by something - either the first or second field.

By contrast, if I define a portal in the same way the value list is defined, the portal is filled instantly. This seems to be something to do with the value list "engine" running amazingly slowly.

Edited July 10, 2009 by Guest