I have two versions of a file that are almost exactly the same except for a couple of layouts.

I imported newer data from an older version to a newer version. I checked and re-checked, and all of the precise number of records are in place exactly as they should be.

But somehow the new version of the file is almost 10 megabytes LESS than the slightly older version with the exact data contents. Either file, saved as an empty clone, differ only about 300kb.

How is it possible that I have lost 10 megs when all of the data appears to be in place??

Try doing some finds on your different fields and watch it grow in size. :B

hmmm..

So does filemaker store all find requests permanently?

If so, is there a way to manually clear that out, or is it some kind of memory leak?

No not a memory leak, but a gradualy turning on indexing. If you should wish for a smaller index should you make all your searches in a carthesian product calc'field made by say:

http://www.briandunning.com/cf/277

--sd

You can also prevent INDEX CREEP by the Storage Options. Minimal eliminates word indexes. FM Help explains this pretty well. Search for 'Defining Field Indexing'. Here is some of it which applies to your situation ...

To create relationships using text fields as match fields without creating word indexes for these fields, use Minimal and disable the option to Automatically create indexes as needed.

To reduce file size and prevent users from creating indexes, use None (or Minimal) and disable the option to Automatically create indexes as needed.

These various settings will make a big difference! Once indexed, a field STAYS indexed and your file will head in an upward spiral. I set most field indexing options myself and rarely use the default Automatic Indexing feature. It won't mean you can't perform a find on a field - only that it will be slower (depending upon your index selection). But many fields are searched so rarely that it is a good trade-off.

LaRetta

Not that Danielles CF is wrong at all, for the purpose she thought it into, but it's not exactly making a carthesian product. For the carthesian purpose is it unfortunately adding a lot of dead meat to the index, while this pair are closer to the idea:

http://www.briandunning.com/cf/394

http://www.briandunning.com/cf/393

--sd

Soren,

Danielles CF is wrong at all, for the purpose she thought it into,

[color:blue] Danielles is[color:red] Daniele

and

[color:blue] she is a [color:red] he

And I know you know that.

Lee

Thank you Lee.

Neither I understand what is the union thread from my CF FiveFieldsPermutations (field1;field2;field3;field4;field5;number) and this TOPIC.

And I know you know that.

No i didn't and I deeply apologize!

To the thread ...Moyer brought it up at devcon this summer, that using a cathesian product calcfield could make a file smaller if the carthesian field is the only field indexed as well as all seaches should be made in it - this was an opimization meant for multiuser scenarios.

Why it works is way beyond my math skills. I brought it up because he havn't enlightend us further with uploads of his slides from devcon ...while hoping that someone versed deeper in math can explain why it apparently works??

--sd