Ok my brain is having one of its many moments again...

Once there was a thread posted on here regarding the speed of storing data in global fields verses creating a cartesian join to a preferences tables from each table. It mentioned the way it came over the network and that it may be faster to have the preferences table because it would be stored on the computer rather than having to come over the network each time with globals.

I cant for the life of me remember where it was...

There is something even faster:

http://www.filemakermagazine.com/videos/using-global-graphic-variables.html

...only you need to be subscriper, to pick his brains on this one!

--sd

Hi Soren,

Yes I did read that article a while ago. I am a subscriber and I try to read all of Matthew's articles. Thank you for pointing to it anyway. However, I was specifically looking for that thread. I thought that Mike Hackett had orig posted it but alas I could not find it.

This is the only thing I could find of mine:

http://fmforums.com/forum/showtopic.php?tid/181666/post/227154/hl/faster+loop/

Which is about using variables vs. global fields. Not exactly what you were looking for.

I should think editing values in regular fields would be slower than editing values in global fields, since regular fields are stored on the host and must go through record locking phases when the data is changed.

There are even more to the topic, remember when Debi Fuchs examined the swiftness in illegal relations (global to global) under fm5??? We can today remove any indexing with cathesian linking to dummy fields and the remove them on both sides later, to prevent any indexing being thrown around unnessersary.

http://www.kevinfrank.com/download/df_invalid_relationships.zip

--sd