Hi,

Obviously, variables should be faster than globals, but in fact that's wrong. Globals are much faster than variables !

I've made a counting loop with FM 8's script variables and the same script with old school golbals fields.

Guess what ? Globals are much faster, either on a standolne solution or through networking with FM 8.0v4 networking.

counting to 500 000 with globals }:| 55s, with $avariable : 79s !

That's for local.

For WAN performanace with 8.04 server that's 11s vs 16s for counting to 100 000, and 56 vs 86 for 500 000.

(at least local and WAN performance is almost the same, which is perferctly normal since everything is processed in client's memory)

I know that I shouldn't say that, but are FM engineers morons that love to make us loose our time ? How on earth can you put a new feature and making it slower than what was already there ?

Didn't they ever hear the word regression ?

They're really laughing at us !

That nullify a large part of the script variable applications !

And by the way, I've also tested in a real world much more complex script and that's still true.

Test performed on dual 2.5Ghz G5. I've attached the file to see for yourself.

I've attached the test file and an applescript that can time any fm script (you have to change the name of the script accordingky in the source code of the applescript)

GlobalvsVariableTest_timer.zip

Edited October 28, 200619 yr by Guest

Globals are much faster, either on a standolne solution or through networking with FM 8.0v4 networking.

I'm afraid your test is flawed. Your variable test is still setting global values. If the Set Field [] of the global is removed from the variable test loop, the variables do perform better (by about a third for me).

I know that I shouldn't say that, but are FM engineers morons that love to make us loose our time ? How on earth can you put a new feature and making it slower than what was already there ?

Didn't they ever hear the word regression ?

They're really laughing at us !

Your whining is getting tiresome and is quite misplaced. If you have a concern, post it in a respectful way, and maybe someone can confirm or resolve it. But it's not useful to anyone to hear you complain about all the problems you have with FileMaker, especially when your assertions are wrong.

Test_variable_vs_global.fp7.zip

I get approximately the same result - variables are faster by ~40%.

thanks guys was wondering this myself ... comment between this and the x-relationship discussed in another post "Cartesian or Constant" the performance of my database just shot through the roof and all in one day... yippee!!!

goodbye globals and constants ... wow what a day!

FileMaker are on the right track for us guys ... can't wait to see whats next.

Well I'm eating my hat right now,

Thanks Ender, lesson well learnt, no more whinning promised !