Revolution much slower then Hypercard

[Geoff Canyon]

At 11:59 PM +0100 2/10/02, Pierre Delain wrote:
>I find out with great surprise and disappointment that Revolution is much
>slower then HyperCard.
>I have a stack with 2163 cards, and a script containing the following loop :

I love a challenge like this :-)

I wrote a script similar to the one given, to go through the Revolution Language reference (1263 cards) looking for "card" in the three fields: comments, summary, examples. This was on a PowerBook G3 400.

Going from one card to the next with the screen unlocked took 50 seconds.

The visual cost is significant:
Going from one card to the next with the screen locked took 14 seconds.

Delivering messages takes time:
Going through the cards with the screen and messages locked took 6 seconds.

But there's no need to actually go to the cards:
Using the form: fld "comments" of cd i of stack "revDocsLanguageReference" took 90 ticks (1.5 seconds)

Setting the defaultStack property and simply using fld "comments" of cd i took 53 ticks (<1 second)

Finally, pre-loading the data into an array variable instead was faster still (not counting the time to load the array): 9 ticks. 

Using one of the built-in functions (Filter is my favorite) on a pre-built data set would probably be even faster.

regards,

Geoff