What is actually transfered on opening the data.fp7 backend?

[Stuart Taylor]

Hi Everyone,

I'm at the stage of testing a separation model solution on the internet.

Now it seems to work just great.

• Takes say a minute to startup

• When open works with very little lag.

Now i have one concern. For test purposes i embedded some movie files in containers, when navigating to those records it seemed like they had already been loaded.

I am hoping that this did not happen when the data file opened as if someone has 1000 movie files you may as well open the solution Monday and come to work on Wednesday.

When did the solution get that data???

Any ideas on when FileMaker gets stuff as i would like to optomize the file for both performance and startup.

best

Stuart

[Fitch]

FileMaker definitely does not load all records at startup. In my experience the biggest slowdown of startup over a WAN is a complex graph, because FileMaker has to evaluate every possible relationship at startup.

[Stuart Taylor]

Thanx Fitch,

Great this is a complex solution. Good to know. Thanks :)

[bruceR]

FileMaker definitely does not load all records at startup. In my experience the biggest slowdown of startup over a WAN is a complex graph, because FileMaker has to evaluate every possible relationship at startup.

It is my understanding that Devcon under the hood sessions definitely declared this incorrect.

[Fitch]

Yes, what was stated at the under-the-hood session was that FileMaker has to evaluate every possible dependency at startup. Which is what I meant, i.e. FileMaker has to see how everything connects. It's not the case that it actually scans all related records somehow. Sorry for any confusion.

[Stuart Taylor]

Hey Fitch,

I have found your original statement to be true so should you statement not read,

"No, what was stated…" (as the response from Bruce is incorrect)

rather than

"Yes, what was stated…"

Just to avoid any further confusion.

Or maybe i am confused.

Edited July 22, 2008 by Guest

[Fitch]

The first part is true: FileMaker does not load all records at startup.

The second part is I believe what Bruce (correctly) objected to: FileMaker doesn't evaluate relationships at startup, it evaluates dependencies.

[Stuart Taylor]

Thanks for the clarification Fitch.

Could you clearly define the difference for me, i think i have a grasp but as you seem to be very clear on these things i would greatly appreciate your definition so that i can make sure my files are optimized for internet based access via a separation model based solution.

All the best

Stuart

[Fitch]

Let's say you have two Table Occurrence Groups (TOGs):

A-B and C-D

A-B is one dependency, C-D is another.

Now let's relate B-C, so our graph looks like:

A-B-C-D

We count these dependencies:

A-B

A-C

A-D

B-C

B-D

C-D

You can see where a large "spider" graph can potentially be costly.

I'm not clear on whether A-B / B-A would count as one or two dependencies. I would love to see some of these under-the-hood Devcon sessions turned into white papers.

The terms "relationship" and "dependency" are loosely used in the same way. In this specific sense though, relationship refers to related data and dependency to the schema.

Hope that helps, and I hope it's correct! I'd love some additional clarification too!

[Stuart Taylor]

Thanks Fitch,

So if i have a Central Table from which all tables relate (a related menu), everything is evaluated on start up even if it does not appear on the layouts.

GULP!

[CurlyD777]

I have a question also in the "dependency" talk...whenever I make an update to our increasingly complex database, i get a pop up screen saying "rebuilding dependencies..." It takes only 3-5 seconds to complete, but as we plan on taking our database offsite, I'm concerned too many dependencies are built to make it unusable remotely.

In a database design report, I have 741 relationships (which i know are not the same as dependencies, but I don't know how to quantify how many I have) As I continue to build the database, I'm starting to creating future modules as related files. I still am making the relationships in the main database, which is creating more dependencies.

Anyone have input if it's best to have the relationships in a single file, or does it make a difference in making these relationships in the related files, and displaying layouts from those files when needed.

During startup, does it still check over all related files for dependencies as well? or will it pause when first accessing these related files to determine the dependencies?

Any feedback or thoughts much appreciated.