Hi

I am a FileMaker Pro 6 Developper. I have read posts on the Sesame forum every day since the very

beginning. I think that Sesame could offer an interesting alternative since it has a full language

and event driven actions (on field exit etc) which are sorely lacking in FMP.

Since linking a parent record to other tables records is one of the most important feature of a

DBS, I examined the user manual (PDF) and search for the term "relational". This search produced

over two dozen instances. I look and read everyone of them. The gist of it is that...Natural Link

is simpler and faster but not very flexible while the relational link is more complicated, slower

but more flexible. I am concerned by the "speed" aspect since Sesame does not do indexing at all.

In order to have a benchmark, I produced a test database In FMP using an application I had already

developped. It is essentially an order/line items application with additional relatinal links for

clients details, products description and price etc.

I have 10,000 records in the order file (with a subform for the line items) and 86,000 line items

in another table.

When I look at any record, the subform shows me instaneously its line items (1 item to 25 items on

average).
Finding subset of records is also very fast since every field is indexed on the fly if needed.

I have essentially 2 questions:

A) How fast/slow would Sesame be to show me line items on a subform if I was to build a similar

application with the same number of records using relational links.

B) How fast/slow is Sesame at finding a subset of record with a simple request (find all ACME Co

orders), with a more complicated request (find all ACME Co with orders for January Larger than

$500).

TIA

Serge

Hi Bob

Thanks for the reply.
I'm not sure we are talking about the samething. In FMP, you can browse your entire database using a key combination to go from one record to the next. Say you have 10,000 records, and you have record number 10,000 on your screen and it's subform with the line items (relational link). If you press Ctrl + Down arrow to goto record number 9,999 you instantly see record 9,999 and it's subform is instantly populated with its proper line items.

Regards

Serge

I have a little experience with FMPro v5.  We have it here at my lab, but don't use it extensively.  In our setup, when you load FMP and select a database, FMP has "remembered" the last search criteria of that database and those records are now the subset of records available to BROWSE (scroll) through, not the entire database.  The user must do a FIND to access other records in the database. It is fairly fast at retrieving records, but seems a bit slow at sorting (this is on a fairly slow computer). 

It sounds like from your example that when you press Ctrl-Down to see the previous record that this previous record is a part of that subset of records available to "browse", and is already loaded in memory, and so seeing that record would be fast.  Am I correct on this?

This would be a different scenario than what you are asking about Sesame doing a complicated search on several criteria.  Sesame does not (by default) load a "result set" of records.  The user must do a search.  It would take time to search the database, but once Sesame has retrieved that "result set" of records, moving from one record to another would be instantaneous. (Am I correct that you are wanting to compare the speed of accessing a record in FMP that is already part of the subset of records that has already been loaded into memory, vs. Sesame doing a search on one or more criteria?)

I cannot offer any better comments about FMP, and am probably not doing a good job of discussing it.  I have not done a comparison to Sesame.  From reading about Sesame, the amount of RAM available is an important consideration.  I'm sure one of the Lanticans (probably Mark) will offer more specifics.

Either here on the Forum, or in the last Inside Sesame newsletter, someone mentioned something that might be of value.  They were asking about the amount of time it took to load the database.  The reply was to use one of the network licenses to keep that database loaded, and then access to it would be instantaneous, rather than having to wait for it to load.  That would mean having an extra license for this purpose, but might be a worthwhile business investment to provide rapid information to customers, clients, etc.

Would it be feasible to export your 10K record database and the 86K record lineitem database, and try to create a Sesame application and import the data to do a test?  A demo version of Sesame is available, and I don't think it is "crippled" in any of the features you would need.  (If I'm wrong, somebody speak up!)

Hope I've offered some food for thought!

T.J.

Hi T.J.

I think you are mixing together my 2 questions. My first question is browsing the entire parent database with 10,000 records (no finding of a subset is necessary if you want to look at the entire database) and simply going from one record (form) to the next or the previous (this done with Ctrl + Down Arrow and Ctrl + Up Arrow in FMP), the relational link with the line item file populates the subform (called a Portal in FMP parlance) instantly. Concerning the speed of a find, Bob indicated that it is quite fast with his database of 37,000 records.

Hope I'm clearer now

Thanks for your reply and Regards

Serge

Hi again T.G.

You are absolutely right on your point 1)

On point 2) and from what I am experimenting with the sample databases provided with Sesame, you put yourself in search-update mode and then pressing f10 0r f9 will let you immediately scroll through the actual set of records. There is no need to do any kind of retrieval if the actual set of records is what you need. At least it is my experience (very brief and superficial)

Regards

Serge

Thanks for the specific details Bob

Serge