v6 "tunneling" relationships to v8

[kiwipaul]

Our homegrown v6 application has a dozen files, most with several thousand records, and the larger files are 200Mb+. It extensively uses the "tunneling" method to pass data between the current records of multiple open files by running external scripts, setting globals and then using look-ups. I believe this was the standard (only?) method to pass data between (artificially) related files in v6.

I realise that the v7/8 relationship model totally supersedes this old kludgy way of getting data between files/tables, and that re-architecting the whole application is the ideal. However that is a monumental piece of work for us.

Will the v6 "tunneling" method still work in v8? If so, is the performance OK and is it a reasonable solution to leave it in place in a v8 application in the medium term?

Our problem is that we developed our application on the fly over many years, it probably has at least a full person-year of development time, and we do not have the free time at the moment to sit down and do a full rewrite!

[Ender]

Yes, Lookups still work in FM7/8, but in large files they seem to be slower than they used to be. You may find auto-enter calcs perform better for pulling data from a parent record into a child record.

[Søren Dyhr]

Tunneling as such is a bygone - correct! Watch this movie:

http://previews.filemakermagazine.com/videos/541/GraphRules_full.mov

And yes this presents a serious problem:

Our problem is that we developed our application on the fly over many years, it probably has at least a full person-year of development time, and we do not have the free time at the moment to sit down and do a full rewrite!

...and is exactly the issue debated here:

http://podcast.Adatasol.com/media/Filemaker_PodCast_03042006.mp3

--sd

[John Mark Osborne]

I recently discussed the speed issue concerning lookups with several of the Systems Engineers. The slow speed of lookups in larger files is a known problem. There are actually two workarounds. One is to use an auto-enter calculation referencing the related field (mentioned previously). The other is to use the Lookup or LookupNext function in an auto-enter calculation in case you need lookup functionality. Believe it or not, the Lookup functions don't suffer the same speed issues as the standard lookup feature, according to the Systems Engineers.

However, some pipelining is still needed in FileMaker 7 and 8. For instance, let's say you have the standard invoicing solution with products and a join table between the two. You can display the product name in a portal in the invoicing table layout from two tables away but prices should be used as an auto-enter or lookup in the join table (pipelining). Let's say you decide to change the price of a product. Do you want the price change to be reflected in all your old invoices? The answer is obviously NO so you do need to use a lookup or an auto-enter to take a snapshot of the price at that moment in time.

[kiwipaul]

Well we bit the bullet, set up a standalone box (Win 2000 Professional)and did a multiple file drag and drop conversion using the free NFR copy of v7 we got with our last v6 licence. To my amazement it runs 99% perfectly, including several older plug-ins. A few minor navigation issues we're fixing on the fly, all else is fine.

It seems we can convert and run with minimal issues and then re-architect at our leisure. Far better than we could of hoped for!