Hi all,

I have encountered a strange problem in my process of migrating existing FM6 files over to FM8.

The previous developer, in FM6, had created a relationship from one file to another keying from FileA global text field to fileB indexed calculation field with text results. Everything works in FM6 but in FM7 and 8 it does not.

So I did a lot of tracing and I have isolated the problem to very poor design (the whole thing is a mess to be accurate and they wont let me redesign it). Anyway, the previous guy made it so that in File A the data in the global key field would have something like:

03-001 Strange Occurance Project

And on the child side File B, there are three fields:

Project number

Project Name

and a calc field that concats the two other fields with an underscore. Basically resulting, 03-001_Strange Occurance Project

Now I would never have something like this act as keys; with the parent having spaces and the child having underscores. So I understand why it doesnt work in 7 and 8 because of the key mismatch.

My QUESTION is why does FM6 allow this???

This is stupid. Now I have to figure out how I am going to have to examine all these relationships (and there are hundreds of them) to see if these kind of mismatched keys are used just because of poor design.

I have attached a sample copy of the FM6 files to demonstrate what I am talking about. It is baffling to me.

underscore.zip

bump.

anybody have any ideas?

I don't remember this specific change, but there were several changes to the way FM7/8 indexes words vs. prior versions. I think this was covered in the Migration Foundations & Methodologies brief. It may have been a discussion about value lists or the leftwords(), rightwords() functions, but your case seems to indicate it applies to matching in relationships as well.

Apparently, FM5/6 does not treat the underscore as a character in the index.

Thanks Ender.

I knew that indexing had changed in FM7. I guess I didnt realize until now that the underscore had previously not been an indexed character in prior versions.

This is really interesting. I could have sworn I'd seen a list of (new) word separators but can't find it anywhere. 'Converting from Previous Versions' makes a very brief reference to Left/MiddleWords etc. and mentions that hyphen is now (V7-8) a word break. I checked earlier documentation and it appears that this is one of those vague FMP areas where the full specs were never spelled out.

The new Help simply says that & and - denote the beginning of a new word. Interestingly, my old V5 manual specifically says to use underscore (_) in place of spaces in field names to "avoid restrictions in ODBC ...".

You have my sympathy. :therethere:

Comment posted about it here which is probably the one you're thinking of.

But since the parent is a global it would be easy to change, I would think; although it sounds like you have more than one possible mismatch here. Why in the world include the name in the key - that's what IDs are for. I also sympathize ...