It is nice to finally have FK RI working. This leads to the next
questions which deal with error recovery.

1) If a column constraint is violated (error 1582), I get back a
text message, but the column name would be nice, too. (Yes, I could
simply return the column name and look up that in an error message
table, but I am hoping for something simpler.)

2) Is it possible to find out which trigger failed? If I have a
table with two FKs, how do I tell which is wrong? In this case, there
is no unique text returned. Even getting the column name would help.
(If I violate the uniqueness primary key constraint, error 1884
returns the column name. I want something like that here.)

3) Is there any guarantee of which order the constraints are checked
/ the errors are thrown? If not, what has been your observation of
the order?

Sincerely,

Gene Wirchenko

Hi Gene,

first of all vfp creates vfp code in the dbc, so everything
the ri code does can be seen and/or modified the way you'd
like it. This also regards the order of tests. Other third party
ri code builders may be a better choice, as they normally
create code better understandable. TaxRI and AMRI are
two I know of. AMRI eg creates a public var _RI_Message
which makes it easy to react to trigger errors.

of the array created.

on element 5. Normally that is NULL, but with error
1539 this has info about the trigger type.

full record (all FKs) in their field order, then it recurses
for cascades. But I'm not that familiar with vfp native RI
code. I only know there were an error with multiple
cascades not rolling back deletes when stepping to some
restrict. As you use vfp6 you may still have these errors.
I'd recommend to use some other RI builder.

Bye, Olaf.

On Sat, 28 Jan 2006 19:36:05 +0100, "Olaf Doschke"
<T2xhZi5Eb3NjaGtlQFNldG1pY3MuZGU (AT) strconv (DOT) 14
<T2xhZi5Eb3NjaGtlQFNldG1pY3MuZGU (AT) strconv (DOT) 14> wrote:

Pointers?

I am already looking at those two. They are not giving quite
enough information.

It says what kind of trigger failed (insert, update, or delete),
but not why. If I have two FKs in a row, which one was wrong?

[snip]

Sincerely,

Gene Wirchenko

You can find AMRI at ftp://ftp.prolib.de/Public/VFP/AMRI400.zip

Bye, Olaf.

On Mon, 30 Jan 2006 23:56:40 +0100, "Olaf Doschke"
<b2xhZi5kb3NjaGtlQHQtb25saW5lLmRl.strconv.14 (AT) t-online (DOT) de> wrote:

Thanks. I have downloaded them and will have a look later.

Sincerely,

Gene Wirchenko