Master/Detail and using GUIDS

rbuser

Hi All,

I have a report with 2 data sets, a master and a detail. The datasets are
linked on a GUID. In the report I have some fields from the master set in
the detail pane and a subreport. I happen to know that for every master
record there should be 2 detail records, this is definitely true. In
practice the first record has about 20 child records, the next 10 or so have
none, the next 1 has loads then there's quite a few with none and so on and
so on.....

I am using RB7 SQLSever2000 and the ADO components in dade.

Are there any known issues with using GUIDS as link fields in RBuilder and
are there any work arounds? Have you any info that might help me in this
situation?

Thanks in advance

James Flannery

rbuser

Addition:

I should have said, the child dataset is ordered on the linking field and

the selection is all records from the master and only those in the child

that match. Also, I have just converted the tables to access and recreated

the whole thing using the a access tables. This exhibits the same issue.

Kind Regards

James Flannery

rbuser

Sorry Guys, another addition/change - it DOES work with the access data set.
SO it seems like the problem occurs on the SQLServer Platform.

Cheers,
James Flannery

digitalmetaphors

Yes, this was an issue we came across before. Check out a thread in the
DADE newsgroup on 8/9/2001 or search our newsgroups on GUID.


Cheers,

Jim Bennett
Digital Metaphors

rbuser

Are you saying then that between RB 5.56 and RB7 there's been no attempt to
fix this so instead force users to hack the source. This when MS promote
the use of GUIDs for linking more and more. Nice.

Richard

digitalmetaphors

Richard,

We have not yet committed resources to research and implement this feature
in RB. Thankyou for the suggestion that it should be added.


Cheers,

Jim Bennett
Digital Metaphors

rbuser

The problem is we can patch ppDB.pas (which I'm doing now) but then the
knock on is we'll have to maintain our own design time packages etc too so
that the RBDesigner correctly reflects the mod.

It's not nice when SQL Server is such a popular DB.

Richard

rbuser

Thanks for pointing us in the right direction

Cheers
James Flannery

rbuser

Jim

I've now reverse engineered the SQL Server collation rules and got a compare
function.

ppDB.pas

In CompareLinkedData I detect GUIDs by testing the string content of
lDetailFieldValue and call my compare function, but still the results are
not right - in fact no data of worth is being displayed now.

I've validate the collation function using another harness and this IS
producing the correct results.

Any other pointers?

Richard

rbuser

We have this working perfectly now.