|
|
|
date: Thu, 29 May 2008 11:19:31 +0200,
group: microsoft.public.access.replication
back
Re: Access 2002/2003 - 2007 replication problem
Hi David
The logic you mentioned sounds right, but if I copy the same two replcas to
a machine with Office 2003,
replication succeed.
It is easy to reproduce this error:
First make a a simple table in Access 2007 2002/2003 mdb format, add 1
column datatype= integer,and add 1 row with the value =1
Then turn that database to a master design, and make a replica of that.
Replicate direct between the master and the replica.
Then change the datatype to "TEXT" and then replicate direct again, then you
get this error.
It should be posible to change datatype from integer to text, but not
opposite as I understand your logic.
Even more strange, if I copy theese two replicas to a box with Office 2003,
it is then posible to replicate these changes without errors.
There must be a difference between Office 2007 and 2003, even we are usisng
.mdb level in both scenarios, and not .accdb file type.
Regards
JensB
"David W. Fenton" <XXXusenet@dfenton.com.invalid> wrote in message
news:Xns9AADC5322DB6Cf99a49ed1d0c49c5bbb2@64.209.0.93...
> "JensB" wrote in
> news:uLkoRsWwIHA.1240@TK2MSFTNGP02.phx.gbl:
>
>> Just as a test, I took the a copy of the masterdesign and
>> dereplicated it and cleaned it and moved it to the server running
>> office 2007. Then made a new Master design and a replica in the
>> 2007 environment. but the same error comes up.
>>
>> Any explantion to this behaviar?
>
> It is usually because the data in the replicas is in violation of
> either validation rules or referential integrity that was
> implemented in the design change that is failing.
>
> Remember:
>
> In a single synch, design changes propagate before data updates.
>
> This means that if you're implementing a validation rule, you need
> to make sure that the data in all the replicas conforms to that
> validation rule *before* you implement the validation rule.
>
> This means:
>
> Update all the data to be consistent with the new validation rule.
>
> Synch all the replicas.
>
> Only *then* can you implement the new validation rule and synch.
>
> If you try to do both steps with one synch, it will fail, because
> the design change will always be applied before any updates within a
> single synch operation.
>
> --
> David W. Fenton http://www.dfenton.com/
> usenet at dfenton dot com http://www.dfenton.com/DFA/
date: Tue, 3 Jun 2008 23:04:38 +0200
author: JensB
Re: Access 2002/2003 - 2007 replication problem
"JensB" wrote in
news:uCGIhtbxIHA.5288@TK2MSFTNGP06.phx.gbl:
> The logic you mentioned sounds right, but if I copy the same two
> replcas to a machine with Office 2003,
> replication succeed.
Then there's something wrong with the original replicas (copying
creates new replicas that then shed certain aspects of the history
they had with the old ReplicaID).
> It is easy to reproduce this error:
> First make a a simple table in Access 2007 2002/2003 mdb format,
> add 1 column datatype= integer,and add 1 row with the value =1
> Then turn that database to a master design, and make a replica of
> that. Replicate direct between the master and the replica.
> Then change the datatype to "TEXT" and then replicate direct
> again, then you get this error.
> It should be posible to change datatype from integer to text, but
> not opposite as I understand your logic.
I don't know. I've never had to make that kind of change. Sounds
like the type of situation where it would be better to create a new
column of the proper data type, synch around the whole replica set,
and then use a SQL UPDATE to copy the numeric data into the text
column.
> Even more strange, if I copy theese two replicas to a box with
> Office 2003, it is then posible to replicate these changes without
> errors. There must be a difference between Office 2007 and 2003,
> even we are usisng .mdb level in both scenarios, and not .accdb
> file type.
Well, of *course* you're using MDB, as ACCDB has no support for
replication.
You're doing something wrong, obviously, but given your willy-nilly
copying of replicas around, I'd suggest that the problem has to do
with sloppy use of replication.
--
David W. Fenton http://www.dfenton.com/
usenet at dfenton dot com http://www.dfenton.com/DFA/
date: 11 Jun 2008 19:19:00 GMT
author: David W. Fenton lid
|
|