A recent mail merge resulted in a Catalog (using Word 2003, so a Directory)
with random missing records. I believe I know what happened. It's a gotcha
that one would need to be aware of.

Using the Mail Merge Wizard, I setup a merge operation knowing that the
Access table had invalid records that would be printed in the Directory. At
one phase in the Wizard, one is given the opportunity to exclude records by
unchecking a box.

Later, I realized I needed to sort these records. (And one has to go
through the whole process of re-connecting to the database in order to get
at the sort mechanism.)

Now, the Dirctory was including the invalid records and dropping random (or
so it seemed) valid records.

Without any rigorous experimentation, it appears that Word 2003 is
remembering the excluded record positions across re-connections, so that
when the newly retrieved recordset is sorted, different records are
excluded than what were excluded earlier - based on record position, not
primary key index values (if present).

Can I get a corroborating experiment on this?

Any manipulation of data should be done in the data source (Access). I
would never dream of:

1. Using Word to filter data if the data source is Access

2. Using the Wizard - Use the Mailmerge toolbar instead

With the mailmerge toolbar, the third button from the left hand end allows
you to view the data source and set and remove filters.


--
Hope this helps.

Please reply to the newsgroup unless you wish to avail yourself of my
services on a paid consulting basis.

Doug Robbins - Word MVP

"Brian Smither" wrote in message
...
A recent mail merge resulted in a Catalog (using Word 2003, so a Directory)
with random missing records. I believe I know what happened. It's a gotcha
that one would need to be aware of.

Using the Mail Merge Wizard, I setup a merge operation knowing that the
Access table had invalid records that would be printed in the Directory.
At
one phase in the Wizard, one is given the opportunity to exclude records
by
unchecking a box.

Later, I realized I needed to sort these records. (And one has to go
through the whole process of re-connecting to the database in order to get
at the sort mechanism.)

Now, the Dirctory was including the invalid records and dropping random
(or
so it seemed) valid records.

Without any rigorous experimentation, it appears that Word 2003 is
remembering the excluded record positions across re-connections, so that
when the newly retrieved recordset is sorted, different records are
excluded than what were excluded earlier - based on record position, not
primary key index values (if present).

Can I get a corroborating experiment on this?

Hi Brian,

This isn't something I've ever tried... But I do know that the checkboxes in
the Recipient's dialog box use a different mechanism than the "Advanced"
filtering options. I NEVER use the checkboxes for anything "serious", just
one-off's.

What you describe jives with the way the checkbox stuff is handled in the
object model (VBA).

If you don't/can't use a query in Access as Doug suggests to pre-filter (and
pre-soft) the data coming across (drop the invalid records), I highly
recommend using the "Advanced" entry at the bottom of the dropdown list to set
a proper filter that WILL hold when the data source sorts or changes.

A recent mail merge resulted in a Catalog (using Word 2003, so a Directory)
with random missing records. I believe I know what happened. It's a gotcha
that one would need to be aware of.

Using the Mail Merge Wizard, I setup a merge operation knowing that the
Access table had invalid records that would be printed in the Directory. At
one phase in the Wizard, one is given the opportunity to exclude records by
unchecking a box.

Later, I realized I needed to sort these records. (And one has to go
through the whole process of re-connecting to the database in order to get
at the sort mechanism.)

Now, the Dirctory was including the invalid records and dropping random (or
so it seemed) valid records.

Without any rigorous experimentation, it appears that Word 2003 is
remembering the excluded record positions across re-connections, so that
when the newly retrieved recordset is sorted, different records are
excluded than what were excluded earlier - based on record position, not
primary key index values (if present).

Can I get a corroborating experiment on this?


Cindy Meister
INTER-Solutions, Switzerland
http://homepage.swissonline.ch/cindymeister (last update Jun 8 2004)
http://www.word.mvps.org

This reply is posted in the Newsgroup; please post any follow question or
reply in the newsgroup and not by e-mail :-)