Can't Open Data Source (Query) where criteria is defined by functi

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

l_stocky - 14 Jan 2005 10:39 GMT

This question relates to the Automation of Word Mail Merge from Access. I
wasn't sure whether to post it here or on the Access Newsgroup.
versions: Word 2000, Access 2000, Windows XP Pro.

I have a Word Document. It's Datasource is set to a predefined query in an
Access DB. The query has criteria which is defined by functions set at
realtime. i.e. the 'CaseID' column has 'getCaseID()' in its criteria. This
will return an integer which will have the query select a single row of data.
This query works correctly within Access and correctly when attempting to
'Edit' the source from Word.

What I am attempting to do is automate the mail merge process from Access.
This is where the problem shows itself. I can open the Word document but if I
try to set the datasource to the Access query I get an error message: "Word
cannot open the datasource" and sometimes "Automation Error".
If I replace the functions in the query's criteria with "[]" (so the
criteria is manually set at runtime), Word opens the datasource successfully
and I can execute the mail merge. If I set the criteria with appropriate
data, again the mail merge process is successful using automation.

So it seems that Word can't open the datasource when directed to do so by
automation where the datasource involves function based criteria.

I have put a breakpoint on the function and it is clear they are being
consulted and are being assigned correctly.

Is this a known problem? I see lots of aritcles about parametrised queries
but none take into account setting parameters by function. (see kb209976).

Is there a workaround?

Reply to this Message

Peter Jamieson - 16 Jan 2005 14:49 GMT

I tried this using a simple example here and managed to get it all to work,
which suggests that something specific either in your function (or perhaps
the query) or your configuration is causing a problem. The main visible
configuration difference was that I was using Win2K not WinXP, but I doubt
if that is a factor.

Does your example work when you do not try to automate Word from the
database, i.e. if you just use Word VBA to set or change the data source to
be the query you want? In other words, is the fact that you are automating
Word from your daatabase a factor (here, it does not appear to make any
difference).

In my simple example, an Access 2000 .mdb called c:\a\a.mdb with a
two-column table t1 with columns k (an integer) and t (text), and a small
number of rows.
I have a query q1 which does
SELECT t1.*
FROM t1
4WHERE ([t1].[k] = myrow())

(There may be some more [] in there than I have written above).

I have a Word document c:\a\atest.doc that just contains mergefields k and
t.

I have a module ("Module1" in the database containing
a. a function myrow() which simpy sets myrow equal to one of the values in
k
b. a function mymerge which does

dim oApp As Word.Application
dim oDoc As Word.Document
Set oApp = CreateObject("Word.Application")
oApp.Visible = True
Set oDoc = oApp.Documents.Open("c:\a\atest.doc")
oApp.Visible = True
oDoc.MailMerge.MainDocumentType = wdNotAMergeDocument
oDoc.MailMerge.MainDocumentType = wdFormLetters
oDoc.MailMerge.OpenDataSource _
Name:= "C:\a\a.mdb", _
Connection:="TABLE t1", _
SQLStatement:= "SELECT * FROM [t1]"
Debug.Print oDoc.MailMerge.DataSource.Connectstring
Debug.Print oDoc.MailMerge.DataSource.Querystring
oDoc.MailMerge.DataSource.Querystring = "SELECT * FROM [t1]"
Debug.Print oDoc.MailMerge.DataSource.Connectstring
Debug.Print oDoc.MailMerge.DataSource.Querystring

' followed by any Set oDoc = Nothing type cleanup required, but I leave the
document open to check the results.

I wonder if a similar simple example works on your system? If so, it would
suggest that there is something different about your database, table or
query. I do not think it can be the multi-user settings or it would still
fail when you remove the function from the query.

BTW, this may work in Word 2000 but automating anything that relies on the
DDE connection method may fail in Word 2002 or later, and using user-defined
functions does require you to use DDE.

Peter Jamieson

> This question relates to the Automation of Word Mail Merge from Access. I
> wasn't sure whether to post it here or on the Access Newsgroup.
[quoted text clipped - 30 lines]
>
> Is there a workaround?

Reply to this Message

l_stocky - 18 Jan 2005 10:51 GMT

Peter - many thanks for your help thus far.

I have managed to successfully replicate your example. However, I don't see
that it tests the myrow() function, which is where the potential problems may
lie.
Can you extend/adjust the code to test the myrow() function, and q1 which is
redundant from what I can see in the example as it stands.

I appreciate you may have avoided this to test the simple functionality of
the mailmerge process and to illiminate the fact that something in my
configuration is causing the problem.
Short of the fact that a second instance of the database is loaded the whole
process is performed properly.

I would really appreciate your further help on this,
Lloyd

> I tried this using a simple example here and managed to get it all to work,
> which suggests that something specific either in your function (or perhaps
[quoted text clipped - 91 lines]
> >
> > Is there a workaround?

Reply to this Message

Peter Jamieson - 19 Jan 2005 10:55 GMT

Apologies, the following code

oDoc.MailMerge.DataSource.Querystring = "SELECT * FROM [t1]"

should say

oDoc.MailMerge.DataSource.Querystring = "SELECT * FROM [q1]"

(and there obviously should not be a digit "4" jut in front of the WHERE in
the SQL clause used to set up q1).
Testing q1 implies testing or myrow()

Peter Jamieson

> Peter - many thanks for your help thus far.
>
[quoted text clipped - 133 lines]
>> >
>> > Is there a workaround?

Reply to this Message

l_stocky - 21 Jan 2005 12:31 GMT

I've made the necessary alteration. However, the resulting word document
scrolls through ALL the records?
The myrow() function looks as follows:
Function myrow()
myrow = 12
End Function

There are two records which should be selected based on that assignment.
Instead, it is possible to scroll through all the records. Interestingly
enough, the myrow() function isn't even being consulted as proved by putting
a breakpoint on it. The breakpoint is never met so I guess the function is
never called.

There still must be something wrong with the code isn't there Peter?
Does it have anything to do with: SQLStatement:= "SELECT * FROM [t1]"

> Apologies, the following code
>
[quoted text clipped - 147 lines]
> >> >
> >> > Is there a workaround?

Reply to this Message

Peter Jamieson - 21 Jan 2005 15:01 GMT

> There still must be something wrong with the code isn't there Peter?

There could be, but if so, at present I can't see it. Unfortunately at the
moment I have to retype code when replying.

> Does it have anything to do with: SQLStatement:= "SELECT * FROM [t1]"

The idea behind that bit of code is simply to set up a connection with a
table, i.e. something that should always work.

If you want, you can attempt the connection to q1 straight away instead,
i.e. in the OpenDataSource statement.

However, I doubt if that is the problem here.

What code do you currently have for q1, and what happens when you run the
query directly in Access?

Peter Jamieson

> I've made the necessary alteration. However, the resulting word document
> scrolls through ALL the records?
[quoted text clipped - 185 lines]
>> >> >
>> >> > Is there a workaround?

Reply to this Message

l_stocky - 21 Jan 2005 17:35 GMT

aha
...I've tried changing the SQL statement to: SQLStatement:="SELECT * FROM
[q1]" and it works. The resulting Word file only has two records. These same
two records are the result if you run the query directly from Access.

A few strange things:
1. TWO additional instances of the database are being opened during the
mailmerge process.
2. Only when I run the query directly from Access does the breakpoint show
that the myrow() function is being called. My theory is that it queries the
second or third instance of Access(see point 1), which since it is opened
during the mail merge process doesn't have a breakpoint on it.
I have more-or-less proved this by: closing the additional instances of the
db, changing the value of myrow, saving the module and then running the mail
merge process again.

Need help to stop the extra instances of Access loading up. I've got to
implement this in an office system - I can't have that happening.

> > There still must be something wrong with the code isn't there Peter?
>
[quoted text clipped - 205 lines]
> >> >> >
> >> >> > Is there a workaround?

Reply to this Message

Peter Jamieson - 22 Jan 2005 05:09 GMT

> 1. TWO additional instances of the database are being opened during the
> mailmerge process.

As a first shot, some guesses....

If the document you are opening from Access has been saved as a mail merge
main document with an Access data source, Word will probably try to open
that data source when it opens the document, and in those circumstances it
may create another instance of Access - almost certainly so if the database
is different from the one you have open already.

If the document you are opening is actually a template and Word actually
tries to create a new document based on the template, both the template and
the new document have a data source attached, so Word will try to open the
data source twice.

However, even if either of those things is true, it does not really explain
why Word would open another instance when you already have the database
open. The following articles might provide a clue:

http://support.microsoft.com/default.aspx?scid=kb;en-us;224697
http://support.microsoft.com/default.aspx?scid=kb;en-us;199963

Although the first article says it applies to Word 97 and that the problem
was corrected in Word 2000, I have a suspicion that a similar problem with
DDE may be occuring, but for slightly different reasons than are mentioned
in the article. One possibility is that if the full path name of your .mdb
is long or contains space characters etc., DDE may not match the database
name correctly and may try to open another instance. Another possibility is
that if you have a window such as a table design window open in Access
(quite likely when you're developing/debugging etc.) DDE may try to open
another instance. I don't think it does - I think it simply fails - but it's
worth checking.

Ignore the ODBC workaround in these articles - it does not work with Access
queries that use user-defined functions such as myrow().

> Need help to stop the extra instances of Access loading up. I've got to
> implement this in an office system - I can't have that happening.

Yes, unfortunately the way Word and DDE work together make it very difficult
to control this. Using ODBC to connect avoids opening any instances of
Access, but then you have to consider the following:
a. does ODBC work when when the database is already open?
b. if the database is protected (either with a database password or with
user-level/workgroup security) you may not be able to access it with ODBC
despite the fact that it is already open (in fact I think you are more
likely to be able to do this with workgroup security than a
password-protected database)
c. you won't be able to use user-defined functions to specify the rows you
want, so you have to use a different approach. That obviously depends on
what you are attempting, but some possibilities are:
- use Access to flag the records you need (or the ones you don't) and
construct a query that relies on selecting flagged/unflagged records. Not
usually a good approach in my experience.
- use Access to create a table with one record for each row you want to
include, and construct a query that uses that info to extract the records
you need. May be OK if only one user accesses the database at a time.
- use code to generate the precise query you need and either
- use that to create a query in Access, then use that as the data
source (never tried it) or
- issue that query in your Word OpenDataSource.
In both cases you would probably face limitations in terms of query
length and probably complexity (e.g. there could be a limit on the number of
ORs in a WHERE clause). A spcific limitation is that OpenDataSource cannot
issue a query longer than either around 256 chars. or 512 chars.

Best I can do for now,

Peter Jamieson

> aha
> ...I've tried changing the SQL statement to: SQLStatement:="SELECT * FROM
[quoted text clipped - 254 lines]
>> >> >> >
>> >> >> > Is there a workaround?

Reply to this Message