.BOF and .EOF are lying!

[Laurel]

Are there steps one should take before requerying a form with different
retrieval arguments? If I reetrieve the data in the form and there are
rows, and then I change the values that are used as arguments to something
that would return no rows .EOF and .BOF are both false.

If the arguemtns that would return no rows are used first, .bof and .eof are
true as one would expect.

Any suggestions?

 

[Ken Snell [MVP]]

Can you post more details, such as sample data, the form's recordsource, the
code you're using, etc.? Kinda hard to suggest anything when we really don't
know what your form is doing.

 

[david epsom dot com dot au]

Use .recordcount > 0 to check if the recordset is empty.
Then use move or find if you want to position at a record.

Remember that by definition .RecordsetClone is a recordset
with different record pointers than those used by the form.

Ignore whatever the help files say about .bof and .eof
for recordset clones: The help was never correct.

(david)

 

[Laurel]

I got into testing .bof and .eof because I learned that you can't trust
..recordcount until you've done a .movelast. (And I have confirmed this
myself.... several times... Before moving to the last record, sometimes the
record count is 0. Sometimes it is the unfiltered count, etc.) Anyway,
testing for .eof and .bof was a way of avoiding an error when .movelast was
attempted on an empty recordset.

So here we are coming around the circle again....

I'm not using a record clone.

 

[david epsom dot com dot au]

you can't trust .recordcount until you've done a .movelast.

No, but I've always had > 0

Before moving to the last record, sometimes the record count is 0.

Ok, I've never seen that.

I'm not using a record clone.

What /ARE/ you doing?

What are you testing for BOF and EOF? Forms don't have a BOF
or EOF parameter. Please show your code.

(david)

 

[Laurel]

See below

No, but I've always had > 0

Well, you may be on to something. I went back to a form where I thought I
was getting 0 instead of the full count, but it turns out it was 1. So I'll
give .RecordCount = 0 a try.

Ok, I've never seen that.


What /ARE/ you doing?

For forms
Me.Recordset.RecordCount
or Me.Recordset.Bof, .EOF

or, sometimes
Dim lrst_Me As Recordset
Set lrst_Me = Me.Recordset

Very often I'm dealing with a recordset created this way:
Set irst_ClassPeriods = CurrentDb.OpenRecordset(ls_sql)

 

[david epsom dot com dot au]

Me.Recordset.EOF

ahh. I've never used me.recordset. I hope .recordcount
works, because I don't know any other solution.

(david)

 

[Ken Snell [MVP]]

You haven't shown us the specific circumstance when the BOF and EOF
properties "fail", but from your posts it appears that you are refreshing
the recordset's data (by a requery?) and then the "failure" occurs. Most
likely, the "failure" is because the refresh/requery has put you at a
different place in the recordset than where you thought you were?

I use BOF and EOF regularly with recordsets -- both those opened in code and
those for forms' recordset or recordsetclone. I have not have any difficulty
with either.

If you can post a specific example of what you are seeing (code and example
data), I'm sure we can assist you.

By the way, although the RecordCount of the form's recordset often can
mislead you because it doesn't have the "correct" count unless you do a
..MoveLast, the form's RecordsetClone.RecordCount property has always had a
correct count for me without having to use .MoveLast.

 

[david epsom dot com dot au]

"An OpenRecordset method internally invokes a MoveFirst
"method. Therefore, using an OpenRecordset method on an
"empty set of records sets the BOF and EOF properties to
"True.

AFAIK, nothing that doesn't invoke a Move method sets
the BOF or EOF properties.

For example

"Any Delete method, even if it removes the only remaining
"record from a Recordset, won't change the setting of the
"BOF or EOF property.

You might read that

"When you use Requery, the first record in the Recordset
"becomes the current record.

and

"If both the BOF and EOF property settings of the Recordset
"object are True after you use the Requery method, the query
"didn't return any records and the Recordset contains no data.

But if you read more carefully, you will notice that in
some cases "the Recordset is re-created from scratch.",
and in other cases "the Recordset is re-populated"

the recordset's data (by a requery?) and then the "failure" occurs. Most
likely, the "failure" is because the refresh/requery has put you at a
different place in the recordset than where you thought you were?

No, the "failure" is because the BOF and EOF properties are
not set at all by some actions.

(david)

 

[Ken Snell [MVP]]

I'm not sure if your post was meant for me or for Laurel.

Nonetheless, the information is consistent with what I am suggesting... that
the "failure" is not a "failure" but rather a misplaced expectation that the
BOF and EOF properties would have a specific value at a certain time, based
on what is being done in the code/recordset. And, as no specific example of
the circumstances surrounding the "failure" have been posted, it's difficult
to postulate the reason for the seeming "failure".

If you're reading my use of the term "put...at a different place" to mean
only a physical "place" then I apologize for the misunderstanding that it
seems to have caused.
--

Ken Snell
<MS ACCESS MVP>

 

[david epsom dot com dot au]

Ken,

We can argue about the difference between Failure and "Failure", but
whichever way you see it, the behaviour of Requery/BOF/EOF is not the same
as described in the help files.

If you're reading my use of the term "put...at a different place"

This is a misunderstanding. The failure/"failure" described by Laurel is not
because requery put something somewhere: it is because requery did not put
something anywhere.

Requery doesn't set BOF/EOF.

BOF and EOF retain their previous values or take random values when you do a
requery.

The values of BOF and EOF that you see after doing a requery are the same as
the values of EOF and BOF that you had before doing the requery or are
uninitialized.

It does not matter where requery puts you in the recordset: that has no
effect on BOF/EOF because the BOF/EOF properties are not
set/reset/changed/updated/affected or written by requery.

There is a particular (common) special case of requery that is an exception
to this rule because as well as executing the select query, it does an
implicit close recordset/open recordset.

posted, it's difficult to postulate the reason for the seeming "failure".

I accept that you haven't seen this behaviour, but I'm starting to get the
idea that you don't believe us?

http://bible.gospelcom.net/passage/?search=John 20:24-29

:~) Peace to you and those you love.

(david)

 

[Laurel]

Thank you for shedding light on a mystery!

"An OpenRecordset method internally invokes a MoveFirst
"method. Therefore, using an OpenRecordset method on an
"empty set of records sets the BOF and EOF properties to
"True.

AFAIK, nothing that doesn't invoke a Move method sets
the BOF or EOF properties.

For example

"Any Delete method, even if it removes the only remaining
"record from a Recordset, won't change the setting of the
"BOF or EOF property.

You might read that

"When you use Requery, the first record in the Recordset
"becomes the current record.

and

"If both the BOF and EOF property settings of the Recordset
"object are True after you use the Requery method, the query
"didn't return any records and the Recordset contains no data.

But if you read more carefully, you will notice that in
some cases "the Recordset is re-created from scratch.",
and in other cases "the Recordset is re-populated"



No, the "failure" is because the BOF and EOF properties are
not set at all by some actions.

(david)