Opening DAO Recordset dbSnapshot vs dbForwardOnly
I have a number of routines in a Access 2003 application I am working on that are purely for data retrieval. No data editing or revision and the records are read in the order returned.
The structure and function of these routines is all the same. They are called interactively from another routine to pull information from a specific record in either a table or query. In most cases only the first record of the dataset is read to get the information required. If the dataset is EOF, the information was not found.
Since I'm making changes anyway I want to code these type of routines in the most efficient way.
In these routines, I noticed some cases the developer used
set rs = db.Openrecordset (selectstring, dbOpenSnapshot, dbReadonly)
and in others the developer uses
set rs = db.Openrecordset (selectstring, dbOpenForwardOnly, dbReadonly)
My question is: In this scenario and type of usage which of these options (dbOpenSnapshot or dbOpenForwardOnly) is the most efficient, or is there any real difference?
Here's an example of this type of routine.
NOTE: In the example below I realize that MoveLast and MoveFirst would need to be removed if dbForwardOnly is used but that is easily done and the recordcount will always return 1. I think those statements are remnants of the original ADO recordset handling.
Example:
Public Sub getInstallPlanInfoFromProp
returnPAY_AGREE_TYPE As String, _
returnPAY_AGREE_START_DATE
returnPAY_AGREE_END_DATE As Date, _
returnPAYMENT_AGREEMENT_TY
returnPAY_AGREE_TYPE = cNoRecFound
returnPAYMENT_AGREEMENT_TY
'
Dim wkPaidToDate As Double
'
selectString = "Select [PAY_AGREE_TYPE], [PAY_AGREE_START_DATE], [PAY_AGREE_END_DATE], [PAYMENT_AGREEMENT_TYPE] " & _
" From qryInstallPay_Main_ForCOPD
" Where ([PropertyID] = " & passedPropertyID & " ) "
'
'
Set db = getCurrentDbC
'
Dim rs As DAO.Recordset
Set rs = db.OpenRecordset(selectStr
'
If rs.EOF Then
Else
rs.MoveLast
rs.MoveFirst
If rs.RecordCount > 0 Then
'
returnPAY_AGREE_TYPE = Nz(rs!PAY_AGREE_TYPE, 0)
'
'////old logic below, revised 09/27/16
'returnPAY_AGREE_START_DAT
'returnPAY_AGREE_END_DATE = Nz(rs!PAY_AGREE_END_DATE, 0)
'
' new logic 09/27/16
returnPAY_AGREE_START_DATE
returnPAY_AGREE_END_DATE = Nz(rs!PAY_AGREE_END_DATE, cNoDateFound) ' return cNoDateFound if Null 09/27/16
'
returnPAYMENT_AGREEMENT_TY
'
End If
End If
'
rs.Close
Set rs = Nothing
'
End Sub
The structure and function of these routines is all the same. They are called interactively from another routine to pull information from a specific record in either a table or query. In most cases only the first record of the dataset is read to get the information required. If the dataset is EOF, the information was not found.
Since I'm making changes anyway I want to code these type of routines in the most efficient way.
In these routines, I noticed some cases the developer used
set rs = db.Openrecordset (selectstring, dbOpenSnapshot, dbReadonly)
and in others the developer uses
set rs = db.Openrecordset (selectstring, dbOpenForwardOnly, dbReadonly)
My question is: In this scenario and type of usage which of these options (dbOpenSnapshot or dbOpenForwardOnly) is the most efficient, or is there any real difference?
Here's an example of this type of routine.
NOTE: In the example below I realize that MoveLast and MoveFirst would need to be removed if dbForwardOnly is used but that is easily done and the recordcount will always return 1. I think those statements are remnants of the original ADO recordset handling.
Example:
Public Sub getInstallPlanInfoFromProp
returnPAY_AGREE_TYPE As String, _
returnPAY_AGREE_START_DATE
returnPAY_AGREE_END_DATE As Date, _
returnPAYMENT_AGREEMENT_TY
returnPAY_AGREE_TYPE = cNoRecFound
returnPAYMENT_AGREEMENT_TY
'
Dim wkPaidToDate As Double
'
selectString = "Select [PAY_AGREE_TYPE], [PAY_AGREE_START_DATE], [PAY_AGREE_END_DATE], [PAYMENT_AGREEMENT_TYPE] " & _
" From qryInstallPay_Main_ForCOPD
" Where ([PropertyID] = " & passedPropertyID & " ) "
'
'
Set db = getCurrentDbC
'
Dim rs As DAO.Recordset
Set rs = db.OpenRecordset(selectStr
'
If rs.EOF Then
Else
rs.MoveLast
rs.MoveFirst
If rs.RecordCount > 0 Then
'
returnPAY_AGREE_TYPE = Nz(rs!PAY_AGREE_TYPE, 0)
'
'////old logic below, revised 09/27/16
'returnPAY_AGREE_START_DAT
'returnPAY_AGREE_END_DATE = Nz(rs!PAY_AGREE_END_DATE, 0)
'
' new logic 09/27/16
returnPAY_AGREE_START_DATE
returnPAY_AGREE_END_DATE = Nz(rs!PAY_AGREE_END_DATE, cNoDateFound) ' return cNoDateFound if Null 09/27/16
'
returnPAYMENT_AGREEMENT_TY
'
End If
End If
'
rs.Close
Set rs = Nothing
'
End Sub
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Of course, the best solution is to not use a cursor at all. Or,since you are only ever using the "first" record, change the query to return the Top 1 rows. Then if the recordset is not at EOF, you have a record.
Running .MoveLast forces Access to bring the ENTIRE recordset down from the server. What is the point if you are only ever using one record?
Running .MoveLast forces Access to bring the ENTIRE recordset down from the server. What is the point if you are only ever using one record?
ASKER
Thanks you confirmed what I was thinking
ASKER
Pat: Great idea and sorry, I had already awarded points when your post showed up.
Could you explain how to do this without a cursor, even though I unintentionally stiffed you on points?
Could you explain how to do this without a cursor, even though I unintentionally stiffed you on points?
If this is an attempt to update a record, just run an update query. Otherwise modify the query as I suggested so it only ever returns a single row.
ASKER
Thanks Pat, no updating in these just data retrieval and usually only the first record is needed. In most of these queries only one matching record or zero records if there is no match will be returned
I'll make the revision to Top 1, dbOpenForwardOnly.
I'll make the revision to Top 1, dbOpenForwardOnly.
<<Otherwise modify the query as I suggested so it only ever returns a single row.>>
He had a WHERE on that SQL, so I would assume it was calling a single row already, but had the flexibility to get more if need be, which is the way I would have left it. Not sure what all the logic needs to be though.
jim.
He had a WHERE on that SQL, so I would assume it was calling a single row already, but had the flexibility to get more if need be, which is the way I would have left it. Not sure what all the logic needs to be though.
jim.
By the description, he was only using the first record and didn't seem to be giving any error messages if there were more than one so I was attempting to remove the ambiguity. The code was probably copied from some other place and that is what caused the confusion.
<< The code was probably copied from some other place and that is what caused the confusion.>>
Yes, it looks like it has morphed over time.
Jim.
Yes, it looks like it has morphed over time.
Jim.
ASKER
You are both right, there have been many fingers in this pie, each with their own techniques.
I noticed something else but I'll post another question for it. As always ,thanks for your help in this one.
I noticed something else but I'll post another question for it. As always ,thanks for your help in this one.
This:
rs.MoveLast
rs.MoveFirst
If rs.RecordCount > 0 Then
is a real performance drain. As you already noticed, .RecordCount will be 1 or more if there are records, 0 if none. So if your just trying to determine if there are records, there is no reason to do a .MoveLast then first. You can easily use BOF and EOF as well.
Also note that ADO is dfferent; it will hand you the correct record count right off.
Jim