Use datediff() result in dsum() function - sql

Question

I have the following table with start and end dates:
dataid TS EndTS
1744 7/27/17 1:57:34 PM 7/27/2017 1:57:38 PM
1743 7/27/17 1:57:31 PM 7/27/2017 1:57:34 PM
1742 7/27/17 1:57:23 PM 7/27/2017 1:57:31 PM
1741 7/27/17 1:57:16 PM 7/27/2017 1:57:23 PM
1740 7/27/17 1:57:04 PM 7/27/2017 1:57:16 PM
1739 7/27/17 1:56:57 PM 7/27/2017 1:57:04 PM
1738 7/27/17 1:56:38 PM 7/27/2017 1:56:57 PM
I would like to get the date/time interval (in seconds) and then calculate a running total.
Here is what I have so far:
SELECT
[dataid] AS [dataid]
DateDiff("s", [TS],[EndTS]) AS [durationsec]
DSum("[durationsec]","[HX32]","[dataid] <=" & [dataid]) AS [add]
FROM [HX32];
I think the datediff() funtion is possibly causing formatting problems. With "[durationsec]" I get all nulls as a result, with [durationsec] I get the following results:
durationsec add
4 6896
3 5169
8 13776
7 12047
12 20640
7 12033
19 32642
I also tried cint(DateDiff("s", [TS],[EndTS])) No change.
I also tried passing durationsec to a table and running a seperate query. No change. (Also I would prefer to do this all in one query)
Here are the results I would like to achieve:
dataid TS EndTS durationsec add
1744 7/27/17 1:57:34 PM 7/27/2017 1:57:38 PM 4 60
1743 7/27/17 1:57:31 PM 7/27/2017 1:57:34 PM 3 56
1742 7/27/17 1:57:23 PM 7/27/2017 1:57:31 PM 8 53
1741 7/27/17 1:57:16 PM 7/27/2017 1:57:23 PM 7 45
1740 7/27/17 1:57:04 PM 7/27/2017 1:57:16 PM 12 38
1739 7/27/17 1:56:57 PM 7/27/2017 1:57:04 PM 7 26
1738 7/27/17 1:56:38 PM 7/27/2017 1:56:57 PM 19 19
Thanks, I'm a beginner.

Answer 1

Time is not seconds but integer days, so you could try:
SELECT
[dataid],
DateDiff("s", [TS], [EndTS]) AS [durationsec],
DSum("[EndTS]-[TS]", "[HX32]", "[dataid] <= " & [dataid] & "") * 86400 AS [add]
FROM
[HX32];
That said, June's method should work as well. If both fail, something else is going on.

Answer 2

SELECT dataid,
startts,
endts,
DATEDIFF("s", startts,endts) AS durationsec,
SUM(DATEDIFF("s", startts,endts)) OVER (ORDER BY endts ROWS UNBOUNDED PRECEDING) AS runningtotal
FROM durtab
ORDER BY 5 DESC;
Result:
dataid startts endts durationsec runningtotal
1744 2017-07-27 13:57:34.000 2017-07-27 13:57:38.000 4 60
1743 2017-07-27 13:57:31.000 2017-07-27 13:57:34.000 3 56
1742 2017-07-27 13:57:23.000 2017-07-27 13:57:31.000 8 53
1741 2017-07-27 13:57:16.000 2017-07-27 13:57:23.000 7 45
1740 2017-07-27 13:57:04.000 2017-07-27 13:57:16.000 12 38
1739 2017-07-27 13:56:57.000 2017-07-27 13:57:04.000 7 26
1738 2017-07-27 13:56:38.000 2017-07-27 13:56:57.000 19 19

Answer 3

DSum is looking at the [HX32] table or query to find a field named [durationsec]. It doesn't exist there.
SELECT
[dataid],
DateDiff("s",[TS],[EndTS]) AS [durationsec],
DSum("DateDiff('s',[TS],[EndTS])","[HX32]","[dataid] <=" & [dataid]) AS [add]
FROM [HX32];`
Note the use of apostrophes to delimit the 's' parameter in the nested DateDiff.
An alternative approach is to do the running sum in Report because textbox in report has a RunningSum property. Domain aggregate functions in queries can perform slowly in large datasets.