Operations with time lapses - sql-server

Question

Let me explain the situation
I have two tables:
-One is to store the equipment faults, with their own Start Date and End Date.
-Other to store the breaks, each break have a Start Date and End Date.
For example the fault begin at 9:10 and finish at 9:16. The break begin at 9:14 and finish at 9:18, The fault time is 4 minutes And 2 minutes of the fault don't count because ocurred on a break. I need get the "2 minutes" value to include it in a report
To calculate it, I use a Scalar-valued Function:
DECLARE #Time int;
SET #Time = 0;
IF (#BreakStartDate > #FaultStartDate) AND (#BreakEndDate < #FaultEndDate)
SET #Time = DATEDIFF(SECOND, #BreakStartDate, #BreakEndDate);
ELSE IF (#FaultStartDate > #BreakStartDate) AND (#FaultEndDate < #BreakEndDate)
SET #Time = DATEDIFF(SECOND, #FaultStartDate, #FaultEndDate);
ELSE IF (#FaultStartDate < #BreakStartDate) AND ((#FaultEndDate > #BreakStartDate) AND (#FaultEndDate < #BreakEndDate))
SET #Time = DATEDIFF(SECOND, #BreakStartDate, #FaultEndDate);
ELSE IF (#FaultEndDate > #BreakEndDate) AND ((#FaultStartDate > #BreakStartDate) AND (#FaultStartDate < #BreakEndDate))
SET #Time = DATEDIFF(SECOND, #FaultStartDate, #BreakEndDate);
RETURN #Time
I need to validate all scenarios, if the fault start first and finish on the break, etc...
My question is, exist a function that do this automatically?
or a more elegant solution?

Answer 1

I was doing something similar with a time entry tool and needed to detect overlap between entered times. If all you really need is "I have a fault. There will be 1 Break during the Fault period. Give me the amount of time during which they're co-incident", then your function is on the right track.
When I did mine, I found exploring test cases visually was better than a truth table, and both were better than prose or code.
Case 1: Case 2:
Fault: |-------| Fault: |----------|
Break: |-------| Break: |----------|
Case 3: Case 4:
Fault: |----------| Fault: |----------|
Break: |----------| Break: |------|
...etc.
As for doing this elegantly, I'd point out that if there is any overlap between the two lapses, you always wind up wanting the larger start date. Like so:
Start Dates:
Fault: |---- ... |---- ... |----
Break: |---- ... |---- ... |----
Verdict: break either fault
...and similar for the smallest end date. Because our zone of interest is when both have started and neither have finished. So your logic can become:
DECLARE #Time int, #biggestStart datetime, #smallestEnd datetime;
SET #Time = 0;
IF (#BreakStartDate < #FaultEndDate) AND (#BreakEndDate > #FaultStartDate)
BEGIN
-- We have overlap.
-- Poor man's Math.Max and Math.Min:
SET #biggestStart = CASE WHEN #FaultStartDate > #BreakStartDate THEN #FaultStartDate ELSE #BreakEndDate END;
SET #smallestEnd = CASE WHEN #FaultEndDate < #BreakEndDate THEN #FaultEndDate ELSE #BreakEndDate END;
SET #Time = DATEDIFF(SECOND, #biggestStart, #smallestEnd);
END
RETURN #Time
If I had to do what you stated, this is probably what I'd go with.
That said, in analyzing your question, I'm wondering if you don't have to worry about multiple breaks per fault, if you are trying to do this for all faults in a result set, if you're aggregating "time per equipment per week" or something. If so, you want another approach. But that's another answer.

Answer 2

AFAIK there is no built in function for this. However, if you convert all dates to seconds since some random start date, I think you can calculate the time of the fault as (in pseudo code):
max(0; b1 - f1) - max(0; b2 - f2)
where b1, f1, b2 and f2 are given in the code below where I have put the SQL to calculate the above expression (untested):
DECLARE #startDate DATETIME
SET #startDate = '2010-01-01' -- Some date that must be smaller than all fault dates and break dates
DECLARE #breakStartDate DATETIME
DECLARE #breakEndDate DATETIME
DECLARE #faultStartDate DATETIME
DECLARE #faultEndDate DATETIME
DECLARE #b1 INT
DECLARE #b2 INT
DECLARE #f1 INT
DECLARE #f2 INT
-- Set values for break and fault dates here
SET #breakStartDate = ....
SET #breakEndDate = ....
SET #faultStartDate = ....
SET #faultEndDate = ....
SET #b1 = DATEDIFF(SECOND, #startDate, #breakStartDate)
SET #b2 = DATEDIFF(SECOND, #startDate, #breakEndDate)
SET #f1 = DATEDIFF(SECOND, #startDate, #faultStartDate)
SET #f2 = DATEDIFF(SECOND, #startDate, #faultEndDate)
DECLARE #time INT
SET #time =
CASE
WHEN #b1 - #f1 < 0 THEN 0
ELSE #b1 - #f1
END
+
CASE
WHEN #b2 - #f2 < 0 THEN 0
ELSE #b2 - #f2
END
SELECT #time
I think it is more elegant and less error prone but it's a matter of taste. It may be harder to read.