The date columns are of data type DATE? It's not confirmed by question. For this answer the dates must be stored as DATE.

;with example_cte as (
 ...

• This reply was modified 2 years, 9 months ago by Steve Collins. Reason: Removed two unnecessary pairs of parentheses

Thank you Jeff.  44.9% or 31%?  Either way there's no implicit type conversion 🙂

Ken McKelvey's solution in the "Pivot with grouping intervals" topic is interesting.  It cross join's a 2...

The top two CTEs could be replaced with only one.  Possibly the bottom two could be combined as well using the "smudge" method?

with 
pvt_cte as (
 ...

Afaik this returns the same output.  The CTEs expand the date intervals, pivot by role name and summarize by day, identify gaps in GROUP BY columns, and create groups using...

From your query results you could CROSS APPLY a row generator created by using the row cardinality from sys.all_columns (which in my test instance is 11,745 rows) and TOP.  The...

• This reply was modified 2 years, 9 months ago by Steve Collins.

You could test if the modulo .1 of HRS equals zero

;WITH cte_data AS (
    SELECT CAST(50.10 AS numeric(5, 2)) AS HRS
   ...

Also there's EOMONTH which has an optional second parameter 'offset'.  Regardless of which date/datetime datatype is passed to the EOMONTH function it returns DATE.  Ha, some SSC members were wondering...

All perfect parents have the same thing in common.  No kids

Also my code assumes the RequestNo is UNIQUE NOT NULL and has appropriate sequential values.  The DDL provided says the RequestNo is NULL-able

drop TABLE  if exists #F6000059;
go
CREATE TABLE #F6000059
(
    [RequestNo] [int] NULL,
    [REAN82] [int] NULL,
    [REAN83] [int] NULL,
 ...

• This reply was modified 2 years, 10 months ago by Steve Collins.

It seems like you're looking for an UPDATE statement

/* select using OUTER APPLY and TOP(1) */
select t.id, t.code, t.result, prev.result prev_result, 
     ...

...

In source table each row is allways for one date, so if the work shift continues over midnight it's broken into two parts. But for expected output the time...

Thanks, Drew for your reply. Did you bother to read my latest post? and did you bother to run the query that I provided?

IF yes, then my question was...

Using the preceding...

Afaik the set up is not clear.  This seems to produce the correct result

with
gaps_cte as (
    select *, case when crit<>lag(crit) over (order by...