Return to Answer

largely arbitrary wording tweaks of my own answer, submitted yesterday, to clarify my purpose for answering this older question which, if we're honest, may really only be interesting as an exercise in date crunching.

Source Link

edited Feb 11, 2016 at 15:36

rob c

NewRelatively green Firebird usern00b here w/a background infair PostgreSQL background, somewhat limited toalthough specialized GIS/spatial work focused mostly to DB admindesign and the sourcing/cleaning/making spatial multi-source data sets (usually of dubious quality) and then querying for map-able results in answer to specific location riddles. TookToolbox is light, therefore, on Admin/Optimization/3rd Party Application Setup etc. You may (err, will) see some Firebird specific questions from me in the coming weeks. Thanks in advance! Took a stab at this one (feedback welcome!)slightly older Firebird question to get thea feel for the new-to-me flavor of date handling and it tookconcatenation. Took more tweaking & testing than expected. May & my dabbling may not be the most efficient/elegant solution but, assuming the existence of column join_date in table employees, something like this will work:

Holler if I goofed it or over-complicated theif I employed bad/overwrought logic. Hope it's helpful to someone at some point! Keep a lookout for new Firebird questions (might already have locked in struggle with a JDBC connection string that just will not lose).

Thanks and hello again! --Rob

I never had a plan-- Anonyman

Source Link

created Feb 10, 2016 at 18:02

rob c

New Firebird user here w/a background in PostgreSQL, somewhat limited to GIS/spatial DB admin. Took a stab at this one (feedback welcome!) to get the feel for date handling and it took more tweaking & testing than expected. May not be the most efficient/elegant solution but, assuming the existence of column join_date in table employees, something like this will work:

SELECT Y ||' Years, '|| M ||' Months, and '|| D ||' Days'
FROM
(
SELECT
CASE WHEN datediff(year, join_date, current_date) <> datediff(day, join_date, current_date)/365
THEN datediff(year, join_date, current_date)-1
ELSE datediff(year, join_date, current_date)
END as Y,
CASE WHEN datediff(year, join_date, current_date) <> datediff(day, join_date, current_date)/365
AND datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)<0
THEN datediff(month, dateadd(year, datediff(year, join_date, current_date)-1, join_date), current_date)-1
WHEN datediff(year, join_date, current_date) <> datediff(day, join_date, current_date)/365
AND datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)>=0
THEN datediff(month, dateadd(year, datediff(year, join_date, current_date)-1, join_date), current_date)
WHEN datediff(year, join_date, current_date) = datediff(day, join_date, current_date)/365
AND datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)<0
THEN datediff(month, dateadd(year, datediff(year, join_date, current_date), join_date), current_date)-1
ELSE datediff(month, dateadd(year, datediff(year, join_date, current_date), join_date), current_date)
END as M,
CASE WHEN datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)<0
THEN datediff(day, dateadd(month, datediff(month, join_date, current_date)-1, join_date), current_date)
ELSE datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)
END as D
FROM employees
)

The subquery turned conditional on me when it became clear that: a) calculations on year & month don't consider the date as a whole; and b) date subtraction will happily return negative values regardless of the order of events.

Year/Month/Day Example: datediff() returns a value of 1 when retrieving the number of days, months, or years between 31.12.2015 and 1.1.2016.

Negative Value Example: about 69% of my test-date values are less than zero when I calculate the number of days less the number of days contained by whole months using:

SELECT datediff(day, dateadd(month, datediff(month, date_column, current_date), date_column), current_date) as D FROM date_table

We can streamline the subquery if substituting 365 days for one year is acceptable. The trick works with my test data but may trigger pitfalls with a broader range of dates. The simplified query looks like:

SELECT Y ||' Years, '|| M ||' Months, and '|| D ||' Days'
FROM
(
SELECT
datediff(day, join_date, current_date)/365 as Y,
CASE WHEN datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)<0
THEN datediff(month, dateadd(year, datediff(day, join_date, current_date)/365, join_date), current_date)-1
ELSE datediff(month, dateadd(year, datediff(day, join_date, current_date)/365, join_date), current_date)
END as M,
CASE WHEN datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)<0
THEN datediff(day, dateadd(month, datediff(month, join_date, current_date)-1, join_date), current_date)
ELSE datediff(day, dateadd(month, datediff(month, join_date, current_date), join_date), current_date)
END as D
FROM employees
)

Holler if I goofed it or over-complicated the logic. Hope it's helpful! --Rob