Write maintainable T-SQL with clear names, explicit results, focused procedures, and comments that explain the business reason behind the code.
Grouping Dates Into Buckets With DATE_BUCKET
DATE_BUCKET groups timestamps into hourly, weekly, and custom intervals, with an origin that determines where every boundary lands.
COUNT_BIG: Count Rows and Non-NULL Values as bigint
COUNT_BIG helps me name the counting rule before reading the result. Compare rows, supplied values and distinct readings. Then check empty results and the bigint return type.
RIGHT: Zero Padding Can Remove Significant Digits
RIGHT can make zero padding look correct while quietly removing significant digits. I validate the allowed numeric range before creating a fixed-width label. The suffix expression alone cannot tell whether the input fits.
Testing a Cumulative Update Before Production
Testing a Cumulative Update means restoring real data, replaying key jobs, comparing plans and timings, and signing off before production.
Funny SQL Server Error Messages and What They Really Mean
Some real SQL Server error messages read like jokes. Here are my favorite funny SQL Server error messages and what each one is really telling you.
RADIANS: Cast Integer Degrees Before Conversion
RADIANS changes degrees into radians, but its input type affects the result. The example contrasts integer and float arguments while keeping the original degree values.