Question

This problem has been bugging me a lot since the past few days.
Here’s what I thought of:
A super big retail store company has stores in many different cities, each store
was opened on a given day.
The table tabulating the data would have (atleast) two columns,
(1) city name (2) date opened
Output desirable: multiple lists, each list
containing all the stores of a given city ordered
chronologically (by date opened) of 10 cities ordered
alphabetically.
To get around this I was thinking of:
Fetch #1: Alphabetically fetching 10 cities from the table
Fetch #2 (multiple fetches in a loop): Iterating each fetched city, and obtaining
all of its respective stores, making the order by date opened.

The above method which I’ve come up with does the job, however I was wondering if there
is a more concise way of doing it, which involves the looping though the stores of the cities while making fetch #1 itself, and thereby not increasing the time complexity by using an external loop and performing many sql executions instead of just one.
Would that be possible? one sql statement?

Answer 1

SELECT * FROM my_table NATURAL JOIN (
  SELECT DISTINCT city FROM my_table ORDER BY city LIMIT 10
) t ORDER BY city, opened

Split into separate lists when looping over the resultset within your application: just inspect city to see if it has changed since the last iteration. In pseudocode:

for each record in resultset:
  if current city is not equal to last city then start new list
  output current record
  set last city to current city