我的SQL中存在日期和日期时间比较问题

Question

Been testing this over and over, and it fails at the date comparison.( item.id_type seems to work fine).

request.date has the datatype DATETIME .

SELECT request.id, request.date, request.total_price,
    item.cod_GERFIP,item.price,item.name,request_item.quantity,
    section.name AS section,user.firstname,user.lastname   

FROM ((((`request_item` 
INNER JOIN `request` ON  request_item.id_request = request.id)
INNER JOIN `item` ON request_item.id_item = item.cod_GERFIP)
INNER JOIN `user` ON request.id_user = user.id) 
INNER JOIN `section` ON user.section = section.id) 

WHERE request.date >= '2019-01-09' AND request.date <= '2019-01-10'  
    AND item.id_type = '1' 
ORDER BY request.date DESC

Answer 1

Just a guess, because you haven't explained what or how the query fails, but to my eyes this condition does not look correct:

 request.date <= '2019-01-10' 

It's a common mistake to expect a condition like this, when used a part of a range, to include all records where the date part of a datetime field is 2019-01-10 . That is, if we have an example value in the database of 1 PM on the same day ( 2019-01-10 13:00:00 ), the expectation is this value will narrow to match the 2019-01-10 literal in the query, the two values will be equal, and so it will meet the condition.

It does work this way.

Instead, the 2019-01-10 literal in the query is widened to a full DateTime, that looks more like this: 2019-01-10 00:00:00.000 . Now the 1 PM value from the table is compared with this full date time, and it fails the condition.

It's much more common for a date range to compare using an exclusive upper bound set for one day in the future:

request.date < '2019-01-11'

Alternatively, you may be tempted to do this:

request.date <= '2019-01-10 23:59:59.999'

It will even work most of the time. Just be warned that in the (rare) case of leap seconds, you can still end up with incorrect results that way.

You may also be tempted to do something like this:

convert(date,request.date) <= '2019-01-10'

This works, but it's not recommended because it prevents the use of any index you might have on the request.date field, and that cuts at the core of database performance.

---

Or maybe the problem is even simpler. With the start of range at 2019-01-09 , maybe you wanted to get the records for exactly one day, and are surprised to see a few values from midnight on 2010-01-10 . Again, the solution is you want an exclusive boundary at the top of the range:

request.date < '2019-01-10'

As a complete side note to the question, I'm a very sad the SQL BETWEEN operator is inclusive at both end of the range. This may make sense for numeric or string data, but for date values defining an exclusive upper bound for the BETWEEN operator would have made much more sense.