SQL GROUP BY Clause

[lex]

This node falls below the community's threshold of quality. You may see it by logging in.

[rnahi]

You are desperately OT in this forum.

Aside from that, let me analyze your question.

Does the order of columns in a SQL Group By clause matter?

Why would anyone ask somathing like that? If you were doing something practical with SQL, you could just throw two queries, changing the order of columns in the GROUP BY, and seeing for yourself if they differ or not (I am not going to give you the answer indirectly :-P).

So youy were not doing anything practical. Then why would you be curious about this? The only plausible answer is because somebody asked you. Hence, it's homework.

One more indication of homework is the "discuss" clause. You don't come to a forum and say "discuss" for something you don't know. You can come here, post something in Meditations, say what youy think about that, and then ask for discussion. But discussing some unknown thing that your teacher asks you to find out is hardly something that can tickle our willingness to help.

[planetscape]

P.S. This is not a school essay that I want you to write for me.

Say, that is convincing... ;-) Please see this.

[rinceWind]

lex: what do YOU think? What's your application? Whether is does matter or not depends on what you are wanting to achieve.

Also, you have not said which database platform you are using, which would significantly affect the answer.

P.S. This is not a school essay that I want you to write for me.

What are you looking for then?

[jZed]

As you can see from people's (lack of) response, if you want discussion, you might be better off giving your own thoughts to kick the discussion into gear. That said, my guess is that no, the order should not matter. If you group SUM(sales) by "year, region", you are asking to see totals for each year and totals for each region and if you group them by "region, year", you are still asking for totals for each region and for each year (in other words, the same two totals). Is that what you'd expect? What about it merits discussion?

[chrism01]

I agree it sounds a lot like homework....
however, .. yes it matters.
The reason is SQL presents the results grouped (nested) in the order you asked, so SALES grouped by yr, mth would group sales by yr, then by mth within that yr. Evidently the other way round would make no sense unless you were looking for seasonal patterns...
Because it's easier to code group by internally by first sorting the data as per request, group by normally implies order by the same values will occur in the output.

[xorl]

Very much not Perl. A better place to ask this question is http://forums.devshed.com/

I don't actually know the answer to your question. I've only had to write queries with only one field in the group by clause.

If you're using MySQL this may help http://dev.mysql.com/doc/mysql/en/group-by-optimization.html