Conditional Order By

Question

Conditional Order By

Forum Etiquette: How to post Reporting Services problems

[/url]

Forum Etiquette: How to post data/code on a forum to get the best help - by Jeff Moden
[/url]

How to Post Performance Problems - by Gail Shaw
[/url]

cengland0 SSCertifiable Points: 6102 More actions · Answer 1

mister.magoo (11/16/2010)
Thanks to those of you who appreciate this question.
To those who think it had unnecessary "tricks" in it, I did try several variants of this question and they all ended up being too easy.
The "tricks" were deliberate and designed to check whether you would notice the implicit conversion and whether you even knew it would happen.
If you don't like the question, I am sorry - for those that like to be challenged to check every small detail I think it provides a sufficient level of complexity to give you a good challenge!
Have a nice day!

I got this one right because I'm always looking for a trick. That's a general problem with the QOTD. Too many of the questions have some trick in it so when there is a straight-forward question without a trick, I find myself looking for one anyway.

Rose Bud Hall of Fame Points: 3147 More actions · Answer 2

I learned a lot from this QOTD. Count me as someone else who didn't know there was such a thing as conditional order by.

I can see the utility of this and I can see that an implicit cast could cause problems, even an error-less unintentional sort as Tony++ mentions.

Thanks to depak.a for providing a solution to the implicit cast problem.

The swapping of string1/string2 was probably an unnecessary distraction but debugging with "distractions" is a part of life. I readily grant license to QOTD authors to add tricks if they prefer that type of question.

b0b555 Mr or Mrs. 500 Points: 591 More actions · Answer 3

b0b555

Mr or Mrs. 500

Points: 591

November 17, 2010 at 5:14 pm

#1252221

Excellent question.

Carlo Romagnano SSC-Insane Points: 22672 More actions · Answer 4

Hugo Kornelis (11/17/2010)Without the reversal, the question would have been a solid test of whether people know about the dangers of using CASE in ORDER BY, and thus an excellent question. With the reversal, it's still a good question - good, not excellent.

The QoTD was not about dangers of using CASE in ORDER BY, but the problem is in the convertion in the CASE clause.

SQLWinther SSCertifiable Points: 6294 More actions · Answer 5

Richard Warr (11/16/2010)
I thought it was an interesting question as well - currently more people have it wrong than right which shows that we can still be surprised at the "behind the scenes" antics of SQL Server.
Please don't describe questions as "nonsense", people take taime to set them and always have good intentions when they do so. It doesn't help your case if you can't spell "nonsense" either.

I agree with you Richard! I admire the people who dare to post questions on this site.

/Håkan Winther
MCITP:Database Developer 2008
MCTS: SQL Server 2008, Implementation and Maintenance
MCSE: Data Platform

Hugo Kornelis SSC Guru Points: 64790 More actions · Answer 6

Carlo Romagnano (11/18/2010)
Hugo Kornelis (11/17/2010)Without the reversal, the question would have been a solid test of whether people know about the dangers of using CASE in ORDER BY, and thus an excellent question. With the reversal, it's still a good question - good, not excellent.
The QoTD was not about dangers of using CASE in ORDER BY, but the problem is in the convertion in the CASE clause.

Hi Carlo,

The implicit converts in the CASE are the root cause of the danger I referred to. And while the same thing may happen in a CASE anywhere, I have thus far only seen this in ORDER BY clauses. So I guess that you and I are saying the same thing, jusst using different words to describe it.

Hugo Kornelis, SQL Server/Data Platform MVP (2006-2016)
Visit my SQL Server blog: https://sqlserverfast.com/blog/
SQL Server Execution Plan Reference: https://sqlserverfast.com/epr/

mister.magoo SSC-Forever Points: 47174 More actions · Answer 7

Hugo Kornelis (11/18/2010)
Carlo Romagnano (11/18/2010)
Hugo Kornelis (11/17/2010)Without the reversal, the question would have been a solid test of whether people know about the dangers of using CASE in ORDER BY, and thus an excellent question. With the reversal, it's still a good question - good, not excellent.
The QoTD was not about dangers of using CASE in ORDER BY, but the problem is in the convertion in the CASE clause.
Hi Carlo,
The implicit converts in the CASE are the root cause of the danger I referred to. And while the same thing may happen in a CASE anywhere, I have thus far only seen this in ORDER BY clauses. So I guess that you and I are saying the same thing, jusst using different words to describe it.

The question was inspired by a forum post where someone was doing exactly that - using a CASE in the order by which would fail on implicit data type conversions. The aim was to highlight the danger of mixing data types in this way - whether it is in an ORDER by or not - but as Hugo says, it mostly happens in ORDER BY clauses.

Also, I just like it when I find something that people may not know and want to share :satisfied:

MM

select geometry::STGeomFromWKB(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

Zagyg SSCrazy Points: 2091 More actions · Answer 8

Just adding my vote to the "good question" list. I'm well aware of data type conversions, and spotted the "trick", and still got it wrong through a moment of fuzzy thinking. Well done on re-highlighting the issue.

SqlOnMyMind SSCertifiable Points: 5062 More actions · Answer 9

I think the "trap" is that the code "looks" like it is writing dynamic sql, but it really isn't.

TomThomson SSC Guru Points: 104773 More actions · Answer 10

Although I'm irritated that the flipping of 1 and 2 was very noticeable and having noticed it I decided that was "the trick" and thought no more (thus getting the wrong answer) I still thionk it's a good question. It's very important when desk-checking someone's code or trying to debug something to do it carefully and follow through on the whole thing, not stop thinking when you spot something odd, so questions like this are valuable - they remind us that leaping to conclusions without considering all the code is a stupid thing to do.

Tom

Toreador SSChampion Points: 11366 More actions · Answer 11

Excellent question - which I got right due to having been buitten by exactly this problem in the past, and it took ages to track down the cause 🙂

SQLRNNR SSC Guru Points: 281334 More actions · Answer 12

Thanks for the question.

Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events

Gopi S Ten Centuries Points: 1091 More actions · Answer 13

Thanks for posting this.

I am new to SQL server programming. Though both question and answer was clear i would like to know the reason or why SQL server is performing CONVERT_IMPLICIT(datetime,[Union1007],0).(ie why String1, String2 is converted to datetime).

Thanks

mister.magoo SSC-Forever Points: 47174 More actions · Answer 14

Gopinath Srirangan (11/25/2010)
Thanks for posting this.
I am new to SQL server programming. Though both question and answer was clear i would like to know the reason or why SQL server is performing CONVERT_IMPLICIT(datetime,[Union1007],0).(ie why String1, String2 is converted to datetime).
Thanks

Please see this post for an explanation : http://www.sqlservercentral.com/Forums/FindPost1021495.aspx

MM

select geometry::STGeomFromWKB(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