The Daily WTF: Curious Perversions in Information Technology

aihtdikh · 2006-03-01 Reply Admin

:-/ oops
I missed the second page, with the 10 or so replies that already pointed this out.
My bad.

I'll just have to hope that nobody bothers reading the third page and sees me making a fool of myself
;-)
waves at the cameras

2006-03-02 Reply Admin

I do not see the WTF in this story at all. It's just another description of some knuckle head programmer. The industry is full of these people. Most have a chip on their shoulder and are quick to blame others rather than considering the possibility that the failure might be a result of something they have done. What? A programmer with a complex! What next, sales and marketing people that use buzzwords?

2006-03-02 Reply Admin

This whole wtf is kinda based on hearsay, isn't it? I mean, the dumb guy probably was dumb, but I can believe that the "top 5" comment meant something like "I didn't write any queries with 'top 5' in them", rather than "I don't know what 'top 5' means. And maybe the logging tools remark didn't mean that the logging tool was mangling his sql, only that it wrote its own sql. I doubt it--the guy is probably just a arrogant halfwit--but the beauty of the whole dailywtf concept for me is that when there is a code sample the dumbass-ness is incontrevertible.

2006-03-02 Reply Admin

Anonymous:
Perl allows general switchouts, it's one of the reasons it's easy to create obfuscated code.
s/this/that/;

s,this,that,;

s[this][that];

It also means if you're, say, switching forward slashes,

s{/}{\}; is the same as s/\//\\/;

I prefer using semicolons. ;)

s;\s+; ;g; s;^ ;; ;s; $;;;

paddy · 2006-03-02 Reply Admin

The big wtf I see......

.....how long does it take two programmers to track a simple failed DB query?
Seriously, how can you continue on for another hour or so when the problem is a basic syntax fault during an attempt to run a query?

Maybe its the sleeping pills kicking in, but wouldn't the DB log the error, even if some "on error resume next" nonesense was causing the app to skip over and make the tracking harder?

Omnifarious · 2006-03-02 Reply Admin

efox:
OK, not a rule of thumb but sometimes the thumbs rule. What I'm getting at is never to rule yourself out as the source of an error. The guy didn't just fail as a result of his ignorance but because of his inflated opinion of himself, that's the true WTF.

That I can definitely agree with. :-)

Omnifarious · 2006-03-02 Reply Admin

Anonymous:
Beep... wrong.
The wise programmer *always* checks his/her own code first and when they suspect a problem in other parts of the system, is always very careful about how they express this.

In my experience a good programmer eschews the whole "my code", "their code" thing and concentrates on statements like "I think the problem may be in this bit of [implicitly, 'our'] code here".

*chuckle* Yeah, probably a good bit of advice.

I tend to be that way on Open Source projects that I contribute to. I guess its because I'm really interested in being accepted as part of the group of people who work on that project. Where at most of the jobs I've had, I've not been that interested. Particularly since most commercial code I've seen has been highly icky, and if an Open Source project is icky, I don't contribute anything but bug reports.

But, it'd probably work out better for me if I showed a little more humility, even if I didn't feel that the respect had been earned.

2006-03-02 Reply Admin

Ahhhhh, the normal reaction of a computer science person:
"It can't be! It simply cannot be this way!"

(Happens to me every day ;) )

jwenting · 2006-03-02 Reply Admin

R.Flowers:

Martin:
Only been veiwing a few weeks and I get first?

It's not based on seniority.[;)]

What strikes me about the above story is how careful everyone is not to just laugh in the guys face. In most forums for IT pros that I've seen (even here, sometimes), IT geeks are not shy about pointing out a colleague's mistakes! [:D]

Unless the guy making the mistake is the fair haired boy of some bigshot manager and you might find yourself out of a job or relegated to deadend jobs if he doesn't like you.
Been there, done that.

jwenting · 2006-03-02 Reply Admin

aihtdikh:
frito:
In SQL SERVER (at least 2005) numbers can optionally be inside single quotes.

I have confirmed this because I am running SQL SERVER 2005. So something like this is valid:

SELECT * FROM users WHERE id>'0'

That's because MSSQL will implicitly convert between string and integral datatypes.
If it's between single quotes then it's a string, not a number.
Your point still stands, though - the number being in a string doesn't say anything about the datatype of the actual column.

Variant? [:'(]

2006-03-02 Reply Admin

Anonymous:

"he knew that it was definitely related to the
backend, far, far behind the depths of the database servers and into
the bowels of the organization's infrastructure"

I like that one. I'll see if I can use it one day.

http://developer.apple.com/qa/me/me05.html

brazzy · 2006-03-02 Reply Admin

Anonymous:
The wise programmer *always* checks his/her own code first and when they suspect a problem in other parts of the system, is always very careful about how they express this.
In my experience a good programmer eschews the whole "my code", "their code" thing and concentrates on statements like "I think the problem may be in this bit of [implicitly, 'our'] code here".

What's important is fixing the problem, not whose code it is. Placing blame is stupid, but so is going through contortions to avoid hurting people's feelings.

In MY experience, a good programmer most of all does not deal in "I think"s and "maybe"s, he adds logging or fires up a debugger and finds out what actually happens, at which point the code he has written and understands well interfaces with someone else's code and gets an unexpected reaction. He checks specs to see whether his input and the reaction are in fact allowed, and then either changes his code to deal with an unexpected but correct reaction, tells the other guy about his code's incorrect reaction to a correct input, or (all too often) sits down with the other guy and an analyst to tighten up the specs.

2006-03-02 Reply Admin

chaim79:
Digitalbath:
John Smallberries:
Anonymous:

The systems guy pointed out the problem right away: you shouldn't be using curly braces within your SQL queries like that; you need to use regular parenthesis.

I'll admit, I don't do much ASP.NET/SQL Server programming. What does using curly braces cause that using parenthesis does not?

I do a lot of SQL/ASP.Net and I'm not sure either. I'm pretty sure they don't mean "top 5".

Also, it looks like APPDOC_NUM would be the PK, so "top" wouldn't do anything anyway.

If APPDOC_NUM was the PK, wouldn't that query just return 5 results (assuming there was at least 5) that had the APPDOC_NUM of 88712? It looks like if you were doing that (for whatever reason) you would need an order by so you just didn't get some random 5 results returned...

If APPDOC_NUM was the PK then there would be only one record in that table with the APPDOC_NUM of 88712, the Query will only return one record ever and TOP 5 is a useless addition to the query. However if APPDOC_NUM Isn't the PK but a FK or a Non-Unique key then there would be a bunch of results and you would be pulling only the top 5, but that would be in whatever order they are stored in the Table itself, not random.

The curly braces/parenthesis bit is the actual problem, not the top 5 thing.

Erik of Ekedahl

I have seen visual query building tools that add things like Top 5 and Top 100 to queries without being asked...

Fred · 2006-03-02 Reply Admin

Martin:
R.Flowers:

What strikes me about the above story is how careful everyone is not to just laugh in the guys face. In most forums for IT pros that I've seen (even here, sometimes), IT geeks are not shy about pointing out a colleague's mistakes! [:D]

Shy about it; we have a big cowboy hat in the office that is cermoniously placed on the offenders head!

.. And we have a big, hairy spider ("the bug") that is placed on top of the monitor of the most recent offender... :-)

Arancaytar · 2006-03-02 Reply Admin

"TOP n" doesn't appear to work in MySQL...

In those database systems where it does work, however, how does it differ from "LIMIT 0,n"? It doesn't strike me as particularly more elegant or readable. In fact, since the limiting statement is obviously executed after sorting by the "ORDER BY" statement, wouldn't the logical place for it be after that statement - as LIMIT is, unlike TOP?

Or am I missing something?

Corwinoid · 2006-03-02 Reply Admin

Disgruntled DBA:
Anonymous:
R.Flowers:

Manni:

If you look at it long enough and cross your eyes, you see Jesus. Or a duck.

I see Ted Nugent.

Did Ted shoot Jesus or the duck?

Ted shot the duck. Dick Cheney would have winged Jesus.

You got that backwards. Dick got the duck... he was aiming for Jesus.

2006-03-02 Reply Admin

Definitely Ziggurat. I hope the prevailing winds come from top/right. And that the foundation is sturdiest on that end.

KenW · 2006-03-02 Reply Admin

Alex Papadimoulis:
Anonymous:
Anonymous:
Alex Papadimoulis:
SELECT TOP 5 RCRD_CD, OWNR_ID
FROM DLX_APPLICATION_DOCS
WHERE APPDOC_NUM = '88712'
Okay... APPDOC_NUM is a string value? Brillant!!
Having a "number" that's actually a string data type is amazingly common in databases, especially older ones.
In the world of data, a number (as account number, social security number, document number) should be stored in character field; this is not an old practice but a sound maintenance practice for a handful of reasons.

A good rule of thumb is, look at a datum (let's say, 5642157). In the context of the system, how would you say it aloud: Five-Six-Four-Two... -- or -- Five Million, Six Hundred and fourty two thousand ... ?

If it's the former, then it's a number in the sense that most people use them -- synonymous with "identifier". If it's the latter, then you should not be calling it a number. Instead, you should use a quantifier like total, sequence, amount, percent, count, etc.

Actually, Alex, a better rule of thumb is: If you're going to do math with it, it's a number; if you're not, it's a string.

Alex Papadimoulis · 2006-03-02 Reply Admin

paddy:
The big wtf I see......

.....how long does it take two programmers to track a simple failed DB query?
Seriously, how can you continue on for another hour or so when the problem is a basic syntax fault during an attempt to run a query?

Maybe its the sleeping pills kicking in, but wouldn't the DB log the error, even if some "on error resume next" nonesense was causing the app to skip over and make the tracking harder?

You have obviously never worked on large applications developed with the "Big Ball Of Mud" design pattern where logging was never a design thought. You will some day -- these apps require hoards of mainteance programmers just to keep running -- and when that day comes, hopefully you'll remember this question of yours and think "now I see; why did I take this job again?" That was my first reaction with such an experience

haveworld · 2006-03-02 Reply Admin

RobIII:
Pope:
gwenhwyfaer:
Anonymous:
Manni:
Anonymous:
Digitalbath:
chaim79:
Digitalbath:
John Smallberries:
Anonymous:

The systems guy pointed out the problem right away: you shouldn't be using curly braces within your SQL queries like that; you need to use regular parenthesis.

I'll admit, I don't do much ASP.NET/SQL Server programming. What does using curly braces cause that using parenthesis does not?

I do a lot of SQL/ASP.Net and I'm not sure either. I'm pretty sure they don't mean "top 5".

Also, it looks like APPDOC_NUM would be the PK, so "top" wouldn't do anything anyway.

If APPDOC_NUM was the PK, wouldn't that query just return 5 results (assuming there was at least 5) that had the APPDOC_NUM of 88712? It looks like if you were doing that (for whatever reason) you would need an order by so you just didn't get some random 5 results returned...

If APPDOC_NUM was the PK then there would be only one record in that table with the APPDOC_NUM of 88712, the Query will only return one record ever and TOP 5 is a useless addition to the query. However if APPDOC_NUM Isn't the PK but a FK or a Non-Unique key then there would be a bunch of results and you would be pulling only the top 5, but that would be in whatever order they are stored in the Table itself, not random.

The curly braces/parenthesis bit is the actual problem, not the top 5 thing.

Erik of Ekedahl

Oh, right...sorry. In my haste to make my first post ever, I didn't read the post close enough...I was thinking FK. Oops.

Pretty neat 3D pyramid we're building here.

If you look at it long enough and cross your eyes, you see Jesus. Or a duck.

It look more like a ziggurat to me.......

And we all know how difficult it is to give up ziggurats.

Well, it's because the companies that make them put all that Ningirsu in them.

preview?

Can we stop quoting? This is starting to hypnotize me... [:O]

I wonder what the size limit of a post is? 8000 characters maybe?

2006-03-02 Reply Admin

frito:
In SQL SERVER (at least 2005) numbers can optionally be inside single quotes.

I have confirmed this because I am running SQL SERVER 2005. So something like this is valid:

SELECT * FROM users WHERE id>'0'

Yes it's legal. Has been since SQL 6.5 (or longer) but for 2000 and before when the predicate datatype dos not match index datatype, the index is not used and the table scanned. Depending on the table size ... That would be an issue.

2006-03-02 Reply Admin

The WTF is basically that the "SQL developer" had never used "TOP n" in a SELECT clause before. Not such a good post Alex. You've posted better. A couple of points here:

This just seems to be a ASP.NET / MS SQL Server bashing exercise by the original author. Not fair considering its considerable usage in the enterprise. I'm not saying it isn't flawless, but it certainly isn't awful.
The developer saw the *top five* query running and he did not recognise it as his. It should be pointed out that a common mistake when using SQL Profiler is to run it (by default) against all database queries on a server. Hence, the query shown might have come from another database, and not his.
Why would his application work in his development environment (with the same SQL code) and then not in the live environment? To me, that shouts "configuration problem".
I'm guessing his SQL syntactical mistake was something along the lines of "select count{myID} FROM myTable", or "select myField FROM myTable where myField={10}", but again, refer to previous statement - bugs in SQL code are bugs in SQL code where ever you run them.

It would have been useful to post the actual offending query, so maybe we could have an idea of what the developer's error was. If the original author was able to supply the *top five* query in full (I'm guessing this is a real table name - Googled "DLX_APPLICATION_DOCS" and got nothing), then why could they not supply the actual bug??

Sorry, but this post is lame. Normally they are much better and complete to boot. It took me quite a while to realise that the single, weak WTF was that a self proclaimed SQL programmer hadn't seen a "TOP n" query before.

N.B. Why *still* does the CAPTCHA never work first time round?

2006-03-02 Reply Admin

I've seen people that didn't know what "Top N" meant. I came across this recently where someone selected latest N news headlines to appear on the front page of a small portal type website, and then had variables counters for each one (anyone say re-use?). This was in classic ASP, he created a recordset, set the recordset pagesize to 20, then looped until "VariableCounter[N]" equaled the recordset pagesize... which had nothing to do with what he was actually trying to do.

2006-03-02 Reply Admin

Anonymous:
The WTF is basically that the "SQL developer" had never used "TOP n" in a SELECT clause before. Not such a good post Alex. You've posted better. A couple of points here:

This just seems to be a ASP.NET / MS SQL Server bashing exercise by the original author. Not fair considering its considerable usage in the enterprise. I'm not saying it isn't flawless, but it certainly isn't awful.
The developer saw the *top five* query running and he did not recognise it as his. It should be pointed out that a common mistake when using SQL Profiler is to run it (by default) against all database queries on a server. Hence, the query shown might have come from another database, and not his.
Why would his application work in his development environment (with the same SQL code) and then not in the live environment? To me, that shouts "configuration problem".
I'm guessing his SQL syntactical mistake was something along the lines of "select count{myID} FROM myTable", or "select myField FROM myTable where myField={10}", but again, refer to previous statement - bugs in SQL code are bugs in SQL code where ever you run them.

It would have been useful to post the actual offending query, so maybe we could have an idea of what the developer's error was. If the original author was able to supply the *top five* query in full (I'm guessing this is a real table name - Googled "DLX_APPLICATION_DOCS" and got nothing), then why could they not supply the actual bug??

Sorry, but this post is lame. Normally they are much better and complete to boot. It took me quite a while to realise that the single, weak WTF was that a self proclaimed SQL programmer hadn't seen a "TOP n" query before.

N.B. Why *still* does the CAPTCHA never work first time round?

The captcha times out. If you write a big post, it's gonna time-out on you.

Got that?

Zlodo · 2006-03-02 Reply Admin

Anonymous:
So...the real WTF is that he was using curly braces in a SQL Statement. The rest of the story leading up to that really has no bearing on anything. Except why WAS he selecting Top 5? Hard to guess without the context. Probably just showing 5 records per screen, I assume.

The WTF is that he was a complete all-knowing, "can't possibly be ME who made an error" asshat.
I worked with people like this.
There was that one guy who updated his code from sourcesafe, and it started crashing because he got a code change that I just made. He immediately went "hey, it's crashing, and it cannot possibly be me since I didn't change anything and it worked just before"

It was a multiplayer game with several mini games, and I just had put in a change to setup IA players in complement of human players.

So, a little irritated but knowing that having a look at it myself would probably be the fastest way of resolving the problem, I went over to his pc and started looking. He immediately left to smoke a cig.
10 seconds later, I had found out that he was simply calling a method on the joypad class to test a button, and since he didn't check that the joypad pointer was null, it crashed.
When he came back and asked me "so, did you fix it ?", I managed to stay calm and just pointed out to him what the problem (with HIS fucking code) was. He didn't even say as much as "oops, sorry".

Moral of the story: fucking spend at least 30 godamn seconds to do a minimal amount of research before complaining about a problem.

2006-03-02 Reply Admin

Anonymous:
Not such a good post Alex. You've posted better.

Sorry, but this post is lame. Normally they are much better and complete to boot. It took me quite a while to realise that the single, weak WTF was that a self proclaimed SQL programmer hadn't seen a "TOP n" query before.

Alex thanks you for your helpful criticism!

No sir, nothing funny at all about a junior programmer parading his ego in front of several co-workers.

2006-03-02 Reply Admin

Anonymous:
Anonymous:
Not such a good post Alex. You've posted better.

Sorry, but this post is lame. Normally they are much better and complete to boot. It took me quite a while to realise that the single, weak WTF was that a self proclaimed SQL programmer hadn't seen a "TOP n" query before.

Alex thanks you for your helpful criticism!

No sir, nothing funny at all about a junior programmer parading his ego in front of several co-workers.

I was just stating my opinion. I do genuinely think we see better WTF examples here every day. I just didn't think this particular one was a well constructed WTF that's all. It's just a bad ego.

Also thanks to CaptchaMan. I didn't realise the CAPTCHA timed out. I've now got it!

2006-03-02 Reply Admin

Anonymous:
Anonymous:
Not such a good post Alex. You've posted better.

Sorry, but this post is lame. Normally they are much better and complete to boot. It took me quite a while to realise that the single, weak WTF was that a self proclaimed SQL programmer hadn't seen a "TOP n" query before.

Alex thanks you for your helpful criticism!

No sir, nothing funny at all about a junior programmer parading his ego in front of several co-workers.

That's worth another ziggurat. And pardon my french, "TOP n" sucks, it's just Another Way To Do What We Usually Do The Other Way That Everybody Knows.

devdas · 2006-03-02 Reply Admin

I think the point is he was claiming to be some kind of hardcore, full-on, seen-it-all experienced legend in SQL and ASP.Net when, in fact, not knowing "TOP" is a valid (and useful) SQL keyword means he is a total n00b.

TOP is MSSQL/Access specific (or maybe transactSQL)..

You would be able to achieve the same results with an order by and a limt clause.

2006-03-02 Reply Admin

Here, here. There's no way I would store a credit card number as an integer type inside a database.

Omnifarious · 2006-03-02 Reply Admin

brazzy:
What's important is fixing the problem, not whose code it is. Placing blame is stupid, but so is going through contortions to avoid hurting people's feelings.

In MY experience, a good programmer most of all does not deal in "I think"s and "maybe"s, he adds logging or fires up a debugger and finds out what actually happens, at which point the code he has written and understands well interfaces with someone else's code and gets an unexpected reaction. He checks specs to see whether his input and the reaction are in fact allowed, and then either changes his code to deal with an unexpected but correct reaction, tells the other guy about his code's incorrect reaction to a correct input, or (all too often) sits down with the other guy and an analyst to tighten up the specs.

I want to work where you work. You seem to have well documented interfaces and stuff there. I sometimes dream of working in a place like that.

Sometimes its clear the code is doing the wrong thing. Sometimes its not so clear and things are well enough documented (if they're documented at all) that you can figure out what the right thing is supposed to be. So then you go get to hash it out with someone and figure it out as you go.

ukemigrant · 2006-03-03 Reply Admin

John Smallberries:

I do a lot of SQL/ASP.Net and I'm not sure either. I'm pretty sure they don't mean "top 5".

Also, it looks like APPDOC_NUM would be the PK, so "top" wouldn't do anything anyway.

Um, yes it will It will return the top 5 matching records (if any) just like it says. By 'a lot' I'm guessing you meant to say 'I don't do a lot of SQL'?

Gene Wirchenko · 2006-03-03 Reply Admin

ukemigrant:
John Smallberries:
I do a lot of SQL/ASP.Net and I'm not sure either. I'm pretty sure they don't mean "top 5".
Also, it looks like APPDOC_NUM would be the PK, so "top" wouldn't do anything anyway.

Um, yes it will It will return the top 5 matching records (if any) just like it says. By 'a lot' I'm guessing you meant to say 'I don't do a lot of SQL'?

If his assumption that APPDOC_NUM is a PK is correct, only one row would be returned since PKs are unique by definition.

Sincerely,

Gene Wirchenko

hank miller · 2006-03-03 Reply Admin

Zlodo:
Moral of the story: fucking spend at least 30 godamn seconds to do a minimal amount of research before complaining about a problem.

More like 30 minutes. And even then don't be too loud about who you blame. I've spent too many days working on a problem that seemed to be in someone else's code (a device, so I couldn't ask the original programmer for help), only to find the bug in my code.

Moral: when something seems like a bug in someone else's code, approach them as in you need an extra set of eyes because you cannot find the problem and it is related to their code. That way if it is their problem (which it is once in a while) they can admit it - approach it wrong and they get defensive about their code lest their boss find out and give them a bad review. If it is your fault you can thank them for the help because you were overlooking something obvious...

2006-03-03 Reply Admin

Manni:
Fist:
Manni:
Whiskey Tango Foxtrot:

The systems guy pointed out the problem right away: you shouldn't be using curly braces within your SQL queries like that; you need to use regular parenthesis.

I'll admit, I don't do much ASP.NET/SQL Server programming. What does using curly braces cause that using parenthesis does not?

I don't think you do much programming at all if you think you can switch out parenthesis for another delimiter and expect the code to work the same.

foo(a, b) <> foo[a, b]

No... What we don't get is why using braces would create a "TOP 5" ah-ha that using parenthesis wouldn't. Also, I have used a language where you can use parens, brackets, braces, etc, as you choose, to help with visual matching of delimeters (they all function the same). So yes, sometimes you can expect such things to work, especially if you have done enough programming to be familiar with such languages.

A surefire way to bring out the naysayers is to declare that something isn't possible. Yes I'm sure such a language exists that gives you the power to pick and choose, but is it one of the major languages like C#, C++, VB6/VB.Net/ASP/ASP.Net, SQL, Perl, PHP?

Major as in most used. Not as in most usefull. Popular programming languages are like popular music. Its not too shabby, its not great either, just mainstrain. And your list only contains imperative programming languages, except sql. So ehm, regardless of the marketing buzz, their all the sort of the same semantic features, different syntax. So presenting it as _the_ list of major programming languages: its the list of the major used imperative programming languages.

Next time try focusing on the actual article rather than proving other people's comments wrong. You would have noticed that there was no correlation between the "top 5" statement and the curly braces. The idiot programmer tried to blame the problem on a SQL statement he'd never seen before, but instead the problem was elsewhere in his code where he tried to use curly braces incorrectly.

2006-03-03 Reply Admin

Anonymous:
N.B. Why *still* does the CAPTCHA never work first time round?

Haven't you figured that out ? Its problely saving only the name of the last CAPTCHA image that is has shown. So the more time inbetween page load and 'post', the bigger the chance some one else has opened the 'post a new message' box at some point during that time.

Its problely a shared session variable that shouldn't be shared.

2006-03-04 Reply Admin

Oh, that's scary! No arachnofobics at your company? I am.. and I live in Norway ;) Guess if I ever end up working at your company then I can not make any single error, otherwise I'll be doomed since I get paralyzed of 8-legged creeps and won't be able to do any work. And not being able to do any work sure will make me get that creep at my monitor for ages.

John Smallberries · 2006-03-06 Reply Admin

ukemigrant:
John Smallberries:

I do a lot of SQL/ASP.Net and I'm not sure either. I'm pretty sure they don't mean "top 5".

Also, it looks like APPDOC_NUM would be the PK, so "top" wouldn't do anything anyway.

Um, yes it will It will return the top 5 matching records (if any) just like it says. By 'a lot' I'm guessing you meant to say 'I don't do a lot of SQL'?

And how many matching records will a a query using the PK return?
C'mon...you can do this...right! At most 1. Now, what is the top 5 of 1 record?

Idiot. I'm guessing you meant to say 'I don't do a lot of thinking'?

vasco · 2006-03-15 Reply Admin

thesis bit is the actual problem, not the top 5 thing.

Erik of Ekedahl

What kind of programmer don't test it's code at least to the level of the syntax error ?

Sql statements can became very complex and an obscure error can luck in.
But the fact that a syntax error can stay unfixed on a production server shows much about the sw development cycle on this company, wich is much worst than the fact that the programmer don't want to accept he made a mistake.

Obviously The Problem

Leave a comment on “Obviously The Problem”