The Daily WTF: Curious Perversions in Information Technology

Nutster · 2014-04-17 Reply Admin

Private Sub ImageList_DataBound(ByVal sender As Object, ByVal e As System.EventArgs) Handles ImageList.DataBound
    Try

    Catch ex As Exception
        If TypeOf ex Is ArgumentOutOfRangeException Then
            Throw New Exception("item not found in the list...")
        End If
    End Try
End Sub

Did the original code have an empty Try block? Well, that would be safe, because nothing would be executed. Personally, I would build the Catch block as:

Catch ex as ArgumentOutOfRangeException
    Throw New ApplicationException("Item is not found in the list.", ex)
Catch ex as Exception
    ' Throw it away, or re-throw and let caller take care of it, or just not catch it in the first place and let the caller take care of it?
End Try

2014-04-17 Reply Admin

Ran the following code against today's article, glad I put the try catch in here because it could have crashed without it!

Try MakeFunnyArticle(joke,humor) Catch Ex as Exception Msgbox("UNFUNNY ARTICLE!") End Try

2014-04-17 Reply Admin

The Fury:
The count method I can possibly understand unlike the others. It is possible that it used to do more, i.e. check for some condition on the document before incrementing the count.
At some point this requirement went away, but the developer didn't bother to do a proper job of refactoring.

function countDocuments() {
    var count = 0;

    for ( var i = 0; i < user.documents.length; i++) {
        if (i % 3 == 0) {
            print "Fizz";
        } else if (i % 5 == 0) {
            print "Buzz";
        } else {
            count++;
        }
    }

    return count;
}

The business logic was temporarily removed because it was not handling properly the FizzBuzz case

da Doctah · 2014-04-17 Reply Admin

String timeStampLength = "                          ";
int lengthOfTimeStamp = timeStampLength.length();

This is the code equivalent of "I once caught a TimeStamp this big!"

2014-04-17 Reply Admin

Sure. That's self-documenting code. I mean:

dateLength="YYYY-MM-DD".length()

A reader knows exactly what the length means. While:

dateLength=10

Who knows?

I probably would have written:

dateLength=10
// date string is YYYY-MM-DD

but arguably that's inferior, because a later programmer could come along and change the length of the string -- maybe add the time of day or add an AD/BC indicator or whatever -- and change the assignment statement but not change the comment. So now it says

dateLength=14
// date string is YYYY-MM-DD

Yeah, that would be dumb, but programmers do it all the time.

But of course taking the length of a string of spaces tells me less than dateLength=10. Now I have to count the spaces. What did that gain?

2014-04-17 Reply Admin

Zacrath:
RFox:
Because this method was decommissioned and they weren't quite sure they got all the calls to it yanked out of all the client code...so they made sure any code that still called it would fail dramatically.
Captcha: plaga - plaga of bad code.

public DataModel getEditionModel() { if ( true ) throw new IllegalArgumentException( "You shouldn't be here" ); return editionModel; }
And, of course, using @Deprecated won't work because everyone ignores warnings.

Or this code was written before the @deprecated tag was added to Java. Or, let's be fair, maybe the programmer isn't familiar with @deprecated.

RE just delete the function and get the compiler errors: But what if this function is in a library that might be called by many programs? The programmer wouldn't necessarily know all the programs that use this library. So if he just deleted the function, then six months later somebody recompiles and gets this mysterious not-found error, and he has to figure it out. If that's the case, sure, it would be better to put in a comment explaining why the function is obsolete, but I don't think this solution is totally idiotic. Not the best thing to do, but not totally idiotic.

2014-04-17 Reply Admin

 for ( var i = 0; i < user.documents.length; i++) {

What gets me about this one is that he used the length member right there. I understand if someone writes his own function to do something even though there's a built-in function in the standard library. So okay, he didn't know it was there. Ignorant, but not stupid. But this ...

Ok, maybe everyone else is saying, Duh, that's why it's on thedailywtf and not on "the daily oh by the way did you know there's a better way to do that".

2014-04-17 Reply Admin

Jay:
 for ( var i = 0; i < user.documents.length; i++) {
What gets me about this one is that he used the length member right there. I understand if someone writes his own function to do something even though there's a built-in function in the standard library. So okay, he didn't know it was there. Ignorant, but not stupid. But this ...
Ok, maybe everyone else is saying, Duh, that's why it's on thedailywtf and not on "the daily oh by the way did you know there's a better way to do that".

This is an attempt to get the most out of the expensive machine. Please note that with "count = user.documents.length" the length would be evaluated only once. But here, it's evaluated user.documents.length times and this with only a minimal overhead in code. This is a drastic improvement in terms of productivity!

2014-04-17 Reply Admin

Not all langauges have compilers/statick analysis tools. It may be that the class has to meet requirements of some interface, which are actually too restrictive, and the programmer is 99% sure, that a given function will not be called, but still prefers to have a nice stacktrace in case it happen.

As to the string of spaces, perhaps it is a buffer used to compose the characters of the date one by one, in which case initializing it with spaces makes some sense.

Using if(true) is either a way to make compiler happy (which in turn indicates existence of some static analysis tools...) or generated by some makro preprocessor which replaces some placeholders with boolean literals, or just a way to make it obvious to the reader that this line is "special".

Counting using loops is very common among people who just started to learn programming and try to encode what they would do if they were tasked with the same problem the computer is. Some of them are not smart enough to realize that the "for mantra" actually contains subexpression which already is the answer to their problem. Also, there might be people for which corner cases of empty array seem to be "better" handled this way. But, the most probable explanation is that the loop used to contain some more conditions/rules.

2014-04-17 Reply Admin

*) An attempt to explain the for-loop construct: With AI on the rise, there is less and less reason to treat structures of silicon other than the human work force. So the developer imagined him/herself rather a manager of the silicon work force. This is not simple code, it's the epitome of wise management as lived for decades!

2014-04-17 Reply Admin

Jay:
 for ( var i = 0; i < user.documents.length; i++) {
What gets me about this one is that he used the length member right there. I understand if someone writes his own function to do something even though there's a built-in function in the standard library. So okay, he didn't know it was there. Ignorant, but not stupid. But this ...

Well, to be fair , he tried this first:

for ( var i = 0; i < countDocuments(); i++)

and when he realized that wouldn't work, consulted the documentation and realized he could replace countDocuments() in his code with user.documents.length and Bob's-yer-uncle.

Loren Pechtel · 2014-04-17 Reply Admin

RFox:
Because this method was decommissioned and they weren't quite sure they got all the calls to it yanked out of all the client code...so they made sure any code that still called it would fail dramatically.
Captcha: plaga - plaga of bad code.

public DataModel getEditionModel() { if ( true ) throw new IllegalArgumentException( "You shouldn't be here" ); return editionModel; }

I disagree. While I think you're on the right track I think it's more about finding use cases that actually trigger the code. If you just want to find the calls remove it and see what fails to compile.

chubertdev · 2014-04-17 Reply Admin

Walky_one:
Note: accessing a constant will always be faster than accessing Array.Length

And neither will be as slow as using both.

chubertdev · 2014-04-17 Reply Admin

Bobby Tables:
Trebla:
ANON:
Or deleting it, so that it just wouldn't compile anymore.
That's no excuse for not deprecating the method and noting that it should be removed in future releases... or for checking if(true).

Deprecating - perhaps it's possible, perhaps the architecture and/or methodology would make it hard or impossible. Which would imply they have the wrong architecture and/or methodology but hacking this and then patching the resulted faults would be simpler than swapping the entire way how they build whatever system this is.

As for if (true) - some compilers would complain or actually refuse to compile this method if you just have it as
public DataModel getEditionModel() {
   throw new IllegalArgumentException( "You shouldn't be here" );
   return editionModel;
}
for the return statement is unreachable. Similarly, doing a straight
public DataModel getEditionModel() {
   throw new IllegalArgumentException( "You shouldn't be here" );
}
Would cause the compiler to tell you there is no return statement. The only way to both always throw an exception and have it compile is to have it in a if (true). Now, of course this is not a good way of doing it, but as pointed above may be the best way in the situation which would make it's the...well, correct incorrect way of doing it.


public DataModel getEditionModel() {
   throw new UnsupportedOperationException( "You shouldn't be here" );
   return null; // just in case!
}

chubertdev · 2014-04-17 Reply Admin

Jay:
Sure. That's self-documenting code. I mean:
dateLength="YYYY-MM-DD".length()
A reader knows exactly what the length means. While:
dateLength=10
Who knows?

I probably would have written:
dateLength=10
// date string is YYYY-MM-DD
but arguably that's inferior, because a later programmer could come along and change the length of the string -- maybe add the time of day or add an AD/BC indicator or whatever -- and change the assignment statement but not change the comment. So now it says
dateLength=14
// date string is YYYY-MM-DD
Yeah, that would be dumb, but programmers do it all the time.

But of course taking the length of a string of spaces tells me less than dateLength=10. Now I have to count the spaces. What did that gain?

int characterLengthOfProperlyFormattedDateString = 10;

2014-04-17 Reply Admin

unction countDocuments() { return
user.documents.length == 0 ? 0 : user.documents.length == 1 ? 1 : user.documents.length == 2 ? 2 : user.documents.length == 3 ? 3 : user.documents.length == 4 ? 4 : user.documents.length == 5 ? 5 : user.documents.length == 6 ? 6 : user.documents.length == 7 ? 7 : user.documents.length == 8 ? 8 : user.documents.length == 9 ? 9 : user.documents.length == 10 ? 10 : user.documents.length == 11 ? 11 : user.documents.length == 12 ? 12 : user.documents.length == 13 ? 13 : user.documents.length == 14 ? 14 : user.documents.length == 15 ? 15 : user.documents.length == 16 ? 16 : user.documents.length == 17 ? 17 : user.documents.length == 18 ? 18 : user.documents.length == 19 ? 19 : user.documents.length == 20 ? 20 : user.documents.length == 21 ? 21 : user.documents.length == 22 ? 22 : user.documents.length == 23 ? 23 : user.documents.length == 24 ? 24 : user.documents.length == 25 ? 25 : user.documents.length == 26 ? 26 : user.documents.length == 27 ? 27 : user.documents.length == 28 ? 28 : user.documents.length == 29 ? 29 : user.documents.length == 30 ? 30 : user.documents.length == 31 ? 31 : user.documents.length == 32 ? 32 : user.documents.length == 33 ? 33 : user.documents.length == 34 ? 34 : user.documents.length == 35 ? 35 : user.documents.length == 36 ? 36 : user.documents.length == 37 ? 37 : user.documents.length == 38 ? 38 : user.documents.length == 39 ? 39 : user.documents.length == 40 ? 40 : user.documents.length == 41 ? 41 : user.documents.length == 42 ? 42 : user.documents.length == 43 ? 43 : user.documents.length == 44 ? 44 : user.documents.length == 45 ? 45 : user.documents.length == 46 ? 46 : user.documents.length == 47 ? 47 : user.documents.length == 48 ? 48 : user.documents.length == 49 ? 49 : 50; }

Matt Westwood · 2014-04-17 Reply Admin

Herr Otto Flick:
Balu:
The physicist takes is ever present chainsaw and after the tree has fallen he takes a folding meter stick, measures the tree, tells the other two: "45 ft! Have a nice day!" and walks on

TRWTF

Oh come on, wake up, stupid. My folding meter stick also has imperial measurements on it in case I bump into American mathematicians arguing over the height of a tree. You don't need it for European ones because they're clever enough to determine tree height without needing a physicist to chop it down for them.

Matt Westwood · 2014-04-17 Reply Admin

faoileag:
dkf:
faoileag:
Count Documents.
Ah yes, the lesser known (and rather less scary) younger brother of Count Dracula.
Nope, the lesser known (and more prone to bureaucracy) older brother of Count von Count.

Any relation to Count von Toothree?

Matt Westwood · 2014-04-17 Reply Admin

Jay:
Zacrath:
RFox:
Because this method was decommissioned and they weren't quite sure they got all the calls to it yanked out of all the client code...so they made sure any code that still called it would fail dramatically.
Captcha: plaga - plaga of bad code.

public DataModel getEditionModel() { if ( true ) throw new IllegalArgumentException( "You shouldn't be here" ); return editionModel; }
And, of course, using @Deprecated won't work because everyone ignores warnings.

Or this code was written before the @deprecated tag was added to Java. Or, let's be fair, maybe the programmer isn't familiar with @deprecated.

RE just delete the function and get the compiler errors: But what if this function is in a library that might be called by many programs? The programmer wouldn't necessarily know all the programs that use this library. So if he just deleted the function, then six months later somebody recompiles and gets this mysterious not-found error, and he has to figure it out. If that's the case, sure, it would be better to put in a comment explaining why the function is obsolete, but I don't think this solution is totally idiotic. Not the best thing to do, but not totally idiotic.

So how is "You shouldn't be here" any more enlightening?

chubertdev · 2014-04-17 Reply Admin

Matt Westwood:
Herr Otto Flick:
Balu:
The physicist takes is ever present chainsaw and after the tree has fallen he takes a folding meter stick, measures the tree, tells the other two: "45 ft! Have a nice day!" and walks on

TRWTF

Oh come on, wake up, stupid. My folding meter stick also has imperial measurements on it in case I bump into American mathematicians arguing over the height of a tree. You don't need it for European ones because they're clever enough to determine tree height without needing a physicist to chop it down for them.

It's obviously 135 hands!

2014-04-17 Reply Admin

Bobby Tables:
Trebla:
ANON:
Or deleting it, so that it just wouldn't compile anymore.
That's no excuse for not deprecating the method and noting that it should be removed in future releases... or for checking if(true).

Deprecating - perhaps it's possible, perhaps the architecture and/or methodology would make it hard or impossible. Which would imply they have the wrong architecture and/or methodology but hacking this and then patching the resulted faults would be simpler than swapping the entire way how they build whatever system this is.

I can easily think of some valid reasons, where deprecating isn't possible and returning an exception is the correct implementation. Implementing interfaces or subclassing come to mind. For example, an ImmutableMap that implements the Map interface. The Map interface lets you put() new entries, but since your object is immutable you want to return an exception if anyone accidentally tries that with an instance of your class. Similarly for subclassing - an ImmutableHashMap could subclass HashMap, and just override the put() methods so they return an exception.

2014-04-17 Reply Admin

Rick:
Bobby Tables:
Trebla:
ANON:
Or deleting it, so that it just wouldn't compile anymore.
That's no excuse for not deprecating the method and noting that it should be removed in future releases... or for checking if(true).

Deprecating - perhaps it's possible, perhaps the architecture and/or methodology would make it hard or impossible. Which would imply they have the wrong architecture and/or methodology but hacking this and then patching the resulted faults would be simpler than swapping the entire way how they build whatever system this is.

I can easily think of some valid reasons, where deprecating isn't possible and returning an exception is the correct implementation. Implementing interfaces or subclassing come to mind. For example, an ImmutableMap that implements the Map interface. The Map interface lets you put() new entries, but since your object is immutable you want to return an exception if anyone accidentally tries that with an instance of your class. Similarly for subclassing - an ImmutableHashMap could subclass HashMap, and just override the put() methods so they return an exception.

Doesn't this violate the Liskov Substitution Principle though?

Matt Westwood · 2014-04-18 Reply Admin

not an anon (again):
Rick:
Bobby Tables:
Trebla:
ANON:
Or deleting it, so that it just wouldn't compile anymore.
That's no excuse for not deprecating the method and noting that it should be removed in future releases... or for checking if(true).

Deprecating - perhaps it's possible, perhaps the architecture and/or methodology would make it hard or impossible. Which would imply they have the wrong architecture and/or methodology but hacking this and then patching the resulted faults would be simpler than swapping the entire way how they build whatever system this is.

I can easily think of some valid reasons, where deprecating isn't possible and returning an exception is the correct implementation. Implementing interfaces or subclassing come to mind. For example, an ImmutableMap that implements the Map interface. The Map interface lets you put() new entries, but since your object is immutable you want to return an exception if anyone accidentally tries that with an instance of your class. Similarly for subclassing - an ImmutableHashMap could subclass HashMap, and just override the put() methods so they return an exception.
Doesn't this violate the Liskov Substitution Principle though?

You say that like it was a bad thing.

dkf · 2014-04-18 Reply Admin

not an anon (again):
Rick:
For example, an ImmutableMap that implements the Map interface. The Map interface lets you put() new entries, but since your object is immutable you want to return an exception if anyone accidentally tries that with an instance of your class. Similarly for subclassing - an ImmutableHashMap could subclass HashMap, and just override the put() methods so they return an exception.
Doesn't this violate the Liskov Substitution Principle though?

Since the exception is a documented part of the interface, it's A-OK with the LSP. Strictly assuming that put() never throws an exception is not warranted by the interface description. (No, just looking at the types doesn't tell you everything about the interface; you've got to read it yourself too to understand the semantics. It's technically possible to encode the semantics in computer-readable form, but You Don't Want That. Computer-readable semantics is usually horrendous!)

Coyne · 2014-04-18 Reply Admin

RFox:
Because this method was decommissioned and they weren't quite sure they got all the calls to it yanked out of all the client code...so they made sure any code that still called it would fail dramatically.
Captcha: plaga - plaga of bad code.

public DataModel getEditionModel() { if ( true ) throw new IllegalArgumentException( "You shouldn't be here" ); return editionModel; }

Okay, I'll buy that. There's just one question that remains: Why does it still return editionModel? I mean, do we still need to return that decommissioned value as opposed to, say, null?

2014-04-18 Reply Admin

Maurizio:
function countDocuments() { var count = 0;
for ( var i = 0; i < user.documents.length; i++) {
    count++;
}

return count;
}
Can't you people read ? The name of the method is "Count Documents", it is not "Tell me how many documents are there". So the code do that, counts documents. What else it should do ?

Captcha: ullamcorper: ancient exotic sexual practice: "last time we did an ullamcorper, i got back pain for a week".

Why does it return them though?

2014-04-18 Reply Admin

Matt Westwood:
faoileag:
dkf:
faoileag:
Count Documents.
Ah yes, the lesser known (and rather less scary) younger brother of Count Dracula.
Nope, the lesser known (and more prone to bureaucracy) older brother of Count von Count.
Any relation to Count von Toothree?

Yess, sere is.

2014-04-18 Reply Admin

Bobby Tables:
As for if (true) - some compilers would complain or actually refuse to compile this method if you just have it as
public DataModel getEditionModel() {
   throw new IllegalArgumentException( "You shouldn't be here" );
   return editionModel;
}
for the return statement is unreachable.

You're right on that one... the version "if(true)" will cause a warning from an IDE, but isn't actually a compiler error as the version above would.

Bobby Tables:
Similarly, doing a straight
public DataModel getEditionModel() {
   throw new IllegalArgumentException( "You shouldn't be here" );
}
Would cause the compiler to tell you there is no return statement.

But you're wrong on that one... this is perfectly valid Java code, though normally seen with UnsupportedOperationException instead. A function must return a value only if it doesn't throw an exception...

2014-04-18 Reply Admin

Loren Pechtel:
If you just want to find the calls remove it and see what fails to compile.

A good start, but no guarantee. It's a "getter" method, so there's a very good chance it's called via reflection, from some reference to an "editionModel" property in an XML file or something...

2014-04-18 Reply Admin

Jay:
What gets me about this one is that he used the length member right there. I understand if someone writes his own function to do something even though there's a built-in function in the standard library. So okay, he didn't know it was there. Ignorant, but not stupid. But this ...

Let me introduce you to TheDailyWTF that, to this day, has caused me to laugh longer and harder than any other: A False Detector.

That WTF is like an Ogre... there're just WTFs layered on top of WTFs. And one of them is what you say: which is that even if you just look at that code and don't use any language features not used there, there are still much better ways to write it.

chubertdev · 2014-04-18 Reply Admin

S:
But you're wrong on that one... this is perfectly valid Java code, though normally seen with UnsupportedOperationException instead. A function must return a value only if it doesn't throw an exception...

I've never really been fond of the throws keyword.

public DataModel getEditionModel() throws IllegalArgumentException {
   throw new IllegalArgumentException( "You shouldn't be here" );
}

2014-04-19 Reply Admin

You introduced a bug there...

2014-04-19 Reply Admin

foo:

String timeStampLength = "                          ";
int lengthOfTimeStamp = timeStampLength.length();

How confusing! I suggest:

String sTimeStampLength = "                          ";
int iLengthOfTimeStamp = timeStampLength.length();

You introduced a bug there...

2014-04-20 Reply Admin

Hannes:
I've seen similar code to the one Dennis sent in.
int newPage = 0; for (int i1 = 0; i1 < (dataGridView1.Rows.Count - 1); i1++) { newPage = i1; }

Yes. The programmer tried to determine the rows count of the DataGridView. So, to do this, he just wrote a for-loop and used the count of rows in the DataGridView as it's terminator. :(

I think I know what is going on here.

Codemonkey programmer only seen use of the array.size / rows.count propperties used inside a for loop, never in a direct assignment. So he/she thinks to be clever and does the for-loop "trick". Wow....

Complete lack of understanding of the fundamentals I reckon.

Do you see the pattern, too?

Matt Westwood · 2014-04-20 Reply Admin

the way Alex has a baby dick:
Ran the following code against today's article, glad I put the try catch in here because it could have crashed without it!
Try
  MakeFunnyArticle(joke,humor)
Catch Ex as Exception
  Msgbox("UNFUNNY ARTICLE!")
Finally
  WriteComment(lame,unfunny)
End Try

FTFY

2014-04-21 Reply Admin

For the last WTF, the joke isn't that they're throwing an exception right after catching one, but that the try block is empty. I assume they either deleted whatever was in the try block, or they wanted to put the try block around every call to the method but couldn't figure out how.

Side-note: If throwing an exception right after catching one (to change it to a more descriptive message), you should reference the first exception in the constructor of the second, so that its stack-trace information of the first is kept (otherwise when the exception comes up, it only points to where you threw the 2nd exception, and you can't check where the original error came from). Here's an example: http://blog.codinghorror.com/rethrowing-exceptions/

2014-04-22 Reply Admin

ComputerForumUser:
for ( var i = 0; i < user.documents.length; i++) {
This probably started out as
foreach (var document in documents) {
and then somebody said "You should use documents.length instead of the foreach."

Or maybe someone changed user.documents.constructor.prototype and as a result the new properties showed up in the in loop.

Pawprint · 2014-04-22 Reply Admin

Because no one did it: "Ohhhhhh, sweet mystery of life at last I found you!"

2014-04-23 Reply Admin

Haha! WTF!

2014-04-23 Reply Admin

ComputerForumUser:
This probably started out as
foreach (var document in documents) {
and then somebody said "You should use documents.length instead of the foreach."

Which would be wrong, because when multi-threading those aren't necessarily equivalent.

In C# length could be a property which returns a different value each time even when single threading.

2014-06-26 Reply Admin

He also may have anticipated the need to filter out some of the items in that array and left that function as an easy-to-find hook to install the feature when the time comes. Frankly, I've written code like that, and at times it has been rewarding. I think the fact that he used the length command indicates it was intentional.

Sweet Mysteries of Life

Leave a comment on “Sweet Mysteries of Life”