The Daily WTF: Curious Perversions in Information Technology

Romeo-Gresta · 2004-12-20 Reply Admin

Omg... still asking myself what kind of number is NOT > 1 AND is NOT = 1 AND is NOT < 1 AND is = -1...

2004-12-20 Reply Admin

Oh, boy. I do like the final option - the one where on the second unreachable case - he then doesn't divide by 100. Smooth.

Ah, unless VB understands the concept of NAN.

znaps · 2004-12-20 Reply Admin

lmao.

You wonder why he didn't add an "if v_Value= 0..." just for good measure.

2004-12-20 Reply Admin

I'm glad this guy isn't on my team... I mean.. two unreachable cases, and 3 identical statements besides (4, but one is unreachable).. why not just:

ConvertPercent = v_value / 100

And be done with it?

Jeff S · 2004-12-20 Reply Admin

Yikes. As often occurs, the post is a WTF but the comments are even worse so far.

All situations in this IF statement ARE reachable.

If v_value is >-1 and <1, then he leaves it alone. Otherwise, it is divided by 100. So he can't just divide it by 100 no matter way.

yes, the code is a WTF, much more convoluted than it needs to be. But these comments are pretty bad ... come on, guys, let's pay attention! I know it's the holidays and all, maybe you've had too much egg nog .....

Jeff S · 2004-12-20 Reply Admin

(quick note: I forgot to add that I am making the assumption that there's a typo on the 3rd IF -- if not, then, yes, there are unreachable situations. But i suspect he meant to say < -1 and not < 1. I'm curious if the code was properly reproduced here on this web page)

Kermos · 2004-12-20 Reply Admin

What the F........?

I've seen some scary code but....this...is...

This person actually has a job!?

2004-12-20 Reply Admin

Jeff S:
Yikes. As often occurs, the post is a WTF but the comments are even worse so far.

All situations in this IF statement ARE reachable.

Please tell us all, what number is not greater than one, less than one, or one?

There must be some new additions to the number line...

2004-12-20 Reply Admin

I wonder how he would have coded the ConvertPolynomial function.

2004-12-20 Reply Admin

I call fake. Nobody is that dumb.

Oh wait, this is VB? Ah, ok nevermind, it is possible then

2004-12-20 Reply Admin

so it seems clear (?) to me that case three was supposed to be "ElseIf v_value < -1 Then"

That makes the function make some sense: A value between -1 and 1 is assumed to be already the decimal representation of the percentage and a value whose absolute value is 1 or greater is assumed to be the percentage representation of a percentage... ie .30 and 30 are both assumed to mean 30% or 0.3

2004-12-20 Reply Admin

So thats what it was probably supposed to be, but it's just buggy:

<FONT color=blue>Static Function</FONT> ConvertPercent(v_value <FONT color=blue>As Double</FONT>)
  <FONT color=blue>If</FONT> (v_value >= 1) <FONT color=#0000ff>or</FONT> (v_value <= -1) <FONT color=blue>Then</FONT>
    ConvertPercent = v_value / 100
  <FONT color=blue>Else</FONT> 
    ConvertPercent = v_value
  <FONT color=blue>End If</FONT> 
<FONT color=blue>End Function</FONT>

DWalker · 2004-12-20 Reply Admin

Oddly, I would think that if you're trying to interpret a number as a percentage, and the number being converted is < 1, then it *shouldn't* be divided by 100. But that's not what happens here.

David Walker

2004-12-20 Reply Admin

I agree with Jeff S that the guy meant to say < -1 for the third statement (although I don't agree with his characterization of the other commenters). Somebody probably made a typo along the way, but even if it was the coder, a typo is not a WTF.

What this guy wants to say is if abs(v_value) >= 1, divide by 100, otherwise don't. So the potential WTFs here are (a) not knowing about >=, (b) not taking absolute value, and (c) assuming that any number less than 1 is not a percent (ex. If I enter 0.5 meaning one half percent, his code will leave it as 0.5, turning it into 50%.).

Maybe he did (a) because he thought it was more readable. Maybe he chose (b) because he thought the abs() call would be too expensive. Maybe (c) isn't a problem because he doesn't have to deal with %s less than 1.

Jeff S · 2004-12-20 Reply Admin

Yes, I am embarrased. [:$] I indicated that others should pay more attention, when I was not paying enough attention myself! I'm the one drinking too much eggnog, apparently ! [D]

2004-12-20 Reply Admin

Don't defend this!!! And all of you who write out what the code should be pat yourself on the back... you can code better then a monkey... (but barely)

The fact of the matter remains that the code posted SUCKS!!! even if it should be a -1... [8o|]

Jacob K · 2004-12-20 Reply Admin

Hmm. Interesting and painful all at the same time. Supefluous is the perfect word for it (like Alex said).

DJDanny · 2004-12-20 Reply Admin

So it's not possible to have 0.5% of something?!

It seems not. According to this guy, I meant 50%!

Yeah, sorry. My bad!

2004-12-20 Reply Admin

How about entering 0.005, then?

Miles Archer · 2004-12-20 Reply Admin

sqr(-1), eg imaginary numbers. That's what the else is for.

Free · 2004-12-20 Reply Admin

Boring

Not much of a WTF, fixed quickly, gone forever.

If v_value > 1
Or v_value = 1
Or v_value < 1
No beer for you! Go back to COBOL.

Static Function ConvertPercent(v_value As Double)
ConvertPercent = v_value / 100
End Function

<rant>
I would say that the real WTF is the ElseIf, a real piece of work ElseIf.
You want help debugging that, first get rid of all the ElseIf!
</rant>

2004-12-20 Reply Admin

:
you can code better then a monkey... (but barely)

...and your English skills are better than a monkey's.

Tom · 2004-12-21 Reply Admin

Alex Papadimoulis:
Ben Murphy found a developer who liked to cover all of his bases

All his code are belong to us.

2004-12-21 Reply Admin

DJDanny:
So it's not possible to have 0.5% of something?!

It seems not. According to this guy, I meant 50%!

Yeah, sorry. My bad!

I agree with this poster. Even if it's some convoluted typo, then it doesn't take away that this is a serious design mistake. Do _NOT_ mix datatypes. A computer hates ambiguity, and for good reason. Either enter 30% or 0.3 but not both.

Or you end up with terrible logic error like the quoted poster shows.

2004-12-21 Reply Admin

The webdesign of this page is becoming a wtf in itself. Why is the width fixed?

I hate scrolling horizontally. I liked the design much better when it was simple (say 2 months ago).

greetings

2004-12-21 Reply Admin

Assuming that you were using a language that supported NaN, wouldn't the result of NaN / 100 be NaN?

What I really like is the suggestion that to get support for 0.5% we should enter 0.005%. We'd need a smart wrapper program that would support it! How about ...

Static Function AlwaysReallyConvertPercent(v_value As Double)
If (v_value > -1) and (v_value < 1) Then
AlwaysReallyConvertPercent = ConvertPercent(v_value / 100)
Else
AlwaysReallyConvertPercent = ConvertPercent(v_value)
End If
End Function

"That's real programming, that is!"

Kermos · 2004-12-21 Reply Admin

:

The webdesign of this page is becoming a wtf in itself. Why is the width fixed?

I hate scrolling horizontally. I liked the design much better when it was simple (say 2 months ago).

greetings

Width fixed? I haven't noticed =)

Then again, 1280x1024 on a 19 inch LCD does help....no horizontal scrolling here!

2004-12-21 Reply Admin

:
Assuming that you were using a language that supported NaN, wouldn't the result of NaN / 100 be NaN?

Well, yes.

But I was thinking about how to get to the second unreachable case. I'm going to assume a language implementation that (a) doesn't raise an exception when a NaN is compared to a value, and that (b) such a comparison always returns false except for the classic Nan != value version.

2004-12-21 Reply Admin

Time: 1995
Place: Boston, Mass
Data: simple 64-bit wide floats (totally normal: not NaN, or wierd, or anything. Just plain "double" variables like ".6")
Test Code:

if (a>b) printf (">\n");
if (a >=b) printf (">=\n");
if (a==b) printf ("==\n")
if (a!=b) printf ("!=\n");
if (a<=b) printf ("<=\n");
if (a<b) printf ("<");

and for an answer I got:

>=
!=
<=

Huh? Two numbers that aren't the same (!=) but where one isn't strictly greater than or less than the other? WTF?

Answer: Compiler bug on the DEC Alpha compiler! The numbers, while not "strange", were not "made" the same way. One was created on a VAX using the VAX "DFLOAT" bit fields and the other on an Alpha, also with DFLOAT bit fields. The Alpha is strictly IEEE floats only; DFLOAT are converted into IEEE floats, and that loses precision.

The > and < would convert the DFLOATs into IEEE format, dropping precision. With the slightly lower precision the numbers were the same. The == and !=, however, just do bit-by-bit compares at higher precision and the numbers are not the same.

Thank goodnessevery single double comparison was macro-ized and so the problem could be worked around with a simple cast-to-double (which made the compiler bug go away).

Peter Smith

2004-12-22 Reply Admin

:

So thats what it was probably supposed to be, but it's just buggy:

<FONT color=blue>Static Function</FONT> ConvertPercent(v_value <FONT color=blue>As Double</FONT>)
  <FONT color=blue>If</FONT> (v_value >= 1) <FONT color=#0000ff>or</FONT> (v_value <= -1) <FONT color=blue>Then</FONT>
    ConvertPercent = v_value / 100
  <FONT color=blue>Else</FONT> 
    ConvertPercent = v_value
  <FONT color=blue>End If</FONT> 
<FONT color=blue>End Function</FONT>

In .NET you no longer have to set a return type? as in:

[code language="vb"]
Function ConvertPercent(ByVal v_value As Double) As Double
[/code]

Katja · 2004-12-22 Reply Admin

> And on another note, I think it's high time we had a logo. If you're the creative type, post your idea (rough draft ok).

I do agree but most here are developers. Creativity is something in the past for us... [<:o)] (Unless you see all the nice work on this site. You must be very creative to create those WTF's...)

2004-12-23 Reply Admin

:
How about entering 0.005, then?

NO! Nononononono! [:@]

I so hope you are joking! This is horrible, horrible code. Even if there's a typo, it's UGLY and fundamentally flawed in design!

Either go by percentage-values or don't! Don't mix and code!

fcarlier · 2004-12-25 Reply Admin

Miles Archer:
sqr(-1), eg imaginary numbers. That's what the else is for.

WTF. If you use imaginary numbers, you can't use < or > anymore. They are not sorted.

2004-12-30 Reply Admin

What he probably meant was this:

<font color="blue">Static Function</font> ConvertPercent(v_value <font color="blue">As Double</font>)
  <font color="blue">If</font> v_value > 1 <font color="blue">Then</font>
    ConvertPercent = v_value / 100
  <font color="blue">ElseIf</font> v_value = 1 <font color="blue">Then</font>
    ConvertPercent = v_value / 100
  <font color="blue">ElseIf</font> v_value < -1 <font color="blue">Then</font>
    ConvertPercent = v_value / 100
  <font color="blue">ElseIf</font> v_value = -1 <font color="blue">Then</font>
    ConvertPercent = v_value / 100
  <font color="blue">Else</font> 
    ConvertPercent = v_value
  <font color="blue">End If</font> 
<font color="blue">End Function</font>

He just forgot one dash.
Now it almost makes sense. He didn't want to convert numbers that already were converted.

Still, couple unnecessary elseifs...

2005-01-03 Reply Admin

You should perhaps read some of the posts above.

Maurits · 2005-01-03 Reply Admin

fcarlier:
[image] Miles Archer wrote:
sqr(-1), eg imaginary numbers. That's what the else is for.

WTF. If you use imaginary numbers, you can't use < or > anymore. They are not sorted.

2005-01-03 Reply Admin

Maurits:

[image] fcarlier wrote:

[image] Miles Archer wrote:

sqr(-1), eg imaginary numbers. That's what the else is for.

WTF. If you use imaginary numbers, you can't use < or > anymore. They are not sorted.

Yep, quoting works.

2005-01-11 Reply Admin

Assuming the type used (Double) is a IEEE754 floating-point type, there are three cases, where given the value v,

v > 1 is false
v < 1 is false
v = 1 is false

When v is NAN (Not a Number as discussed), positive infinity or negative infinity.

Here is some fun in Java (where floating-point types are per the IEEE754 standard):

// What is the output? Calculators are permitted.
public class X
{
    public static void main(String[] args)
    {
        System.out.println(641 * 6700417);
        System.out.println(16777216*256);
        System.out.println(-2147483648 * -1);
    }
}

2005-01-11 Reply Admin

Assuming the type used (Double) is a IEEE754 floating-point type, there are three cases, where given the value v,

v > 1 is false
v < 1 is false
v = 1 is false

When v is NAN (Not a Number as discussed), positive infinity or negative infinity.

Here is some fun in Java (where floating-point types are per the IEEE754 standard):

// What is the output? Calculators are permitted.
public class X
{
    public static void main(String[] args)
    {
        System.out.println(641 * 6700417);
        System.out.println(16777216*256);
        System.out.println(-2147483648 * -1);
    }
}

Drak · 2005-01-12 Reply Admin

if v_value * v_value >= 1 then
ConvertPercent = v_value / 100
else
ConvertPercent = v_value
end if

Might be quicker than using the abs() mentioned before.

Still, baaad code!

2005-01-13 Reply Admin

:
Assuming the type used (Double) is a IEEE754 floating-point type, there are three cases, where given the value v,

v > 1 is false
v < 1 is false
v = 1 is false

When v is NAN (Not a Number as discussed), positive infinity or negative infinity.

Here is some fun in Java (where floating-point types are per the IEEE754 standard):

// What is the output? Calculators are permitted.
public class X
{
 public static void main(String[] args)
 {
 System.out.println(641 * 6700417);
 System.out.println(16777216*256);
 System.out.println(-2147483648 * -1);
 }
}

Trick question. None of the values are floating point, they are all ints, so the issue of how floating point is handled, and NaN, is irrelevent. And all three expressions will overflow (not that the second or third will be reached) and 641*6700417 will cause a *looks it up* ArithmeticException to be thrown, which is not being caught by anything. So the output will be an unhandled exception message, and a backtrace. What do I win?

AndrewVos · 2005-01-18 Reply Admin

:

I call fake. Nobody is that dumb.

Oh wait, this is VB? Ah, ok nevermind, it is possible then

GEEZ! i cant believe i found this! get back to ur C! die coding pathetic C "apps" too!

2012-07-31 Reply Admin

Romeo-Gresta:
Omg... still asking myself what kind of number is NOT > 1 AND is NOT = 1 AND is NOT < 1 AND is = -1...

bingo : this calls for "unreachable Code Error" from the compiler - everything below < 1 is not needed and will never get executed

Covering All Cases ... and then some

Leave a comment on “Covering All Cases ... and then some”