Comment On The Friday Farrago

Are ya WTF-ed out yet? Let's hope not, otherwise you'd miss out on Duane Homick's discovery of the strftime documentation ...

%S is replaced by the second as a decimal number [00,61].

This is a WTF on the face of it, but there's a very good reason for
seconds to includ e  61.  That would be when leap seconds
happen, which they do periodically to allow clocks to  synchronize
with the gradual slowing rotation of the Earth.

Alex Papadimoulis:

could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Or ""

I laughed at most of the above, but I didn't see the problem with the strftime man page. I believe the values are valid because of leap seconds aren't they? Or am I missing something else that should be funny?

sinistral:

Are ya WTF-ed out yet? Let's hope not, otherwise you'd miss out on Duane Homick's discovery of the strftime documentation ...

%S is replaced by the second as a decimal number [00,61].

This is a WTF on the face of it, but there's a very good reason for
seconds to includ e  61.  That would be when leap seconds
happen, which they do periodically to allow clocks to  synchronize
with the gradual slowing rotation of the Earth.

Indeed, the linked manpage says just that:
"The range of values for %S is [00,61] rather than [00,59] to allow
for the occasional leap second and even more infrequent double leap second."

What's the obscure synonym of 'jumble'???

Alex Papadimoulis:

  if (this == null)

  {

    throw new ArgumentNullException("this");
  }
  

I've always hated those zombie classes.  They're dead but they just keep coming at ya.

Anonymous:

What's the obscure synonym of 'jumble'???

Farrago

Alex Papadimoulis:

We'll start out with this simple piece of code that Jake Vinson tripped over while chasing bugs ...

if(3 < 4)
{        
  if(chk.value == 'Yes')
    fld.value = 'No';
  else

    fld.value = 'Yes';    
}

I've written code vaguely similar to this to trick the compiler,
strictly debugging purposes.. I don't see how it applies here though.



When I do it, it's usually because I want to return from a function early just to test something, say:






But if I just stick a "return 5;" in where my comment is, the compiler
will complain that statementTwo can never be reached.  So I'll do
this:






And that is enough to trick the compiler.  But I don't see any
reason for the WTF code, since it is not avoiding any compiler
complaints.

Alex Papadimoulis:

IF (current_num - prev_num = current_num) THEN

If this is an odd way of testing "prev_num == 0", then it's a WTF.

But without knowing the types, it's hard to say.  If, for instance, they're floats, then the expression could be true with prev_num != 0.

Maurits:

sinistral:

Are ya WTF-ed out yet? Let's hope not, otherwise you'd miss out on Duane Homick's discovery of the strftime documentation ...

%S is replaced by the second as a decimal number [00,61].

This is a WTF on the face of it, but there's a very good reason for
seconds to includ e  61.  That would be when leap seconds
happen, which they do periodically to allow clocks to  synchronize
with the gradual slowing rotation of the Earth.

Indeed, the linked manpage says just that:
"The range of values for %S is [00,61] rather than [00,59] to allow
for the occasional leap second and even more infrequent double leap second."

Do they ever thought about a even more infrequent triple leap second?

Regarding the last example (dot net DrawImage method), I'm pretty sure
the (this == null) check is useful when the method is invoked via
reflection, and the "this" object is actually a parameter to the method
invocation.  Can anyone confirm this?

Alex Papadimoulis:

public void DrawImage(Image image, int x, int y)

{

  if (this == null)

  {

    throw new ArgumentNullException("this");

  }

}

I've seen this idiom quite a bit in C++ to help cushion against null pointer problems, since 0->func(a,b,c); compiles with thiscall as _decorated_func(0,a,b,c) -- but in .Net? WTF indeed...

So who's got the COBOL -> 2005 translation? :)

Scott B.:

So who's got the COBOL -> 2005 translation? :)

CONST FIVE = 4

Nothing new certainly, but fun to see it in COBOL.

 

Anonymous:

Regarding the last example (dot net DrawImage method), I'm pretty sure
the (this == null) check is useful when the method is invoked via
reflection, and the "this" object is actually a parameter to the method
invocation.  Can anyone confirm this?

'this' is a reserved keyword, that code would never compile:



Test.cs(17): Identifier expected, 'this' is a keyword

Well, I was forced to take cobol in HS (I swear I'm only 31 not 50) so I'll take a stab at it...



The field name is FIVE but its being set with the value of 4. 



I guess I didn't need my cobol for that either.


/took cobol in college even
//Never learned any JCL so all my cobol knowledge is probably a waste

///thats fine by me.

How in the world could "this" ever equal null?????

Hey Alex, when you gonna post my Coldfusion wtf?

They didn't consider "even more infrequent triple leap second" because the standard defines single and double leap seconds, but triple leap seconds are forbidden.



Actually, leap seconds are all going away.  They (damn, can't
remember the standards body) is going to let the time drift until it
reaches some large quanta like a minute in a few centuries.

Anonymous:

Regarding the last example (dot net DrawImage method), I'm pretty sure
the (this == null) check is useful when the method is invoked via
reflection, and the "this" object is actually a parameter to the method
invocation.  Can anyone confirm this?

So what you're essentially saying is, this check guards against callers
trying to call this method after dynamically creating the owner
object?  In the immortal words of Gene Wirchenko:



I hope you're not serious



Sincerely,

Gene Wirchenko

The strftime(3) man page on OpenBSD give more detail as to why %S goes all the way to 61.

"%S is replaced by the second as a decimal number
(00-61). The range of seconds is (00-61) instead of (00-59) to allow for the periodic occurrence of leap seconds and double leap seconds."

Anonymous:

Actually, leap seconds are all going away.  They (damn, can't remember the standards body) is going to let the time drift until it reaches some large quanta like a minute in a few centuries.

And $DEITY only knows what all the computer systems in the world will make of the time leaping forward a minute. (Especially any still using Unix-style timekeeping, of which there probably will be some, barring the total collapse of civilisation.)

Alex Papadimoulis:

And speaking of pointless code, John pulled this line of triply-redundant code from a Java production system that could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Not familiar with Java, but couldn't isActive have been set earlier to something like null, NaN, or undefined?

Jenny Simonds:

Alex Papadimoulis:

And speaking of pointless code, John pulled this line of triply-redundant code from a Java production system that could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Not familiar with Java, but couldn't isActive have been set earlier to something like null, NaN, or undefined?

If it was a Boolean, yes, but since its a boolean (small 'b'), no.

Alex Papadimoulis:

#define private private

#define genius idiot



The guy who wrote this is surely the greatest genius born to this world in a long time.

Matt B:

Anonymous:

Regarding the last example (dot net DrawImage method), I'm pretty sure
the (this == null) check is useful when the method is invoked via
reflection, and the "this" object is actually a parameter to the method
invocation.  Can anyone confirm this?

'this' is a reserved keyword, that code would never compile:



Test.cs(17): Identifier expected, 'this' is a keyword

That's essentially the point of the WTF in the code.  this refers to the instance of the object, and if it was null that method would never execute.



I have no clue regarding Renaldo's comment, as my experience with
Reflection in .NET is confined primarily to type checking and property
information.

ever tryed :

if (true) return 5;



or better:



if (IsTrue(5 > 4)) return5;



private bool IsTrue(int x, int y)

{

bool istrue;

if (x ==y) { istrue = true }

else { istrue = false }

return istrue



}



-C

Mung Kee:

Jenny Simonds:

Alex Papadimoulis:

And speaking of pointless code, John pulled this line of triply-redundant code from a Java production system that could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Not familiar with Java, but couldn't isActive have been set earlier to something like null, NaN, or undefined?

If it was a Boolean, yes, but since its a boolean (small 'b'), no.

Sorry, on second glance, this wasn't very helpful.  The value of
an uninitialized boolean is false in Java and there is no no-arg
constructor for the object type Boolean, so left uninitialized it would
be null.

Maurits:

Indeed, the linked manpage says just that:
"The range of values for %S is [00,61] rather than [00,59] to allow
for the occasional leap second and even more infrequent double leap second."

And the moral to the story: RTFMB4UWTF.

Maurits:

Alex Papadimoulis:

could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Or ""

Not always.

I've seen another form of the if (this == null) construct in Java.

Foo foo = new Foo();

if (foo == null) return;

kipthegreat:

But if I just stick a "return 5;" in where my comment is, the compiler
will complain that statementTwo can never be reached.

This is because the Java compiler especially is full of WTFs like emitting errors when it should only be emitting warnings.

Anonymous:

Do they ever thought about a even more infrequent triple leap second?

Not in this tense.

dubwai:

I've seen another form of the if (this == null) construct in Java.

Foo foo = new Foo();

if (foo == null) return;

You've gotta be careful, sometimes those pesky constructors will actually return a null reference.

IF (current_num - prev_num = current_num) THEN


This is a method of testing that the change in numbers is
too small to have any effect. I've seen and used that construct
a lot in doing numeric analysis there the algorithm converges
on some value.

sinistral:

  That would be when leap seconds
happen, which they do periodically to allow clocks to  synchronize
with the gradual slowing rotation of the Earth.

You're right about the leap seconds, but it's not that the earth is
slowing.  It's because the length of a solar year isn't exactly
the 365.24x days that the gregorian calendar allots for it.  If
you have to adjust a wall clock every few months, it's because it's too
fast, not because it's getting faster.

That .NET one was pasted from Reflector, and you'll find a lot of that nonsense in the reflected code.


Here's a pop quiz:  What's going on in this reflected code?

public sealed abstract class Foo

{

    //stuff

}

Mung Kee:

Jenny Simonds:

Not familiar with
Java, but couldn't isActive have been set earlier to something like
null, NaN, or undefined?

If it was a Boolean, yes, but since its
a boolean (small 'b'), no.

Autoboxing?

Ytram:

You've gotta be careful, sometimes those pesky constructors will actually return a null reference.

Not in any of C#, C++ or Java, so what language are you talking about?

Anonymous:

That .NET one was pasted from Reflector, and you'll find a lot of that nonsense in the reflected code.


Here's a pop quiz:  What's going on in this reflected code?

public sealed abstract class Foo

{

    //stuff

}

Seems perfectly alright to me if you don't ever want your class to be
instantiated or subclassed. The C# compiler won't allow a mere mortal
to write this, though. It would be better to just define only a private
constructor.

Alexis de Torquemada:

Ytram:

You've gotta be
careful, sometimes those pesky constructors will actually return a null
reference.

Not in any of C#, C++ or Java, so what language are you talking about?

Go into a C# project and create a class with a constructor.  In the constructor, do the following:



1.  Have the constructor take any value type, reference types won't work.

2.  Within the constructor, set the parameter to null.

3.  Add the line, "return null;" to the end of the constructor.

4.  Realize that I am being sarcastic.



Hopefully that should show you what I'm talking about.

Anonymous:

You're right about the leap seconds, but it's not that the earth is
slowing.  It's because the length of a solar year isn't exactly
the 365.24x days that the gregorian calendar allots for it.  If
you have to adjust a wall clock every few months, it's because it's too
fast, not because it's getting faster.

Not only because of the solar year length, but also to keep 00:00 at
precisely the moment when the sun is opposite the 0° meridian. The
earth's rotation isn't 100% uniform; it slows and speeds up due to the
moon and the other planets.

I suppose, something written here http://www.andymcm.com/dotnetfaq.htm, have produced such a confusion of ideas in the head of a programmer who wrote "if (this == null)". Actually, destructor may have been called during the method call in some circumstances, you know. 

Ytram:

4.  Realize that I am being sarcastic.

Steps 1 to 3 were easy, but now I'm stuck at number 4. Tough one.

I'm sorry, I should have clarified that you need to compile after step 3. [:P]

Anonymous:

IF (current_num - prev_num = current_num) THEN


This is a method of testing that the change in numbers is
too small to have any effect. I've seen and used that construct
a lot in doing numeric analysis there the algorithm converges
on some value.

this is why mathematicians cannot program.  QED.

The strftime() documentation is still wrong. Unless you modify a struct
tm's fields yourself, the seconds will never be more than 59. struct
tm's are generated by the gmtime and localtime functions, which take a
time_t so there it no way they can distinguish between

2004/12/31 23:59:60 and 2005/01/01 00:00:00, since they both would have a time_t value of 1104537600.

Mung Kee:

Jenny Simonds:

Alex Papadimoulis:

And speaking of pointless code, John pulled this line of triply-redundant code from a Java production system that could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Not familiar with Java, but couldn't isActive have been set earlier to something like null, NaN, or undefined?

If it was a Boolean, yes, but since its a boolean (small 'b'), no.

What if it's not a boolean?

Couldn't it be some other type with the == operator overloaded?

(I'm not that familiar with Java; does it even support operator overloading?)

Anonymous:

The strftime() documentation is still wrong. Unless you modify a struct
tm's fields yourself, the seconds will never be more than 59. struct
tm's are generated by the gmtime and localtime functions, which take a
time_t so there it no way they can distinguish between

2004/12/31 23:59:60 and 2005/01/01 00:00:00, since they both would have a time_t value of 1104537600.

Usually true, not always.
http://en.wikipedia.org/wiki/Leap_second

The next leap second will occur at the very end of December 31 2005.
Let me count the new year down for you...
2005/12/31 23:59:51 TEN!
2005/12/31 23:59:52 NINE!

2005/12/31 23:59:53 EIGHT!

2005/12/31 23:59:54 SEVEN!
2005/12/31 23:59:55 SIX!
2005/12/31 23:59:56 FIVE!
2005/12/31 23:59:57 FOUR!
2005/12/31 23:59:58 THREE!

2005/12/31 23:59:59 TWO!

2005/12/31 23:59:60 ONE!

2006/01/01 00:00:00 HAPPY NEW YEAR!

It will be interesting to see if the ball drops a second early.

John Smallberries:

Mung Kee:

Jenny Simonds:

Alex Papadimoulis:

And speaking of pointless code, John pulled this line of triply-redundant code from a Java production system that could be fairly easily simplified to ";".

isActive = (isActive == true) ? true : false;

Not familiar with Java, but couldn't isActive have been set earlier to something like null, NaN, or undefined?

If it was a Boolean, yes, but since its a boolean (small 'b'), no.

What if it's not a boolean?
Couldn't it be some other type with the == operator overloaded?
(I'm not that familiar with Java; does it even support operator overloading?)

Nope...no operator overloading in Java

The time_t will have the same value, 1136073600, on both of those last
two seconds. So there's no way localtime/gmtime can distinguish them,
unless it does something really weird.