What the hell?

[AntiVirus]

Long story short: it's because of a bug when deciding whether or not a file is unicode.

And incidentally, it's not all about 4 3 3 5.  For example, "Bush hid the truth" doesn't work.

It's not really a bug.  A heuristic is being used to guess whether a text file with no indication as to whether it is Unicode or not is probably Unicode.  As such, it's really just making an educated guess based on whether something is likely Unicode or not, and like most guesses, sometimes it gets it wrong.

And it doesn't learn from its mistakes either!  Sheesh.

[Sidoh]

And it doesn't learn from its mistakes either!  Sheesh.

If it could identify a mistake, there wouldn't be any...

[AntiVirus]

And it doesn't learn from its mistakes either!  Sheesh.

If it could identify a mistake, there wouldn't be any...

....  It was a joke.

And there would be a mistake.  It would have to make atleast one mistake first before it learned from it.   That's how you learn from mistakes, you make them.

[Sidoh]

....  It was a joke.

And there would be a mistake.  It would have to make atleast one mistake first before it learned from it.   That's how you learn from mistakes, you make them.

How's it going to learn from a mistake if it's unable to identify it?  That was my point.

[AntiVirus]

....  It was a joke.

And there would be a mistake.  It would have to make atleast one mistake first before it learned from it.   That's how you learn from mistakes, you make them.

How's it going to learn from a mistake if it's unable to identify it?  That was my point.

... My first post was about it identifying the mistake and then learning about it..

[Sidoh]

I don't know about you, but I'd hate to have to click "Yep, it's encoded correctly" every time I opened something with notepad.

[iago]

Hmm, I wonder if there's some way to fool it for longer text?

That might be useful for somebody who wants to raise the bar for viewing source on their website, since IE opens source in notepad. 

[Sidoh]

Hmm, I wonder if there's some way to fool it for longer text?

That might be useful for somebody who wants to raise the bar for viewing source on their website, since IE opens source in notepad. 

An interesting question... I'd guess it gets increasingly hard as the text it has to sample gets longer, though.

[iago]

That's assuming they take the entire file as a sample.  Perhaps they take the first 1024 bytes, in which case it'd be much more possible.

[Skywing]

That's assuming they take the entire file as a sample.  Perhaps they take the first 1024 bytes, in which case it'd be much more possible.

Notepad will use up to the first 1024 bytes in conjunction with IsTextUnicode (using all available tests).  Determining just what probabilities the tests use is left as an exercise for the reader.

[AntiVirus]

I don't know about you, but I'd hate to have to click "Yep, it's encoded correctly" every time I opened something with notepad.

-.-;;

I mean the program had almost like AI or something. Might not be possible, but it'd be nice.