A fair share of the WTFs for python are probably due to the WTForms library. I count 6 out of 10 WTForms matches on the first page of github file results.

https://github.com/search?q=wtf+extension%3Apy&type=Code&ref...

And codebases like Webkit have WTF throughout - Web Template Framework - http://stackoverflow.com/questions/834179/wtf-is-wtf-in-webk...

The Web Template Framework is also the explanation for ObjC's high WTF count.

I thought this was about anti-patterns or cryptic code passages. It's literally about the string 'wtf'

One could argue that the 'WTFs' themselves are a bit of a (completely unscientific) marker of just such anti-patterns etc...

They might be, but equally they could be a marker of an unorthodox but brilliant piece of work (fast inverse square root springs to mind). I suppose that's the unscientific part, but it does strike me as being a bit of a useless marker.

One would be arguing wrongly. There are a lot of things that gets abbreviated to wtf other than "what the fuck".

I should start a library called Write Template Format with cross language support....

Is the graph normalized to account for lines of code per repository? WTFs by repository is a useless number: I'd be more interested in seeing the WTFs per line.

Lines of C are certainly not equivalent to lines of, say, Erlang, so I don't think that metric would be extremely useful either.

Perhaps something like WTFs per commit would be best?

Well that is C:s problem, isn't it? If it want a better measure value then it should be more expressive. I think it is fair.

Yes, C wants a better measurement of WTFs. Get a grip.

Last year I did something similar, but instead of searching the code I looked at commit messages that matched certain patterns to categorize them by emotions, you can see the results here

http://geeksta.net/geeklog/exploring-expressions-emotions-gi...

People would probably only use wtf if the language did something that they didn't expect of it, regardless of how wtf the language actually is. This just shows that Python programmers programming in Python use WTF x times and Java programmers using Java use WTF y times.

(This would also explain why the wtf rate for Javascript is so low ;) )

Lua is mostly just people using wtf in a l10n var (SoundWTF) or as a variable or something trying to be funny:

https://github.com/search?q=wtf+extension%3Alua&type=Code&re...

The answer is in the comments of that site. Lua is used in lots of places, but the earliest "famous" one is World of Warcraft, where it is used to make extensions. The folder where those extensions are stored is called WTF (Warcraft Text File), so on every warcraft extension the string WTF appears multiple times. See http://www.wowwiki.com/WTF

The graph doesn't show up in Opera, Firefox, or Chrome for me.

It's on the 'raw data' page, not on the post itself.

NoScript?

Nope, no extensions whatsoever.

Isn't "fuck" a better pattern to measure this? However, funny project :D

How many of those Java WTFs are actually Android Log.wtf() logs?

From the Android API docs:

"What a Terrible Failure: Report an exception that should never happen."

http://developer.android.com/reference/android/util/Log.html..., java.lang.String)

Wow, I'm not sure if I should be proud or ashamed, the two languages I use everyday now - C and Objective-C are in the winner's circle.

reminds me of commit logs from last night

http://www.youtube.com/watch?v=V44kscaJe3M