Hacker News Reader

Comment by conover | Hacker News Reader

That's a staggering amount of code. Nearly unbelievable. To put that in perspective, Windows XP was 45 million lines of code[0].

[0] https://m.facebook.com/windows/posts/155741344475532

View on HN

My guess is they're counting code from existing packages used to create the site.

So for example, if they used Java, that line count is including the lines of code found in the JDK.

I see no other way of getting to such a large number.

I don't see it approaching that even with the lines of code for their existing packages. For instance, this is old and out of date, but you could extrapolate a ballpark estimate out of it... looks like back in 2007 or so JDK was about 6 million LoC (http://cr.openjdk.java.net/~darcy/OpenJdkDevGuide/OpenJdkDev...). Even if it increased tenfold since then, it's barely making a dent in that 500 million LoC number.

Don't forget to add in the lines of code for the Linux kernel on which they are running and the MySQL database etc etc.

I kept reading the article and on the last you find this gem:

According to one specialist, the Web site contains about 500 million lines of software code. By comparison, a large bank’s computer system is typically about one-fifth that size.

That to me suggests the number is bogus, because it is so obviously a quote from someone who has no idea what they are talking about.

Simply measuring the web site in 'lines of code' as if that's meaningful indicates that it's a quote by someone who is not aware of how modern programming works.

Disagree, perhaps. If you were talking to a New York Times reporter, with an intended lay audience, how would you describe scope issues like this?

And while it's a poor metric, lines of code is not completely useless.

The linux kernel is ~15 million LOC. That also doesn't make a dent in 500 mil.

It's hard to imagine that even those could total more than a small fraction of the purported 500 MLOC. The JDK is only a few million lines of code.

What was the site written in?

Assembly, to protect your privacy.

it's either Java or .net , but it feels more like Java. i'm more interested in knowing what the data layer is made of though.

Somewhat surprisingly, Visual Studio 2012 has 50 million lines of code. [0]

It's kind of weird to think that VS 2012 has more lines of code than Windows XP.

[0]: http://blogs.msdn.com/b/visualstudio/archive/2012/06/20/the-...

VS 2012 does a lot more.

That's a lot of eating your own dog food

Yeah, I'd call total BS on that number, unless they did something absolutely insane... I mean, we're talking "instead of saving user information in the database, we codegen'd a set of classes specifically for each regisered user and save them to our repository" level of insanity. Has there ever been software that's even come close to approaching 500 million lines of code?

Imagine someone threw together some automatic code generation thing. It tries every possible input and then creates a special if-then statement for it.

Don't worry, the compiler will optimize it.

My reaction as well. Even 500 million bytes would be hard to believe.

While "efficient" is not a word that comes to mind when thinking about the government, 500 lines of code would be 1.6 lines for every man, woman, and child in this country.

This is where my hypothesis comes in. Instead of a database, they're codgening a small class and adding it to their codebase for every registered user, and recompiling the entire site immediately following that operation, blocking requests until the recompilation is complete. This hypothesis neatly explains both the terrible performance and the unimaginable LoC estimation.

Reply on news.ycombinator.com