The Incredible Importance of NULL Testing

[billroper]

If you do not check for NULL pointers, you can get away with it right up until the point when the code presents you with one, at which time your code will merrily GPF. I am currently at about line 1400 in someone's 2800 line file, adding in the NULL pointer checks which aren't there at the moment.

Before I finish this, I am going to Sweet Baby Ray's for dinner.

Because I deserve it.

Comments

[catalana.livejournal.com]

*hug*

[billroper]

Thanks. I tested my fixes and things no longer GPF, which is good. Now I get to merge this to two more code lines.

There's still another interesting thing going on in the code, but I'm going to pass that off to the guy who wrote it and see what happens. :)

[madfilkentist]

That's called "design by contract," isn't it? If the caller doesn't provide valid data, the subroutine is allowed to act unpredictably.

As you might guess, I'm not a fan of design by contract.

[billroper]

Sadly, no, it wasn't design by contract. The NULL is an intended input -- as nearly as I can tell -- in the case where no entity is specified or required. It just wasn't being handled correctly.

I was willing to let a couple of instances pass where I could reasonably guarantee that a function call would return a valid pointer, but only because I got tired of reindenting this fellow's code. :)

(If he used my indenting and brace style, I probably would have done it, but I was doing too much hand fixup.)