Comments on: Draft Paper about Integer Overflow

By: regehr

regehr — Thu, 13 Oct 2011 03:26:24 +0000

Sean, we do have these patches. Let me see about getting them uploaded.

By: Sean M

Sean M — Mon, 10 Oct 2011 03:52:17 +0000

Awesome stuff! Can’t wait to try it.

Do you have patches against llvm/clang svn trunk instead of against 2.9?

By: Daniel

Daniel — Sat, 08 Oct 2011 04:13:40 +0000

Some grammar nitpicks from a first reading:

pg 1, last paragraph: “because latent until a compiler upgrade…” should probably be “because they are latent” or similar

pg 6, last paragraph of left column: add comma in “and many_,_ like gcc, …”

pg 10, last full paragraph: the sentence beginning with “The tools vary from static analysis, …” doesn’t parse; perhaps replace comma with “and”?

By: Mans

Mans — Fri, 07 Oct 2011 23:10:59 +0000

This looks useful. However, I’d like a way to disable certain checks. Specifically, shifts of negative numbers is well-defined with most compilers and lots of code intentionally relies on it working. On such code, the checker as available today produces far too many false positives to be of much use.

By: regehr

regehr — Fri, 07 Oct 2011 22:07:09 +0000

Thanks David, I’m not a C# programmer and hadn’t known about this.

It’s insane that any compiler would permit silent overflow in new[]. Probably not insane from a language lawyer point of view, but insane in a practical sense.