Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-

Static code analyses

Name: Anonymous 2013-12-28 4:57

What kinds of static code analyses has humanity implemented over the years? Which of them turned out to be the most practically useful?

We know that the common System F-ish type systems are tractable but not really useful for error detection.
Dependent types are more powerful but they seem to be too unwieldy for practice (nobody's really embracing them for programming despite the fact they've been around for decades).
Contracts are practical but they're just runtime asserts, no static analysis.
Perhaps, SMTs are the future? Have you ever used them for your projects?
Then there are variable lifetime analyses, region typing for safe storage management... there's so much stuff but nothing makes it to the industry. When Haskell's primitive System F-like type system is a beacon of static safety, something's clearly wrong about the state of the world. Or maybe it's the whole idea of static code analysis that's wrong and useless for real life?

Name: Anonymous 2013-12-28 14:27

p.s. install gentoo

Name: Anonymous 2013-12-29 0:27

>>1

It's not completely useless; in fact it's a particularly useful development tool; One of many. Use Smalltalk or Common Lisp for about 4 years or so to get a feel for other power tools :)

Name: Cudder !MhMRSATORI!fR8duoqGZdD/iE5 2013-12-29 0:48

Or maybe it's the whole idea of static code analysis that's wrong and useless for real life?
It's the academic idea of static analysis that is useless. None of the bullshite you've mentioned is useful, because what's useful is the stuff that actually works and gets to the heart of the problem you're trying to solve: simulated execution/dataflow tracking, variable range analysis, etc.

Name: Anonymous 2013-12-29 4:58

static dubs analysis

check 'em

Name: Anonymous 2013-12-29 5:03

I've got a static code analyser that solves the halting problem.
Keeping it closed source though.
Also taking a page from all the cold fusion guy's books I've decided to only let people see it in action after they pay me a few thousand dollars and sign a contract. No refunds.

Name: Anonymous 2013-12-29 8:34

>>6

Do you believe Douglas Lenat actually pulled this off?

Don't change these.
Name: Email:
Entire Thread Thread List