News for hackers. Stuff that matters.

A comment on Making robust software: the “Mars Code” talk (2012). Parent comment

Re:Making robust software: the “Mars Code” talk (2012) by pieceof314 on Monday February 15, 2016 @10:25AM (#11103809)
Loops. A recursive algorithm can be transformed into an iterative looping algorithm that uses a fixed stack of memory.

There's probably no change in the algorithms. They just use non-recursive versions.

[ Parent | on HN ]
Re:Making robust software: the “Mars Code” talk (2012) by marcosdumay on Monday February 15, 2016 @11:01AM (#11104013)
What's the gain? If you turn your recursive algorithm into an iterative one with a stack, you are sill subject to all the same problems.

They probably don't use trees and quicksort at all.

[ Parent | on HN ]
Re:Making robust software: the “Mars Code” talk (2012) by ktRolster on Monday February 15, 2016 @11:14AM (#11104104) 

  > What's the gain?
You can prove that you won't have a stack overflow. You can calculate the maximum depth of the stack at compile time, not runtime.
[ Parent | on HN ]
Re:Making robust software: the “Mars Code” talk (2012) by marcosdumay on Monday February 15, 2016 @03:14PM (#11105609)
Well, you can either prove that you won't have a stack overflow at the original program, or you can not guarantee that your new code will run out of stack space.

If you can calculate the size of the stack at compile time, you don't need to manually manage it.

[ Parent | on HN ]
Re:Making robust software: the “Mars Code” talk (2012) by dclowd9901 on Monday February 15, 2016 @11:23AM (#11104159)
Can't you also just pass a bit through the recursion that contains its depth, and halt the recursion if that number gets too high?
[ Parent | on HN ]
Re:Making robust software: the “Mars Code” talk (2012) by ktRolster on Monday February 15, 2016 @11:30AM (#11104199)
Yes, but that depends on making sure you get the programming logic correct, whereas with banning recursion outright, it can be tested automatically.
[ Parent | on HN ]