Linus Torvalds' good taste argument for linked lists, explained

Re:Linus Torvalds' good taste argument for linked lists, exp... by erik_seaberg on Sunday December 06, 2020 @05:28PM (#25327240)

Lines of code are expensive, because we expect human beings to reread them with almost no help from tools. Concise code is a temporary problem for novices, until they grow into experts. Bulky code is an ongoing problem for everyone and we don’t have a solution.

[ Parent | on HN ]

Re:Linus Torvalds' good taste argument for linked lists, exp... by Taniwha on Sunday December 06, 2020 @05:46PM (#25327393)

While I mostly agree with you, I think that use of double indirect pointers is rare enough that in effect each use of one probably counts as a "line of code" when trying to understand what's going on.

The one Linus doesn't like likely runs faster (at a microarchitectural level), it's also the thing I've done for 40 years now, it's what comes out of my fingers when I code linked lists, for me at least it's more understandable because I already understand it

[ Parent | on HN ]

Re:Linus Torvalds' good taste argument for linked lists, exp... by varjag on Sunday December 06, 2020 @05:59PM (#25327489)

It's very unlikely it is faster.

[ Parent | on HN ]

Re:Linus Torvalds' good taste argument for linked lists, exp... by erik_seaberg on Sunday December 06, 2020 @07:18PM (#25327990)

A good compiler should keep *p in a register because there are no aliases that might modify it, but a poor compiler might follow p a second time.

[ Parent | on HN ]