Re: table library changes (was Re: table.new in 5.3?)

[Sean Conner <sean@...>]

It was thus said that the Great Coda Highland once stated:
> On Mon, Nov 25, 2013 at 7:03 PM, Sean Conner <sean@conman.org> wrote:
> > It was thus said that the Great John Hind once stated:
> > 
> >> Clearly, the optimised algorithm could be written in C as a multi-value
> >> remove library function, and like any algorithm it will be a faster in C
> >> than in Lua. But I do not think it will be sufficiently faster to justify
> >> its place in a C library.
> >
> >   I actually went to the trouble to measure the code and indeed, the
> > "optimized" version does indeed run faster (about 10x), *BUT* it does *NOT*
> > produce identical results.
> >
> >   I realize you were trying to prove a point here, but it might be made a
> > bit better if both code fragments actually did identical work.
> 
> What's the difference? 

The size of the resulting arrays (original: 9909 "optimized": 9899) and 90
incorrect values:

index	orig	optmized
-------------------------
10      11      10
11      13      102
12      15      103
13      17      104
14      19      105
15      21      106
16      23      107
17      25      108
18      27      109
19      29      110
20      31      111
21      33      112
22      35      113
23      37      114
24      39      115
25      41      116
26      43      117
27      45      118
28      47      119
29      49      120
30      51      121
31      53      122

	... and so on

> And would it help to make the final loop
> "table.remove(t, i)" instead of "t[i] = nil"?

  Nope.  Makes it worse.

  -spc