Re^2: Yet Another Sort Benchmarking Question

I don't know what gave you that idea.

What gave me the idea is are statements (probably misconstrued by me) in Guttman and Rosler's paper on sort optimization asserting or implying that Perl's default lexicographic sort is faster than any sort that uses a user-specified comparison sub.

As an aside, note that perl specifically optimizes {$a cmp $b} and {$a <=> $b}, and doesn't actually call the perl code for each pair comparison

This answers my question. It suggests that ~~some of the claims in the cited paper are no longer valid~~ there are a few special cases for which the assertion in that paper that the default lexicographic sort is faster than any sort using a user-supplied sortsub no longer holds. Thanks.

Update: I realized, after reading dave_the_m's and tilly's replies, that, in its original form, the second sentence of my last paragraph may have given the impression that the paper I cited is somehow out-of-date. That was not at all my intent; I've tried to edit the problematic sentence to avoid this false impression.

the lowliest monk

Comment on Re^2: Yet Another Sort Benchmarking Question

Replies are listed 'Best First'.
Re^3: Yet Another Sort Benchmarking Question by dave_the_m (Monsignor) on May 08, 2005 at 15:16 UTC
This answers my question. It suggests that some of the claims in the cited paper are no longer valid The basic message of the paper is still true though. Dave.	[reply]
Re^3: Yet Another Sort Benchmarking Question by tilly (Archbishop) on May 08, 2005 at 19:43 UTC
That claim remains true for any sort operation that has not been special cased inside of perl. The number of exceptions to their claim can currently be counted on one hand.	[reply]