|
From: | Dmitry Gutov |
Subject: | Re: master 4b79c80c999 1/2: New function 'sort-on' |
Date: | Thu, 7 Mar 2024 03:34:12 +0200 |
User-agent: | Mozilla Thunderbird |
On 06/03/2024 22:12, John Wiegley wrote:
Dmitry Gutov<dmitry@gutov.dev> writes:Here's a version from Daniel as well:This version from Daniel looks pretty optimal to me. It trades building a whole new list spine that is sent to GC, for an extra O(n) traversal to “restore” each of the original list cells. Done in C, I have a feeling that would beat the GC.
Speed-wise, IIRC this version already beats the committed implementation.I'm talking about moving to C in the context of merging this feature with the basic 'sort' function. Because it really looks more fitting for a new optional argument now, rather than a separate function.
[Prev in Thread] | Current Thread | [Next in Thread] |