3 ms·
Bookmarked "toread" -- it looks to be a really insightful look at byte-code that compares and analyzes a bunch of language implementations. What an interesting
by haberman 16y ago
Bookmarked "toread" -- it looks to be a really insightful look at byte-code that compares and analyzes a bunch of language implementations.
What an interesting guy. He runs a blog masquerading as a mailing list (that's what this link is from: "Kragen-tol" is a mailing list for "Kragen thinking out loud") which looks to be full of lots of interesting and incredibly detailed essays across a wide range of subjects.
His homepage (http://www.canonical.org/~kragen/ http://www.canonical.org/~kragen/) doesn't mention the mailing list at all (?), but it does have a bunch of other intriguing and random things, like a list of all the people he knows, for the purpose of telling search engines about it: http://www.canonical.org/~kragen/pik/ http://www.canonical.org/~kragen/pik/
- Someone 16y agoI am not sure about that insightful. He complains that literals on the F21 take two instructions. The Forth way is to define constants you need often like this: : 1 1 ; (that denies a function called 1 that 'returns' the integer 1) With that function in place (3 instructions, I guess), you can replace n usages of the literal 1 (2n instructions) by n calls of that function (n instructions), for a total of n+3 instructions. So, this pays of if n>=3. Oh, and subroutine calls are cheap on Forth hardware (one cycle for a call and zero cycles for a return is easily achievable)