The algorithm used in malloc.coh may be improved

[Laci1953]

It affects the performance of Cowfe ...

See https://github.com/Laci1953/Cowgol_on_CP_M/blob/main/Improved%20alloc/improved%20alloc.cow

I also write you a mail about this on 7 November, including a simple proof, comparing the alternatives...

[davidgiven]

Hello --- yes, I did see the email; thanks! Do you know what's different with the new algorithm? A quick look suggests it's best-fit rather than first-fit, which I think the old malloc was. This will reduce fragmentation at the expense of slower allocations and more code, but given the small systems I suspect it's worth it.

Nothing much has happened with cowgol for a while because there are two major issues which need dealing with: firstly, interfaces don't work properly in .coo files, and secondly, the assembler is all wrong and needs a complete replacement. But I have too many projects and not enough time...

[Laci1953]

Hi,Yes, i am aware of the advantages of this new allocation algorithm... But, are you aware how much you are loosing by using the old malloc with Cowfe, on Z80 machines running CP/M? A new programming language gets attention when the community of users (mostly retro computing...) is able to use-it on large-enough source files (mostly old games...). Ladislau Yahoo Mail: Căutare, organizare, reușită Pe joi, dec. 11, 2025 la 13:30, David ***@***.***> a scris: davidgiven left a comment (davidgiven/cowgol#171) Hello --- yes, I did see the email; thanks! Do you know what's different with the new algorithm? A quick look suggests it's best-fit rather than first-fit, which I think the old malloc was. This will reduce fragmentation at the expense of slower allocations and more code, but given the small systems I suspect it's worth it. Nothing much has happened with cowgol for a while because there are two major issues which need dealing with: firstly, interfaces don't work properly in .coo files, and secondly, the assembler is all wrong and needs a complete replacement. But I have too many projects and not enough time... — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>