01:46 <KipIngram> remexre: How involved was the process of getting your QEMU set up so you could target it?

01:48 <remexre> not too bad; overengineered script for starting it is https://git.sr.ht/~remexre/stahl/tree/main/item/run.py#L74

01:49 <KipIngram> That doesn't look too bad.

01:50 <KipIngram> I'm contemplatinng a computer buy. If I spring for it setting up something along those lines is of interest to me.

01:50 <remexre> oh, this isn't doing hardware virtualization

01:50 <KipIngram> Oh, ok.

01:50 <remexre> but you'd just add "-accel","kvm" to cmd to get it to do so

01:50 <remexre> and it looks about the same to the guest

01:51 <remexre> oh, and you might need to switch -cpu cortex-a53 to -cpu host

01:52 <KipIngram> I'd still continue the work I'm already doing on the other end of things, but it would be neat to get to start pecking at the bare metal side of things too.

01:52 <KipIngram> And to set up an ARM target so I could test that as well.

01:52 <remexre> but all the peripherals and such look the same, whether KVM is being used or not

01:53 <remexre> yeah, if you end up getting a pinebook pro or a rockpro64 (the two target devices I have), I wrote up https://remexre.xyz/pbp-osdev/hello-world/ as a very basic intro to just getting something that boots

02:02 <KipIngram> Cool. I may decide not to drop the money - the nice ones aren't exactly cheap.

02:02 <KipIngram> It would offer a number of convenience factors though - I'd suddenly again be able to do all the stuff my work has shut down.

02:03 <KipIngram> I think I mentioned it - they've turned off writing to the usb ports.

02:03 <KipIngram> Except they failed to turn off DVD burning, so they really accomplished nothing.

02:28 <remexre> oof

02:54 <DKordic> https://youtu.be/pQHX-SjgQvQ

10:04 <KipIngram> So, a few weeks ago someone mentioned inlining code even on a direct or indirect threaded machine. That looks like it would be pretty easy. When you arrive at the CFA of the inline code word, IP will be pointing at the start of the code, so you just need to do jmp <ipreg> to start it. Then that code does whatever it does, adds some cell size multiple to IP to jump it past the code, and runs NEXT. Should

10:04 <KipIngram> work really simply.

10:05 <KipIngram> The jmp <ipreg> and the add <ipreg> <val> are overhead, though - you wouldn't have those if it were a regular primitive.

10:05 <KipIngram> So I'm not sure you really gain anything other than a bit of code anonymity.

10:06 <KipIngram> Well, maybe the jmp <ipreg> isn't overhead.

10:06 <KipIngram> Or maybe it is. :-) Need caffeine.

10:07 <KipIngram> Anyway, it's cute.

10:17 <KipIngram> Ok, on reflecting on it I do think the jmp <ipreg> is overhead. So this adds two instructions to your machine code. Depending on what the code was doing, there might be some compensating savings, but that wouldn't be guaranteed.

10:22 <KipIngram> If you could control the NEXT part of the word just before this, you could have it jump straight to this code; that would save the jmp <ipreg>. But that would be convoluted as heck to arrange.

10:23 <KipIngram> Same kind of problem there as trying to mix sizes - you could switch around amongst 16 bit, 32 bit, and 64 bit sizes all you wanted - but only if the word BEFORE the new size knows it needs to do that.

10:24 <KipIngram> I've thought about something like that at times, and always wind up coming up dry.

10:25 <KipIngram> It's interesting to note, though, that you COULD do something like this:

10:25 <KipIngram> 32bitA 32bitB 32bitC 32->16 16bitA 16bitB 16->32 32bitD ...

10:26 <KipIngram> If you're willing to inject those "size change" words, that can work.

11:19 <KipIngram> You know, it occurs to me that the inline code thing might make a lot more sense for producing inline versions of several straight primitives. You'd avoid all the extra NEXTs. That actually may be worth considering. Just recognize opportunities for that automatically.

11:30 <KipIngram> Did you notice the pdf I posted a few weeks ago on cuckoo hashing?

11:30 <KipIngram> A CS friend of mine brought that to my attention, and it was interesting that he did so just before I got on the Lisp bandwagon. I think cuckoo hasing would be a really good way to handle symbol storage for Lisp.

11:31 <KipIngram> It's quite clever how it arranges to get high utilization density in the hash table.

11:33 <KipIngram> Oh sorry guys - thought I was in a private channel. :-|

13:13 <cmtptr> please take it back

13:54 <KipIngram> :-)

18:25 <KipIngram> Well, just sprung for a new notebook.

18:26 <KipIngram> First computer I've bought for MYSELF in nearly 9 years.

18:26 <KipIngram> To busy buying them for daughters to be able to think about myself.

20:54 <dave0> maw

22:29 <KipIngram> Howdy dave0.

23:18 <KipIngram> So, calculator groupies - anyone know if the TI nSpire CX is an interesting tinkering target?

23:25 <KipIngram> Ah, I see this:

23:25 <KipIngram> https://hackspire.org/index.php?title=Main_Page

23:59 <dave0> maw