[ back to Tom 7 Radar ]

p
e
r
s
o
n
a
l
June 2024 (30 Jun at 21:56)
As predicted, I posted my new video called Badness 0. If you are waiting for me to post the video here, you are a poor detective! I think it's best without spoilers, which is why the title and thumbnail have very little useful information in them, although the thumbnail is intended to be an attractive nuisance for people with eyes/brains like mine:

Badness O
Badness O


Of course the whole system is set up to punish you if you don't have a clickbait title with a human face making an astonished look and an arrow at something in the video with some bold text that says "OVERSTATED PREMISE!" So if you did like it, please share the old fashioned way. Someone plausibly told me that they showed it to Knuth, which I choose to believe is true.

The video was not really that much more work than usual (although I am practicing new techniques and using new equipment) but since it came after several rounds of crunch on the same project (papers, SIGBOVIK talk, NYC talk) I did rather feel like a break. But a few bugs and unfinished things in BoVeX were haunting me, so I got sucked back into programming this month, although at a relaxed pace. I fixed the bug reported in previous comments where the files wouldn't render on some platforms, or would have really screwed up kerning. I added compression, so that you can't see my unoptimized PDF code if you open it in a text editor. I fixed the bugs with mutually recursive function bundles that capture polymorphic variables. I fixed the bug where the internal bytecode routine that creates nodes always flattens them into normal form, which fixes the weird spaces in some situations. I got it compiling in recent GCCs. I think I'm successfully done hacking on it for now, although I suspect I will use BoVeX for future papers, so I will be back!

Other things: I let myself play video games and promptly got addicted to Hades, which is excellent, as you probably know. I have "beaten" it but I'm not really finished; a more detailed review in a future post.

Right now we are working on the 2024 ICFP Programming Contest so I should probably get back to that. I am enjoying this one, not only because it continues the story of the 2006 contest that I led (it still works! and is also best without spoilers!). I'll write about this too, but there's still 10 hours left in the contest so fair play forbids me from showing any images until after. (Not that we are doing great or anything, but we're pretty close to solving every puzzle, at least!) So I should probably get back to that!
c
o
m
m
e
n
t
mixmix (122.135.13.133) – 07.01.24 00:36:02
I once again encourage you to make a technical, supplementary video on BoVeX. The world needs a 2 hour crash course on compilers by Tom 7.
c
o
m
m
e
n
t
Anonymous (179.217.229.149) – 07.01.24 13:44:00
I highly agree with mixmix, thank you for bringing this issue to attention in the Tom 7 Radar comment section
c
o
m
m
e
n
t
david (180.150.25.222) – 09.29.24 18:58:36
Hi Tom. Agree with mixmix. Also would be really worthwhile to just note down the git revision of llama.cpp and any tips on which llvm version works with the weird json implementation they have?
c
o
m
m
e
n
t
Tom 7 (154.47.25.137) – 09.30.24 23:32:24
Sorry about llama not being checked in. I did make a few light modifications to that code, too. The problem is that their codebase is changing super fast (makes sense) and I've been bit before by API changes or behavior changes when syncing. So I don't want to try to update until I have a solid chunk of time to do it. But my plan is to fork a copy (I'm only using the core library which isn't really that big) for BoVeX so that it doesn't have an external dependency you need to track (and so that I can check in my modifications).
p
o
s
t

a

c
o
m
m
e
n
t
Please enter the code, unless you are spammer!!
[ Tom 7 Radar  •  Tom 7 on Google+  •  on Twitter  •  on Facebook ]