Omar Rizwan (@omar@social.omar.website)

Omar Rizwan @omar

(I'm sure people who've actually used CAD software know this:)

CAD software is fascinating honestly, it's basically
programming

You replied to yourself

CAD software is fascinating honestly, it's basically
programming

it has 'parametric history’, like if you make a hole
in an object and then copy and paste it so there
are 4 holes, you can then select any of those
holes later and it'll filter the undo history to only
operations that affect that hole

and you can go back to the original operation
where you created the original hole and adjust the
size, make it bigger or w/e, and the new size will
propagate through the history and change in all
the copies also

Open thread

Omar Rizwan @omar

one of the subtleties in this area is I found that in Folk you generally want semantics that are more like a condition variable or signal than like a blocking channel -- you raise it, they get notified if they're waiting on it, but they don't queue up if you raise it multiple times in a row
we initially used various pipe/channel/message stuff (both Tcl and Unix/TCP/websocket), but I found that I ended up spending all my time trying to find weird blockages and latencies that emerged from that
shared memory provides a much more reasonable default that matches the Folk statement/database semantics
where the reader can miss intermediate states and it's ok

Open thread

Omar Rizwan @omar

my current understanding is that both clang and gcc basically support nested functions / closures but with different and incompatible syntax and implementation… wonder if people ever just ifdef between them

Open thread

Omar Rizwan @omar

Have ppl ever used `git notes`...

Open thread

Omar Rizwan @omar

tiny usability nicety that I like (the green)

screenshot of Folk calibration board pdf embedded in web page with a bunch of fiducial markers on it and the inner perimeter of one marker marked with a green line, then an HTML form underneath telling you to enter the "inner side length" (in green) in millimeters

Open thread

Omar Rizwan @omar

initialization is often a place where you want to break the 'normal rules' (of concurrency, process management, error handling), just write everything imperative style and not have a bunch of callbacks/indentation/try-catch/synchronization

Open thread

Omar Rizwan @omar

"how do you represent phenomena that cannot be measured by the human eye?" (https://www.frieze.com/article/claire-lehmann-lucy-ives-interview-2021) https://social.omar.website/@omar/statuses/01GSSDXRFXWKTSDD359MYQW7AG

LI: You also take the side effects of contemporary image-producing technologies and make us look straight at them. CL: Yes! The touchstone for Model for Normals [2020], for example, is a file from an imaging experiment where scientists measure vectors of light bouncing off various surfaces, which are then used to calculate reflectance in CGI engines. This visualisation fulfils a purely bureaucratic need, but I’m offering a glimpse of this hidden process. These by-products from the contemporary imaging industry are, for me, also redolent of some of the same issues that presented themselves in painting centuries prior: how do you represent phenomena that cannot be measured by the human eye? I think, for example, of how a divine visitation could only be gestured at through beaming vectors of gold leaf. LI: Speaking of the unrepresentable, the last time I was looking at your work I noticed that medical technology plays a certain role. I hadn’t previously thought about how much imaging goes into contemporary medicine.

Open thread

Omar Rizwan @omar

my aesthetic problem with TypeScript was always that it doesn't actually let you do anything new, almost by definition -- it lets you do increasingly specific correctness checks (if you can figure out how to encode them), but your program will never get any shorter or more expressive

Open thread

Omar Rizwan @omar

hmm, I have a stereotypical association of both C and Lisp codebases with 'maintained by 1 person or a small tight-knit team' but for sort of opposite reasons -- C not expressive enough, can't maintain invariants around memory management, encapsulation, concurrency unless everyone has them in their head; Lisp too expressive, everyone has their own DSL, hard to read other people's code

Open thread

Omar Rizwan @omar

I do want to try this line of stuff https://github.com/CASIA-IVA-Lab/FastSAM / https://github.com/YavorGIvanov/sam.cpp but I don't want to be dependent on PyTorch and Python and CUDA and 1000 other things + I don't want to be dependent on an M-series Mac CPU either, I want to like run it on the GPU on a Pi 4 or 5 or on a $150 mini-PC

Open thread

social.omar.website

About

Recent posts