paulsmith
paulsmith on Hacker News
This book is fantastic. Highly recommended if you have any interest in how the stored program computer came to be. The Dyson name apparently led to unrivaled access to first-hand accounts.
Posted by
paulsmith on Thursday October 01, 2026 @03:24PM
from the earendil.com dept.
Posted by
paulsmith on Tuesday March 10, 2026 @04:03PM
from the 2389.ai dept.
Alternately, hold the power button and either volume button together for a few seconds.
I like this approach. Something related I've been tinkering with are "protected bookmarks" - you declare what bookmarks (main, etc) are protected in your config.toml and the normal `jj bookmark` commands that change the bookmark pointer will fail, unless you pass a flag. So in your local "CI" script you can do `jj bookmark set main -r@ --allow-protected` iff the tests/lints pass. Pairs well with workspaces and something that runs a local CI (like a watcher/other automated process).
I haven't yet submitted it to upstream for design discussion, but I pushed up my branch[1]. You can also declare a revset that the target revision must match, for extra belts and suspenders (eg., '~conflicts()')
[1] https://github.com/paulsmith/jj/tree/protected-bookmarks
I'm glad you shared this because, in my head this Dr. Nick quote is so canonical it must be from the original golden 8 seasons, so it's nice to be reminded there are occasionally good things after! ;^)
lol I came here to post this too - perfect
As an OCaml-curious, is this the community recommendation, to choose the Jane Street stdlib if you’re just getting started?
Posted by
paulsmith on Wednesday October 01, 2025 @09:48AM
from the 2389.ai dept.
> Unfortunately, they removed Plan mode
If I hit shift-Tab twice I can still get to plan mode
In that case, each county that corresponds to that ZIP Code is shown and the user can disambiguate manually.
For example, health care plans in the US are county-specific with regard to premiums, co-pays, etc. (based on demographics). Allowing someone to type in their ZIP Code to get started can be a better user experience than having them pick their county.
https://www.healthcare.gov/see-plans/
One thing I just recalled is that if you maintain a small exceptions lookup table (i.e. the ones that span state boundaries), you can use ZIP Codes as a way to uniquely look up a county name.
This is a fact of ZIP Codes that a lot of people stumble one. I've worked on GIS/mapping projects in the past where stakeholders wanted or assumed ZIP Codes to be polygons.
Another complexity that surprises folks is you can't guarantee a one-to-many state-to-ZIP Code relationship. There are several (I forgot offhand how many, I used to have them memorized) that span across state boundaries.
Similar timing, I had a high school internship at the National Cancer Institute at Ft. Detrick, MD in 1994-95, and the lab down the hall had some SGI iron and a glove (I don't remember what the glove hardware was, if it was SGI or 3rd-party or custom) for manipulating 3D renders of folded proteins. Incredible stuff, same "in the future" feeling.
Aside from the Liquid Glass stuff, has anyone detailed the changes to the Unix bits of the OS? What's new, deprecated, moved, locked-down, etc. ... ?
Another good rule of thumb to remember is that a 50mm lens on a 35mm sensor ("full-frame") is roughly the equivalent FOV of the human eye, i.e., what you see naturally.
It's a great point, and I did consider it, the trouble is, how do you get the pattern resource data out of ResEdit running in the emulator and onto the modern machine? And ResEdit doesn't seem to run in any kind of compatibility mode on modern Macs anymore either.
It's too bad because ResEdit is an amazing program, and even has a surprisingly full-featured graphical editor, including for those patterns, with a live preview mode:
https://raw.githubusercontent.com/paulsmith/classic-mac-patt...
OK, wallpapers are on there now - tap/click on the virtual display after picking a pattern.
Exactly. One of the cool things about doing this the hard way was discovering that Apple still hosts old system and programmers manuals like the one for QuickDraw on its website.
Interesting - what's the provenance of this? Is this a leaked copy of the System source, or a disassembly/decompile?
> My favourite is the wide weave one (#15)
Same, that and the bubble one #36 have so much personality.
Yeah this is what keeps me from considering old PCs for NAS.
Maybe stating the blindingly obvious but seems like there is a gap in the market for a board or full kit with a high efficiency ~1-10W CPU and a bunch of SATA and PCIe ports.
> say model returns the probability of the token "cat" on the second position as p_2("cat") = 0.3, while p_2("dog") = 0.6. We may want to replace "cat" with dog, and use it in the subsequent iterations.
Might one tradeoff of speed/quality be a tree "search" for better outcomes by branching on logit choices? If a diffusion model is so much faster overall than AR, then I might not mind that I hunt or backtrack for the best probabilities overall.
Posted by
paulsmith on Tuesday April 08, 2025 @02:56PM
from the show-hn dept.
https://github.com/paulsmith/tailscale-mcp-server
Done mostly as an exercise to familiarize myself with MCP, but might be useful to others. Also might be a bad idea to let an LLM near your tailnet! Since we are no closer to cracking the in-band signaling problem of prompt injection. Even though passing through the eye of a RPC needle, UAYOR