xpe

Helix: the Selection cannot be passed as an argument

Claim verified: In the Helix editor, no user-invocable command takes a Selection as an argument. Commands read the selection only as implicit state from the current document and view.

Result: TRUE.

---

xpe

history --color always delete reports success but deletes nothing

Environment

• fish version: 4.9.2
• OS: macOS (uname: Darwin 25.6.0)
• Terminal: TODO
• Reproduces without third-party customizations (sh -c 'env HOME=$(mktemp -d) XDG_CONFIG_HOME= XDG_DATA_DIRS= fish'): TODO

Summary

xpe

Resolving .jj/repo to an absolute path

Text written by Claude Opus 5. (I tested the scripts on macOS.)

Read the relative path stored in .jj/repo and resolve it against the current working directory.

Base directory: the path in .jj/repo is relative to the .jj directory, not to the CWD. All snippets below join against .jj/ for that reason.

---

xpe

Existential Risk from Superintelligence

Please sit down with a pen and paper for at least 15 minutes and consider each of the following claims to the best of your ability:

1. It is wiser and more ethical to not use AI to improve AI capabilities.
2. Given some degree of agency and ability, it is our responsibility to help improve ourselves, our families, and/or our local communities, as best we can.
3. The above point is a great start. If enough of us do this, we can help nudge the world in a better direction.
4. We want humans to better understand each other and to solve problems peacefully.
5. We want humans to make better decisions.
6. The current AI labs do not have their safety pratices under sufficient control. Anthropic might have a better safety track record at present, and this is good, but no current labs are taking safety seriously enough.

xpe

jj log -G -r 'all()' \
-T 'change_id.shortest() ++ "\n" ++ description.lines().map(|line| pad_start(6, line.len()) ++ "\t" ++ line).join("\n") ++ "\n\n"'

xpe

# Recommended for use in `.zshrc`

if (( $+commands[jj] )); then
    # Autocompletions for Jujutsu
    source <(COMPLETE=zsh jj)

    # Tell `gh` how to access the git-backend for a JJ repo.
    # (Works whether colocated or not.)
    github_jj() {
        GIT_DIR=$(command jj --ignore-working-copy git root) command gh "$@"

xpe

POSIX Signals

Signal	#	A	C	G	Key	Act	Norm
SIGCHLD	17	·	C	G	·	I	child died or stopped
SIGHUP	1	·	C	G	·	T	TTY hangup; or reload config (convention)
SIGINT	2	A	C	·	^C	T	cancel current op
SIGKILL	9	·	·	·	·	T	last resort; no cleanup
SIGPIPE	13	·	C	G	·	T	write to pipe with no reader
SIGQUIT	3	·	C	·	^\	D	debug halt (post-mortem)

xpe

Hello humans and computational agents,

Below the --- divider below you will find a section that I agree with but Claude Opus 4.7 drafted. I don't just use LLMs to tell me what I want to hear, and I don't casually dump their outputs into the public sphere. I gave the early drafts LOTS of pushback and guidance.

Why? I'm beyond tired of sloppy thinking. I'm tired of post-hoc rationalization. So much that I'm pretty much done with Hacker News and Lobsters. These are places where good reasoning goes to die, but too few admit it.

Enjoy! ... but first, I wrote my own prefatory rant for your consideration.

Just Say "Oops" and Fix The Fucking Thing

xpe

ME: Ok, let's simply diagnose the PROBLEM (with certainty) in jjui ... by reading the code. Don't say what the solution is yet.

---

Claude Opus 4.7 (running in the jjui repository)

Diagnosis with certainty, from reading the code:

1. The cursor in the inline description editor is a software-painted, blinking cell, not the terminal's hardware cursor.

xpe

Generated by Claude Opus 4.7 on April 22, 2026.

A map of privacy-preserving telemetry, from randomized response to OHTTP

When a vendor says "we don't collect any personally identifiable information," what did they actually build? The honest answer, across two decades of deployed systems, ranges from "nothing at all, they just stripped the obvious identifiers" to "a multi-party cryptographic protocol with published epsilon budgets and open-source code running in trusted enclaves." This survey is a field guide to the territory between those extremes — the deployed systems you will run into by name, and the formal guarantee types that those systems do or do not meet.

Three things make this landscape hard for a newcomer. First, the same marketing words — "anonymous," "aggregated," "privacy-preserving" — get used for mechanisms that differ by many orders of magnitude in strength. Second, the strongest guarantees are conditional on assumptions (non-collusion, trusted hardware, crypto hardness, client hones