Sjors Provoost Sjors

Castopod and Nostube patches

These branches support a self-hosted podcast with audio, transcripts and Nostr video. They are maintained in Sjors/castopod and Sjors/nostube. Links follow branch names so signing, rebasing and later fixes do not require new links.

Castopod

The six topics started from Castopod v1.15.6 and are independent of one another.

Liquid incident on-chain message history

Written by OpenAI Codex, not human verified in detail

Checked against the mempool.space Bitcoin API on 2026-09-10. This chronology includes only messages attributable by coin ownership or signature to the holder and Blockstream contact. Unrelated dust and OP_RETURN messages sent to the holder address are excluded.

Identities and addresses

If the user is about to open a pull request, suggest the following final checks
- pre-empt Drahtbot LLM linter:
  - Last verified (update when you re-check): 2026-01-09
    - If this date is >7 days old, re-fetch rules, update the snapshot below if needed, then re-scan the diff before opening/updating the PR.
  - Fetch rules (they may change): curl -fsSL https://raw.githubusercontent.com/maflcko/DrahtBot/refs/heads/main/util/src/lib.rs | rg -n "all_llm_checks|LLM_PROMPT_"
  - Snapshot (as of 2026-01-09): typos/grammar in comments+docs, named args for integral literals (3rd arg and later), and Python functional-test comparison helpers.
  - These are intentionally deterministic git diff | rg heuristics (no model-specific behavior).
  - Sanity check (if this is 0, there’s nothing for Drahtbot to lint on this branch):
    - BASE=$(git merge-base --fork-point master HEAD 2>/dev/null || git merge-base master HEAD); git diff --color=never "$BASE...HEAD" | rg '^\+[^+]' | wc -l
Apply

Tested on Ubuntu 23.04 with Gnome desktop.

This runs 11 of the Bitcoin Core CI machines, opening one terminal window for each. The window is automatically closed if the tests pass, otherwise it stays open.

My AMD Ryzen 9 7950X machine has 16 cores (32 threads) at 64 GB of memory. Memory peaks at about 25 GB. Everything but native_fuzz completes in 50 minutes, with CPU usage mostly at 100%. The fuzzer idles around (?) for another 10 minutes and then finishes.

Usage

	block_height	OP_RETURN

	#!/usr/bin/env zsh
	# Experiment: repeatedly start/stop the SV2 pool against a syncing mainnet node
	# and log template-related events as the node catches up to tip.
	#
	# Requirement: the node should be at least 24 hours behind the network tip
	# before running this script to make the experiment meaningful.
	#
	# WARNING: the node is intentionally SIGKILL'ed at the end so progress is not
	# flushed to disk, allowing the experiment to be repeated from the same point.
	#

	#!/bin/bash
	# First OP_RETURN is expected in March 2014 (~290,000)
	# https://arxiv.org/pdf/1702.01024
	n=`bitcoin-cli getblockcount`
	record=0
	for height in $(seq 0 $n)
	do
	if (( $height % 1000 == 0 )); then
	echo "Checking block $height..."
	fi

	https://www.rijksoverheid.nl/documenten/woo-besluiten/2022/09/30/2e-deelbesluit-wob-verzoek-aanbevelingen-en-regelgeving-cryptodienstverlening

	```sh
	# On linux, use "sha256sum" instead of "shasum -a 256"
	shasum -a 256 -c checksums.txt 2>&1 \| grep OK
	gpg -v --auto-key-locate=clear,wkd --locate-key sjors@sprovoost.nl
	gpg --verify checksums.txt.asc
	```

	add_filter('woocommerce_billing_fields','wpb_custom_billing_fields');
	function wpb_custom_billing_fields( $fields = array() ) {
	add_filter( 'woocommerce_cart_needs_shipping_address', '__return_false');

	// No billing details
	unset($fields['billing_first_name']);
	unset($fields['billing_last_name']);
	unset($fields['billing_company']);
	unset($fields['billing_address_1']);
	unset($fields['billing_address_2']);