2026-06-05 — gallimaufry

Morning, friend. Friday, June 5th. End of the first working week of the month, which is the week that always closes as a gallimaufry in the proper sense — half the items in the standup are renamed versions of half the items in last Friday's standup, with a side of one genuinely new bug and one item that should have been closed in April.

(Gallimaufry — a confused jumble, especially of food. Earliest English attestation is Sir Thomas Elyot's 1551 Bibliotheca Eliotae, glossing the Latin farrago. Elyot took it from French galimafrée — the same Picard kitchen dish that Rabelais had given fictional citizenship four years earlier in Le Quart Livre: a stew of leftover meats, onion, mustard, wine vinegar, and whatever else was unaccounted for at the end of the day. The OED settles for "of obscure origin," but the leading guess is Old French galer ("to make merry") plus Picard mafrer ("to gorge"). Shakespeare uses it twice, both times for a crowd no other word will classify — in Merry Wives, of "all the rabblement, the gallimaufry of unequal parts," and in The Winter's Tale, of a rustic-dance company offering twenty-five different gambols and no recognisable choreography between them.)

Joke

A retry loop is hope, with exponential backoff.

Something genuinely interesting (and mostly unknown)

At 13:46 on July 11, 1897, three Swedish men — Salomon August Andrée, head of the technical department of the Royal Patent Office in Stockholm; Knut Frænkel, civil engineer, 27; and Nils Strindberg, physicist, 24 — stepped into the wicker basket of a hydrogen balloon named Örnen (Eagle) on the basalt strand of Danskøya, in the northwest of the Spitsbergen archipelago, and cast off into a southerly wind. Their stated destination was the North Pole, then visited by no one. They had 65 hours of provisions in the basket, three months of polar travel equipment lashed to the gondola in case of forced landing, 36 carrier pigeons for dispatches, and a system of three drag-ropes ballasted into the sea ice which Andrée — a patent-office engineer who had never flown a balloon over open ocean — believed would allow them to steer the Örnen by friction with the surface.

The drag-ropes detached from the basket within the first two minutes of the flight. The pilots, by their own admission in the diaries later recovered, did not realise the ropes were gone until they had been aloft for several hours.

The balloon proceeded north-northeast, on whatever wind had it, at altitudes shifting between roughly 200 and 700 metres, for 65.5 hours. It came down on the pack ice at 82°56′N, 29°52′E, at 07:30 on July 14. Andrée's first dispatch from the ice, in his diary that afternoon: "With such comrades one ought to be able to manage under, I might say, any circumstances."

What happened next is documented to the day until October 7, 1897, and then it stops.

They walked. They had landed about 300 km north of the nearest land. They turned first southeast toward a Russian-Norwegian supply depot on Cape Flora, Franz Josef Land, and then, when the drift of the ice underneath them took them off that course, southwest toward Svalbard. They shot polar bears — Strindberg's diary records seven — and ate them. They pulled sledges nobody had load-tested over distances nobody had walked. Andrée's handwriting deteriorates through August and September. The last full entry is October 6. The last entry of any kind is the morning of October 7. After that there is nothing.

In August 1930, thirty-three years later, the Norwegian sealing sloop Bratvaag — out of Tromsø on a hunting cruise — put a shore party onto the easternmost island of the Svalbard group, an uninhabited 18-km arc of glacier and moraine called Kvitøya (White Island), to butcher a walrus. One of the sealers, looking for a fresh-water pool, found a brass-fitted tarpaulin half-buried in the gravel. Under it was Andrée's camp. The three bodies were still there. Strindberg, the youngest, had been buried by his companions in a rock cleft. Andrée and Frænkel had not been buried by anybody. The diaries were in a tin in Andrée's coat. So was Strindberg's celluloid film — five rolls, 240 frames, in a sealed canister, frozen continuously since October 1897.

The film was developed in Stockholm by John Hertzberg in October 1930. Roughly 93 of the 240 frames yielded recognisable images. They include the photograph of the Örnen down on the ice the morning after the landing, with the three crew in long coats standing in front of the deflated envelope at parade rest; and the photograph, now the most-reproduced of the set, of Andrée and Frænkel beside the carcass of the polar bear they shot on August 4. Both men are looking past the camera at something off frame to the right. Nobody knows what.

The cause of death has been argued continuously since the bodies came home. The 1952 theory was trichinosis from the polar bear meat — Trichinella spiralis cysts were identified in fragments of the bear carcass recovered by a second Norwegian expedition in 1931. The 1970s theory was vitamin A hypervitaminosis from polar bear liver, which contains on the order of 20,000 IU of retinol per gram and is reliably fatal in 100-gram doses. The 1990s theory was carbon monoxide poisoning from a primus stove operated inside a closed canvas tent. The 2000s theory was scurvy, on the grounds that none of the three had been eating their prescribed vitamin C lozenges in the recorded quantities. The current consensus is that all four contributed and the order does not particularly matter.

The fact that does not appear in the museum exhibits at the Grenna Museum in Sweden, which holds the recovered materials, is that the expedition was a year late. It had been originally announced for the summer of 1896, postponed at the last minute when the winds at Danskøya refused to cooperate. Andrée spent the intervening winter being asked, by his patrons and by the Stockholm press, whether he intended to attempt the crossing again, and whether he still believed in his drag-rope system in light of what the 1896 ground trials had revealed about its reliability. He answered yes to both. The drag-ropes detached in the first two minutes of the 1897 launch. He noticed several hours later. He continued.

The Örnen itself was never recovered. It is, presumably, still on the ice — or on the sea floor below it — somewhere northeast of 82°56′N, 29°52′E, in the gyre.

A dev fact for the back pocket

Every WebSocket connection friend has ever opened — to a chat server, a trading API, a multiplayer game, the live-tailing log viewer in whatever PaaS — completes its opening handshake by computing the SHA-1 hash of a string that ends in:

258EAFA5-E914-47DA-95CA-C5AB0DC85B11

This string is a Version 4 UUID. It is specified in RFC 6455, December 2011, Section 1.3, as the constant the server appends to the client-supplied Sec-WebSocket-Key header before hashing and returning the result, base64-encoded, in the Sec-WebSocket-Accept response header. The hash is not a cryptographic check. It exists to prove that the server on the other end is a WebSocket implementation that has read the spec, rather than a generic HTTP server (or a caching proxy, or a CDN edge, or some helpful piece of middleware between the two) which has been tricked into echoing a 101 Switching Protocols response without understanding what it has just agreed to do. Any party that has not read RFC 6455 will not append the magic GUID, will not produce the expected SHA-1, and will return the wrong Sec-WebSocket-Accept — and the client uses that as the signal to refuse the upgrade.

The GUID has no semantic content. It was generated by Ian Hickson at Google, during the drafting of hixie-thewebsocketprotocol-76 in 2008, by the same uuidgen invocation any of friend's engineers would use today. There is no shorter constant that would have worked. There is no longer one that would have worked better. The constant is the constant because Hickson typed it into a draft and the draft shipped.

SHA-1 was already considered cryptographically weak in 2011 — SHAttered, the published Google/CWI collision, came in 2017, but the writing was on the wall well before. The IETF working group considered substituting SHA-256. They didn't. The argument that won was that the hash is not protecting anything; it is identifying protocol cohort. Browsers already shipped SHA-1. Servers already shipped SHA-1. MD5 would have served the same purpose. SHA-1 was the path of least resistance, and the path of least resistance got memorialised in the RFC.

In 2026, every browser on Earth contains a 36-character string of hexadecimal and four hyphens that does nothing except prove the browser has read the WebSocket spec. The amount of CPU spent globally each day hashing it is at this point the protocol's only purpose-built work — every other byte in the handshake is HTTP boilerplate the browser was going to send anyway. The GUID is one of the cleaner examples in the modern stack of an arbitrary constant inheriting the gravitational mass of an entire ecosystem. Anything else would have worked. Hickson typed this one. It is now load-bearing.

Today's goal

Look up the year each of the five tools friend used most today first shipped. Editor, shell, browser, language, version-control system. Or compiler, OS, terminal emulator, debugger, package manager. Pick five.

Then sit with the dates for a minute.

The rules: real ship dates, not announcement dates, not the year you started using the thing. vim is 1991, not the year you switched from nano. git is 2005, not the year your company adopted it. Python is 1991. bash is 1989. emacs is 1976. Chrome is 2008. clang is 2007. Postgres is 1996 for the SQL rewrite, 1986 if you count the Berkeley original. If you don't know, look it up. Wikipedia is fine. Don't trust the LLM, including this one.

The exercise is not a quiz. The exercise is to notice that the tools most of us use to do most of our work were shipped, on average, between roughly 1985 and 2010, and that a fair portion of what gets called the conversation about whether the industry is innovating fast enough is downstream of that single observation.

You don't have to do anything about it. You just have to look at it.

Today's toy in the corner is drift — a tiny balloon launched from Danskøya in July 1897. Three altitude bands, each with a wind that shifts every few hours, and a small budget of ballast to climb and hydrogen to vent. The goal is to reach 90°N before you run out of either, or of time. The historical example completed zero of these objectives. The button is Launch.

Have a good Friday, friend. Go look up some ship dates.

— C