v1.6.0

See CHANGELOG.md for the full release notes.

What's Changed

• deploy/blobs: what the first deploy read on Fly by @codetalcott in #351
• blobs: shapes keep their vertex order, and nothing pops by @codetalcott in #352
• The milestone report sees an issue the doc-fact ratchet marked by @codetalcott in #341
• host: a command line and a doctor (SPEC E30–E31, D37) by @codetalcott in #353
• blobs probe: a twist is judged by the kernel's own measure by @codetalcott in #354
• layer: htmx 4 is the built-in vocabulary (SPEC N22, D6 retired, D38) by @codetalcott in #355
• m0: the wheel and its template-free CLI (SPEC N23-N26, D39-D43) by @codetalcott in #356
• m0 new: two templates that compile where they lie (SPEC N27-N29, D44) by @codetalcott in #357
• m0 dev, m0 image, and the workflow that will publish m0 (SPEC N30-N32) by @codetalcott in #358
• The Mojo stack's pages, llms.txt, and smoke-quickstart-mojo (R4; SPEC N33-N35, D45) by @codetalcott in #359
• CI: two smoke jobs, no cap within reach of a green run, and nothing waiting on another job by @codetalcott in #360
• The nightly canary can speak, and one break no longer hides the rest by @codetalcott in #330
• sabotage-scaffold: the no-fragment arm keeps still in use by @codetalcott in #361
• m0 0.1.0 is published: the record, and the sentences that said otherwise by @codetalcott in #362
• scaffold AGENTS.md: plain-form login, a cookie in a test, where the worked example is by @codetalcott in #363
• m0-sqlite: the fork test names the signal that killed a child by @codetalcott in #364
• test_host: the producer tests wait on the step counter, not on a sleep by @codetalcott in #366
• accept sharing: the knob-off arm runs five rounds, not one by @codetalcott in #367
• Query: a query-string builder beside url_for (SPEC N36) by @codetalcott in #369
• push=True: a swap that moves the address bar (SPEC N37, D46) by @codetalcott in #370
• soak.py: an abandoner pauses between laps by @codetalcott in #368
• test-sqlite: OS_ACTIVITY_MODE=disable, and what the macOS fork crash is by @codetalcott in #371
• Datastar: pin v1.0.4, and D46's condition names a price not a release by @codetalcott in #372
• An early 413 reaches a client that uploads before it reads (SPEC A20) by @codetalcott in #374
• ASGI: judge each send by its connection; answer at the final body (SPEC L20–L24) by @codetalcott in #375
• An app's WebSocket close is one datagram, so its Close is never answered twice by @codetalcott in #376
• G16: a process the application starts inherits none of the server's descriptors by @codetalcott in #377
• The response head as the application sent it (A21, K12, L27) by @codetalcott in #378
• A request as its client sent it, and work scheduled at import refused by name (L25, L26) by @codetalcott in #379
• CI hygiene: conflict markers, the container copy, the spawn-workers wait, PR 3's minors by @codetalcott in #380
• The shim's follow-ups: a bounded shutdown drain, a WSGI HEAD's head alone (D11, K13), PR 1's minors by @codetalcott in #381
• blobs keeps sending a picture, and now says why (D47) by @codetalcott in #373
• A WebSocket's close code reaches the app, and an oversized inbound message is refused with 1009 (L28, I26) by @codetalcott in #382
• chore(deps): bump granian from 2.8.2 to 2.8.3 in the python-deps group across 1 directory by @dependabot[bot] in #365
• m0-datastar against the Datastar SDK's own conformance cases (I27-I29) by @codetalcott in #383
• A Datastar replay too big for the connection is not served in part (I30) by @codetalcott in #384
• fix(uri): an @ ends a userinfo only inside the authority (A22) by @codetalcott in #385
• Release 1.6.0, and m0 0.2.0 by @codetalcott in #386

Full Changelog: v1.5.0...v1.6.0