v1.0.0

See CHANGELOG.md for the full release notes.

What's Changed

• Re-soak against 0.19.0: four applications, six rows, zero failures; manifest pins follow the fixtures by @codetalcott in #255
• BENCHMARKS.md's prose may not carry a bare figure: spans or a marked observation, checked by @codetalcott in #256
• BENCHMARKS.md: trim the intro and caveats; Reproducing names the script behind every table by @codetalcott in #257
• Gate FastAPI through the ASGI bridge, and name it on the landing page by @codetalcott in #258
• BENCHMARKS.md: the pool follows Granian's architecture, it did not copy it by @codetalcott in #260
• Re-record the ASGI rows at 0.19.0, gate that script's machine, and retire the inversion's "nothing at saturation" verdict by @codetalcott in #261
• m0serve could not see its own CPU budget: usable_cpus reads affinity and the cgroup quota by @codetalcott in #262
• The Linux container's sync could rebuild nothing, or the wrong tree, and report success either way by @codetalcott in #263
• Design item 6: the inverted mode's drain is stepped, so a request mid-await at SIGTERM is answered when it finishes by @codetalcott in #264
• The loop inversion is visible and documented, and deliberately not a default by @codetalcott in #265
• Two of the benchmark page's four conclusions invert on Linux, and the artifact system could not have told you by @codetalcott in #266
• poe bench-linux-conclusions: the Linux check is a task now, and an empty comparison fails by @codetalcott in #267
• bench-linux-conclusions --remote: the same arms on a rented box, over ssh by @codetalcott in #268
• A second Linux box says otherwise: the WSGI inversion does not reproduce by @codetalcott in #269
• Gate the pool's other half: a view parallelises across --blocking-threads only if it releases the GIL by @codetalcott in #271
• 1.0.0 by @codetalcott in #274

Full Changelog: v0.19.0...v1.0.0