v1.0.0-RC1

What's Changed

• Wait out a release whose binaries are still uploading by @jnnngs in #86
• Mark N02 fixed and plan bounded host stewardship by @jnnngs in #87
• Let a pool choose its hosts, and say which machines that is by @jnnngs in #88
• .github/workflows: Migrate workflows to Blacksmith runners by @blacksmith-sh[bot] in #89
• Prewarm every pool's image again on a timer, so a moving tag reaches the hosts by @jnnngs in #91
• Close the findings from the security review of auth and the web UI by @jnnngs in #90
• Give the runner image the toolchain a build assumes it has by @jnnngs in #92
• Say when the controller is behind, without ever updating it by @jnnngs in #93
• Name pools and hosts for what they are, and make the runner image multi-OS by @jnnngs in #94
• Name a runner for its pool's shape, and generate the image catalogue by @jnnngs in #95
• Bump useblacksmith/setup-docker-builder from 2.1.0 to 2.2.0 in the actions group by @dependabot[bot] in #96
• Say which architectures a runner variant is really built for by @jnnngs in #97
• Recapture the screenshots from the fleet as it is now by @jnnngs in #98
• ZF-101, first pull request: a job knows which installation it belongs to, and no pool crosses that line by @jnnngs in #99
• ZF-101, second pull request: poll each installation on its own evidence, and its own quota by @jnnngs in #100
• ZF-101, third pull request: say when a pool's runners went to the default runner group by @jnnngs in #101
• Recover from a route chunk that does not arrive by @jnnngs in #102
• ZF-102, second pull request: refuse a second controller over the same database by @jnnngs in #103
• Let a restarted agent keep the runners already on its host by @jnnngs in #104
• ZF-102, fourth pull request: late reports and restart boundaries by @jnnngs in #105
• ZF-301a: make the end-to-end harness able to fail by @jnnngs in #106
• ZF-301b: the drill tier — run the product for real against a fake GitHub by @jnnngs in #107
• Usage page fixes, and ZF-105's cleanup record by @jnnngs in #108
• Give the image builder the cache key it has required since v2 by @jnnngs in #109
• Collect a runner's abandoned sidecar, and stop three pages alarming or failing by @jnnngs in #110
• Two ways a runner's cleanup quietly stops happening by @jnnngs in #111
• Let the usage report read down a phone instead of across it by @jnnngs in #112
• Wait as long as GitHub asked, and stop asking while it says no by @jnnngs in #113
• Move a tab onto the build the fleet is running, and stop a removal racing its own exit record by @jnnngs in #114
• Let a host say how much machine and how much disk its runners have by @jnnngs in #115
• ZF-104 (1/5): prove the access and secret boundary by @jnnngs in #116
• Overview: count this fleet's jobs by default, with a switch for everybody else's by @jnnngs in #117
• Usage scope, host disk, and ZF-104 PR5 (uniform proxy trust) by @jnnngs in #118
• ZF-104 (3/5): a live stream ends when its credential is revoked by @jnnngs in #119
• 📚 Docs: close the pool.dangerous / pool.cache_shared gap in problem-codes.md by @jnnngs in #121
• Phase 1 complete: ZF-104 PR4, ZF-105's last finding, the Gate F note, and the gap it found by @jnnngs in #120
• Put the real dog on the About page, and say what Zoomies is by @jnnngs in #122
• Reconcile the roadmap with the code, and hand it to the next session by @jnnngs in #123
• Let the migration wizard reach every repository, and every workflow in one by @jnnngs in #124
• Say what Zoomies is, to the people who have not heard of it by @jnnngs in #125
• Plan Windows runners, and close the last Enterprise Server claim by @jnnngs in #126
• Let a migration point one job somewhere other than the label mapping says by @jnnngs in #127
• Finish the SEO pass, and hold the landing pages to the catalogue by @jnnngs in #128
• Hold the two landing pages to the runner image catalogue by @jnnngs in #129
• Give every page one refresh, in the same place by @jnnngs in #130
• Say when a host's slots and its heartbeat move by @jnnngs in #131
• Stop the product implying a Windows platform it has not got by @jnnngs in #132
• Charge a runner for the machine it takes, not just the slot by @jnnngs in #133
• Make the problems drawer survive the thing it exists to report by @jnnngs in #134
• Say whether the fallback poller is still sweeping by @jnnngs in #135
• Tell the two causes of a stuck runner apart, on its own page by @jnnngs in #136
• Say why a job GitHub is holding has not started by @jnnngs in #137
• Give the tests a fleet with something wrong with it by @jnnngs in #138
• Answer "why is this job not running?" once, on the side that knows by @jnnngs in #139
• Let the drawer and the CLI stop reasoning for themselves by @jnnngs in #140
• Put the whole instance in one document, for a bug report by @jnnngs in #141
• Read the buffer rather than its string by @jnnngs in #142
• Give backup and restore something to stand on by @jnnngs in #143
• Make taking a backup a thing an operator can do by @jnnngs in #144
• Put a backup back, and refuse before anything moves by @jnnngs in #145
• Let a restored fleet decide without doing by @jnnngs in #146
• Notice an agent that has fallen out of protocol by @jnnngs in #147
• Say which hosts are on another release, once by @jnnngs in #148
• Prove a database from an older release still opens by @jnnngs in #149
• Finish ZF-204 and ZF-205: upgrades, telemetry and the load measurement by @jnnngs in #150
• Let an operator hold a machine back, and see what it has promised away by @jnnngs in #151
• Three things the first-run journey said that were not true by @jnnngs in #152
• Finish ZF-201, and let a controller under load ask its agents to poll less often by @jnnngs in #153
• Keep the drill rows, so they can be compared with the ones before them by @jnnngs in #154
• Let a Nixpacks-based PaaS build and run the controller by @jnnngs in #155
• Keep a vendor's queued jobs behind the switch that hides the rest by @jnnngs in #156
• Kill the controller, then the agent, while a job is running by @jnnngs in #157
• Take the quota away, and refuse to mint a runner by @jnnngs in #159
• Roll back, and run a job on what came back by @jnnngs in #160
• Say a repository cannot be migrated before the wizard offers it by @jnnngs in #158
• Take a host's Docker socket away, and make the fleet say so by @jnnngs in #161
• Pin the Docker-socket drill to the host whose socket is missing by @jnnngs in #163
• Say which host the wizard's count lost, and why by @jnnngs in #162
• Keep the end of a runner's log on screen, and let the jump button be clicked by @jnnngs in #164
• Fix the Go check on PR #165: promote prometheus/client_model to a direct dependency by @jnnngs in #166
• Fix RC1 cleanup reporting and lifecycle timing discrepancies by @jnnngs in #165
• Serve index.html by what it holds, not by a date that never moves by @jnnngs in #167
• Make revoking somebody's access revoke their tokens too by @jnnngs in #168
• RC1 review fixes: restore safety and webhook trust by @jnnngs in #169
• RC1 blockers: seven fixes, plus red main and the pool page on a phone by @jnnngs in #170
• Enrol a host on the agent this controller is, and stop the join being dropped by @jnnngs in #171
• Publish the agent image, give :latest to releases, and make a containerised agent able to start by @jnnngs in #172
• Allow release candidates to reuse an existing published release by @jnnngs with @Copilot in #173
• Release setup: allow prerelease-tag rebuilds while preserving full-release immutability by @jnnngs with @Copilot in #175
• Make Upgrade CI choose the latest installable release, not just the latest tag by @jnnngs with @Copilot in #174
• Run release builds when a GitHub release is published by @jnnngs in #176

Full Changelog: v0.2-beta...v1.0.0-RC1

What's Changed

• Wait out a release whose binaries are still uploading by @jnnngs in #86
• Mark N02 fixed and plan bounded host stewardship by @jnnngs in #87
• Let a pool choose its hosts, and say which machines that is by @jnnngs in #88
• .github/workflows: Migrate workflows to Blacksmith runners by @blacksmith-sh[bot] in #89
• Prewarm every pool's image again on a timer, so a moving tag reaches the hosts by @jnnngs in #91
• Close the findings from the security review of auth and the web UI by @jnnngs in #90
• Give the runner image the toolchain a build assumes it has by @jnnngs in #92
• Say when the controller is behind, without ever updating it by @jnnngs in #93
• Name pools and hosts for what they are, and make the runner image multi-OS by @jnnngs in #94
• Name a runner for its pool's shape, and generate the image catalogue by @jnnngs in #95
• Bump useblacksmith/setup-docker-builder from 2.1.0 to 2.2.0 in the actions group by @dependabot[bot] in #96
• Say which architectures a runner variant is really built for by @jnnngs in #97
• Recapture the screenshots from the fleet as it is now by @jnnngs in #98
• ZF-101, first pull request: a job knows which installation it belongs to, and no pool crosses that line by @jnnngs in #99
• ZF-101, second pull request: poll each installation on its own evidence, and its own quota by @jnnngs in #100
• ZF-101, third pull request: say when a pool's runners went to the default runner group by @jnnngs in #101
• Recover from a route chunk that does not arrive by @jnnngs in #102
• ZF-102, second pull request: refuse a second controller over the same database by @jnnngs in #103
• Let a restarted agent keep the runners already on its host by @jnnngs in #104
• ZF-102, fourth pull request: late reports and restart boundaries by @jnnngs in #105
• ZF-301a: make the end-to-end harness able to fail by @jnnngs in #106
• ZF-301b: the drill tier — run the product for real against a fake GitHub by @jnnngs in #107
• Usage page fixes, and ZF-105's cleanup record by @jnnngs in #108
• Give the image builder the cache key it has required since v2 by @jnnngs in #109
• Collect a runner's abandoned sidecar, and stop three pages alarming or failing by @jnnngs in #110
• Two ways a runner's cleanup quietly stops happening by @jnnngs in #111
• Let the usage report read down a phone instead of across it by @jnnngs in #112
• Wait as long as GitHub asked, and stop asking while it says no by @jnnngs in #113
• Move a tab onto the build the fleet is running, and stop a removal racing its own exit record by @jnnngs in #114
• Let a host say how much machine and how much disk its runners have by @jnnngs in #115
• ZF-104 (1/5): prove the access and secret boundary by @jnnngs in #116
• Overview: count this fleet's jobs by default, with a switch for everybody else's by @jnnngs in #117
• Usage scope, host disk, and ZF-104 PR5 (uniform proxy trust) by @jnnngs in #118
• ZF-104 (3/5): a live stream ends when its credential is revoked by @jnnngs in #119
• 📚 Docs: close the pool.dangerous / pool.cache_shared gap in problem-codes.md by @jnnngs in #121
• Phase 1 complete: ZF-104 PR4, ZF-105's last finding, the Gate F note, and the gap it found by @jnnngs in #120
• Put the real dog on the About page, and say what Zoomies is by @jnnngs in #122
• Reconcile the roadmap with the code, and hand it to the next session by @jnnngs in #123
• Let the migration wizard reach every repository, and every workflow in one by @jnnngs in #124
• Say what Zoomies is, to the people who have not heard of it by @jnnngs in #125
• Plan Windows runners, and close the last Enterprise Server claim by @jnnngs in #126
• Let a migration point one job somewhere other than the label mapping says by @jnnngs in #127
• Finish the SEO pass, and hold the landing pages to the catalogue by @jnnngs in #128
• Hold the two landing pages to the runner image catalogue by @jnnngs in #129
• Give every page one refresh, in the same place by @jnnngs in #130
• Say when a host's slots and its heartbeat move by @jnnngs in #131
• Stop the product implying a Windows platform it has not got by @jnnngs in #132
• Charge a runner for the machine it takes, not just the slot by @jnnngs in #133
• Make the problems drawer survive the thing it exists to report by @jnnngs in #134
• Say whether the fallback poller is still sweeping by @jnnngs in #135
• Tell the two causes of a stuck runner apart, on its own page by @jnnngs in #136
• Say why a job GitHub is holding has not started by @jnnngs in #137
• Give the tests a fleet with something wrong with it by @jnnngs in #138
• Answer "why is this job not running?" once, on the side that knows by @jnnngs in #139
• Let the drawer and the CLI stop reasoning for themselves by @jnnngs in #140
• Put the whole instance in one document, for a bug report by @jnnngs in #141
• Read the buffer rather than its string by @jnnngs in #142
• Give backup and restore something to stand on by @jnnngs in #143
• Make taking a backup a thing an operator can do by @jnnngs in #144
• Put a backup back, and refuse before anything moves by @jnnngs in #145
• Let a restored fleet decide without doing by @jnnngs in #146
• Notice an agent that has fallen out of protocol by @jnnngs in #147
• Say which hosts are on another release, once by @jnnngs in #148
• Prove a database from an older release still opens by @jnnngs in #149
• Finish ZF-204 and ZF-205: upgrades, telemetry and the load measurement by @jnnngs in #150
• Let an operator hold a machine back, and see what it has promised away by @jnnngs in #151
• Three things the first-run journey said that were not true by @jnnngs in #152
• Finish ZF-201, and let a controller under load ask its agents to poll less often by @jnnngs in #153
• Keep the drill rows, so they can be compared with the ones before them by @jnnngs in #154
• Let a Nixpacks-based PaaS build and run the controller by @jnnngs in #155
• Keep a vendor's queued jobs behind the switch that hides the rest by @jnnngs in #156
• Kill the controller, then the agent, while a job is running by @jnnngs in #157
• Take the quota away, and refuse to mint a runner by @jnnngs in #159
• Roll back, and run a job on what came back by @jnnngs in #160
• Say a repository cannot be migrated before the wizard offers it by @jnnngs in #158
• Take a host's Docker socket away, and make the fleet say so by @jnnngs in #161
• Pin the Docker-socket drill to the host whose socket is missing by @jnnngs in #163
• Say which host the wizard's count lost, and why by @jnnngs in #162
• Keep the end of a runner's log on screen, and let the jump button be clicked by @jnnngs in #164
• Fix the Go check on PR #165: promote prometheus/client_model to a direct dependency by @jnnngs in #166
• Fix RC1 cleanup reporting and lifecycle timing discrepancies by @jnnngs in #165
• Serve index.html by what it holds, not by a date that never moves by @jnnngs in #167
• Make revoking somebody's access revoke their tokens too by @jnnngs in #168
• RC1 review fixes: restore safety and webhook trust by @jnnngs in #169
• RC1 blockers: seven fixes, plus red main and the pool page on a phone by @jnnngs in #170
• Enrol a host on the agent this controller is, and stop the join being dropped by @jnnngs in #171
• Publish the agent image, give :latest to releases, and make a containerised agent able to start by @jnnngs in #172
• Allow release candidates to reuse an existing published release by @jnnngs with @Copilot in #173
• Release setup: allow prerelease-tag rebuilds while preserving full-release immutability by @jnnngs with @Copilot in #175
• Make Upgrade CI choose the latest installable release, not just the latest tag by @jnnngs with @Copilot in #174
• Run release builds when a GitHub release is published by @jnnngs in #176

Full Changelog: v0.2-beta...v1.0.0-RC1