Skip to content

Feature/11 - #12

Merged
ginccc merged 2 commits into
masterfrom
feature/11
Nov 7, 2016
Merged

Feature/11#12
ginccc merged 2 commits into
masterfrom
feature/11

Conversation

@Mobe91

@Mobe91 Mobe91 commented Oct 31, 2016

Copy link
Copy Markdown
Contributor

improved startup scripts and fixed assembly build

@ginccc
ginccc merged commit ce19abf into master Nov 7, 2016
@ginccc
ginccc deleted the feature/11 branch November 7, 2016 15:56
ginccc added a commit that referenced this pull request Aug 15, 2017
ginccc added a commit that referenced this pull request Apr 19, 2026
Review fix #1: Delete dead SlackChannelRouter + SlackChannelRouterTest
  - Class was @ApplicationScoped but never injected, causing double
    startup cost scanning all agents for channel connectors
  - Both files removed (265 + 351 LOC)

Review fix #2: Migration tool now merges duplicate channelId entries
  - Old: created one config per (agent, channel) pair, last-write-wins
  - New: groups connectors by platformChannelId and creates a single
    multi-target config per channel with agent-derived trigger keywords
  - Multi-agent channels show mergedAgents in dry-run output

Review fix #3: Deep-copy config before resolving secrets
  - resolvePlatformSecrets was mutating the store's instance in-place,
    which would leak plaintext secrets if a caching layer is added
  - Added deepCopyConfig() — router works on copies, REST returns
    vault references

Review fix #5: Null/blank trigger guard in validation
  - Null triggers from loose JSON input now return 400 instead of NPE

Review fix #6: Remove dead fields
  - Removed newStyleChannelIds (assigned but never read)
  - Removed cacheFactory field (only used in constructor)
  - Removed unused ConcurrentHashMap import

Review fix #7/#12: Reject observeMode=true until implemented
  - Validation now blocks observeMode=true with clear error message

Review fix #8: Preserve stack traces in router error logging
  - All LOGGER.warnf(msg, e.getMessage()) changed to
    LOGGER.warn(msg, e) for production diagnosability

Review fix #10: Rename 'channelId' to 'resourceId' in MCP responses
  - Eliminates confusion between Slack channelId and Mongo resourceId

Review fix #11: Fix deployAgent description typo
  - 'production' was listed twice in 4 environment descriptions

Review fix #17: Temper platform-agnostic Javadoc claim
  - Javadoc now says 'currently Slack-only with platform-agnostic model'
ginccc added a commit that referenced this pull request Jul 3, 2026
…pod cancel honored

- #12: deleteGroupConversation now runs the same paused-state cleanup as
  cancel-of-paused (delete timeout schedule + cleanupAfterTerminalState) so a
  deleted paused discussion no longer leaks its armed schedule, ephemeral
  dynamic agents, and lastVerifiedIndex entry.
- #13: cancelDiscussion decides the cancel path (token / DB CAS) FIRST and
  deletes the timeout schedule only after the CAS wins; on CAS loss it leaves
  the schedule alone and reports 409, so a cancel racing a fresh pause can no
  longer strip a finite timeout while reporting success.
- #27/#45: the running leg re-reads persisted state at each phase boundary and
  completes via updateIfState(running-state) so a cross-pod cancel/ABORT is
  honored at the next boundary and the terminal COMPLETED write can no longer
  resurrect IN_PROGRESS over a cluster-wide CANCELLED.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants