feat: sync away team assignments by blevouschurverantunt · Pull Request #295 · STFC-Mod/mod

blevouschurverantunt · 2026-09-19T04:39:43Z

Summary

Adds Away Team data to the sync feature so consumers can tell which officers are currently away and until when. Today processors::jobs() skips JOBTYPE_AWAYASSIGNMENT and the dispatcher ignores the AwayAssignments* entity groups, so a crew-planning tool will happily recommend an officer who is off on a 2-day away mission.

Everything is additive, read-only, reuses the existing ProcessResultInternal dispatcher (no new hooks), and is gated like every other data type.

What changes

New sync option away_assignments ([sync] and per-target, same semantics/defaults as the other data flags) → new data type away_assignment. It is one new row in SyncOptions, so config.cc needs no changes.
away_assignment events, built by one shared helper (away_assignment_event) so both sources share the same de-dup state and slot-ordered officer_ids:
- Entity group AwayAssignmentsList (111, AwayAssignmentUserListResponse) — sent when the Away Teams screen is opened and after collecting a mission.
- Entity group AwayAssignmentsInstance (113, a bare AwayAssignmentInstance) — sent when an assignment is started: first with state=1 (READYTOSTART), officers already chosen and an empty job_uuid, then ~1.5s later with state=2 (INPROGRESS) and job_uuid set.
Example event:
```
{"type":"away_assignment","aid":2814298639311718967,"template_id":4186662802,"state":2,
 "officer_ids":[3479870516,440622131,2520801863],"job_uuid":"60f7…264b","duration":43200,"rarity":2}
```
officer_ids is emitted in slot order (the proto map iterates in arbitrary order). Events are only emitted when the instance's observable state changed since it was last sent. When an instance disappears from the list a {"type":"collected_away_assignment","aid":…} event is sent and its state is pruned, mirroring completed_job.
job events now include JOBTYPE_AWAYASSIGNMENT with {"aid": …}, so start_time/duration flow through the existing job / completed_job events (gated by the existing jobs option) for the instance with the matching aid.
All 10 localized example configs get the new key (first entry of Subgroup: Data, alphabetical). The da/de/es/fr/nl/ru comments are best-effort translations in each file's existing style — corrections welcome.

Officer events (processors::officers()) are untouched by this PR — away-assignment status is not mixed into them. A prior revision added an away_assignment_id field there, but slot order isn't available in that hook, so instead of shipping an event that loses it, assignment starts are now reported directly via group 113 above.

Verified against the live client (Windows, probe build, 2026-09-20)

Group 111 parses as AwayAssignmentUserListResponse and re-serialises byte-exact; officerIds is slot index → officer id. Sent when the Away Teams screen is opened and in the response to collecting a mission (collected instance appears once with state 5, then drops out → collected_away_assignment). Lists were always complete, so pruning is safe.
Group 113 parses as a bare AwayAssignmentInstance and re-serialises byte-exact (285/285 and 317/317 bytes on the two captures); cross-parsing the same bytes as AwayAssignmentUserListResponse yields 0 instances, confirming it is not a wrapped list. It fires when an assignment is started, twice: state=1 (READYTOSTART) with officerIds already populated and an empty jobUuid, then ~1.5s later state=2 (INPROGRESS) with jobUuid set — the second one arrives in the same server response as the Jobs and Officers groups. It was not seen at login, when opening the Away Teams screen, or on collect (those use the 111 list).
The option is read from [sync], inherited per target, and written to community_patch_runtime.vars like the other data flags (it goes through the same SyncOptions loop).
Not handled on purpose: group 110 (static refresh settings; seen once at login) and group 112 (never observed).

Consumers only need away_assignments to know who is away (the state-2 event arrives at the moment an assignment starts); enable jobs as well for the exact start_time/duration, from the job event whose aid matches.

Note: aid is an int64 above 2^53, emitted as a JSON number like the other ids in the sync stream; JavaScript consumers should be aware of precision.

Checks

Fork CI on this exact commit: build-win, build-mac (arm64), build-mac (x86_64), package-mac, Validate example configs — all green.
git diff --check clean, LF only, clang-format applied to the changed lines only.
No changes to CHANGELOG, README, version.h, or xmake-requires.lock.

🤖 Generated with Claude Code

Add SyncConfig::Type::AwayAssignments as the first enumerator (mirrors Jobs/Missions/Traits naming), a corresponding away_assignments member (default false, matching every other sync data flag), and its SyncOptions row so TOML loading, per-target overrides, runtime.vars write-back and JSON type serialization all pick it up automatically. config.cc needs no changes since everything iterates SyncOptions. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Add processors::away_assignments_list (EntityGroup 111, AwayAssignmentUserListResponse) and processors::away_assignment_instance (EntityGroup 113, bare AwayAssignmentInstance), following the same two-processors-plus-shared-helper split used for entity_slots / entity_slots_data + single_slot_locked, since no entity group carries a single-instance wrapper here. away_assignment_event() builds the sync event for one instance and only emits it when a hash of its observable fields changes, tracked in the new trackers::away_assignment_states map (mirrors slot_states). The list processor also prunes ids that drop out of the response and emits a collected_away_assignment event for each, mirroring how jobs() prunes completed jobs. Wire both entity groups into HandleEntityGroup's dispatcher as the first (alphabetical) block, gated by the new away_assignments option. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

jobs(): add JOBTYPE_AWAYASSIGNMENT so away-team jobs flow through the existing job/completed_job events with an "aid" pointing at the instance, in the same style as the other job-type cases. officers(): surface Officer.awayAssignmentInstanceId as "away_assignment_id" on the officer event. The officer de-dup state is a struct local to officers() (kept separate from the shared RankLevelShardsState used by forbidden_tech) so this extra field doesn't change forbidden tech's comparison at all, while still causing an officer to re-emit if only their away assignment changes. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Add the away_assignments key as the first entry of the [sync] "Subgroup: Data" block in all 10 localized example config files, matching the existing comment/blank-line rhythm of the other data keys. da/de/es/fr/nl/ru translations follow each file's own missions line pattern and are best-effort; tlh follows the file's established "Do you want Daq sync X De'?" template. The unsuffixed pointer file and [sync.targets.*] example blocks are intentionally left untouched, matching how every other sync data key (e.g. traits) is handled there. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

officerIds is a protobuf map, whose iteration order does not match slot order (observed e.g. 2=.., 0=.., 1=..). Sort by slot index before emitting officer_ids so the array, and the de-dup hash built from it, are stable run to run. Remove the AwayAssignmentsInstance (group 113) processor and dispatch case: it was never observed on the wire during probing (not at login, not on opening the screen, not on collect), so its payload shape is unverified and we should not ship unverified parsing. AwayAssignmentsList (group 111) is the only observed source of away assignment sync data and remains the single caller of away_assignment_event(), whose comment now states plainly that the caller must hold trackers::away_assignment_states_mtx. Also runs clang-format over the touched line ranges in config.h and sync.cc to keep the branch's own added/changed lines clean. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

netniV

Changes look good. @lightbull-stfc may want to weigh in on whether we need anything more or compatibility for Spocks Club.

lightbull-stfc

Let's explore if there's a way to emit updates to away_assignments at a point when an assignment is started (in addition to when the whole list is refreshed) to avoid using officer updates for that.

lightbull-stfc · 2026-09-20T14:56:41Z

                                    {"level", officer.level()},
-                                    {"shard_count", officer.shardcount()}});
+                                    {"shard_count", officer.shardcount()},
+                                    {"away_assignment_id", officer.awayassignmentinstanceid()}});


This shifts the burden to the sync target to piece back the data together. Slot order is unavailable in this hook and is thus lost. I would prefer to remove it and emit a started assignment through other means.

lightbull-stfc · 2026-09-20T15:19:33Z

+                {"duration", instance.parameters().duration()},
+                {"rarity", instance.rarity()}};
+
+  const auto state_value = static_cast<size_t>(std::hash<json>{}(event));


Redundant static_cast

lightbull-stfc · 2026-09-20T15:22:40Z

@@ -1446,7 +1548,8 @@ static void officers(std::unique_ptr<std::string>&& bytes)
      std::scoped_lock lk(officer_states_mtx);

      for (const auto& officer : response.officers()) {
-        const RankLevelShardsState officer_state{officer.rankindex(), officer.level(), officer.shardcount()};
+        const OfficerState officer_state{officer.rankindex(), officer.level(), officer.shardcount(),
+                                         officer.awayassignmentinstanceid()};


I would prefer not to have Away Assignments mixed with officers. For users to benefit from this feature addition, they'd currently have to enable away_assignments, officers, and jobs options in the config.

Address sync maintainer review feedback on PR STFC-Mod#295: - Revert processors::officers() to be byte-identical to dev. Away assignment status stays off officer events; slot order isn't available in that hook, so instead of losing it we emit assignment starts through a dedicated event (see below). - Drop a redundant static_cast in away_assignment_event(): std::hash already returns size_t, matching away_assignment_states's value type. - Restore processors::away_assignment_instance for EntityGroup 113 (AwayAssignmentsInstance), parsed as a bare AwayAssignmentInstance and routed through the existing away_assignment_event()/ away_assignment_states_mtx so list and instance updates share one de-dup state and the slot-ordered officer_ids. Live capture against the Windows client confirms group 113 fires when an assignment is started: first at state=1 (READYTOSTART) with officerIds already populated and an empty jobUuid, then ~1.5s later at state=2 (INPROGRESS) with jobUuid set. Wired into HandleEntityGroup's dispatcher right after the existing AwayAssignmentsList case, gated by the same away_assignments option. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

blevouschurverantunt · 2026-09-20T16:48:34Z

Thanks for the review — all addressed in 6bd1893:

The officer change is dropped entirely; officers() is identical to dev again.
The redundant cast is gone.
Start-time emission: I captured a start with groups 110–113 logged. The server sends group 113 (AwayAssignmentsInstance, a bare AwayAssignmentInstance that re-serialises byte-exact) when an assignment is started — once with state 1 and the chosen officers, then ~1.5 s later with state 2 and the job uuid. It doesn't appear at login, when opening the Away Teams screen, or on collect, which is why I hadn't seen it before. It now goes through the same helper as the 111 list (shared de-dup state, slot-ordered officer_ids), so consumers only need away_assignments. With this build, a started assignment reached my sync target as a state-2 away_assignment event about a second after the job's start_time.

The PR description is updated with the details. Fork CI is green on this exact commit (Windows, both macOS archs, package, config validation); the workflows here still need a maintainer's approval to run.

netniV

Thanks for the update. Unfortunately, there are now conflicts with this branch. Could you resolve those so myself and @lightbull-stfc can review once more.

…signments # Conflicts: # mods/src/config.h # mods/src/patches/parts/sync.cc Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

blevouschurverantunt · 2026-09-23T15:12:04Z

Conflicts resolved by merging the current dev into the branch (1db8a05). The Haven sync and this change now sit side by side: both options in SyncOptions, both dispatcher blocks, no changes to either feature's behaviour. Builds clean locally and on my fork's CI. Ready for another look.

netniV · 2026-09-27T18:59:20Z

I have approved, but I'll let @lightbull-stfc have the final say.

…signments Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Client build 267 (2026-09-29) delivers battle result headers as entity group 219 (BattleResultHeadersResponse) and owned ships as group 251 (PlayerShipsResponse) instead of the Json group. Implement the 219 processor and add a 251 processor, sharing helpers with the Json handlers. Add the client's EntityGroup values 242-252, and declare only the proto fields that are read. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

…assignments Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

…signments Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

blevouschurverantunt · 2026-10-02T01:37:28Z

Conflicts resolved again by merging the current dev (now including #315) into the branch (b8c53b1). The diff against dev is unchanged: the away-assignment feature only. Fork CI green on this exact commit.

blevouschurverantunt and others added 5 commits September 18, 2026 20:33

netniV approved these changes Sep 19, 2026

View reviewed changes

lightbull-stfc suggested changes Sep 20, 2026

View reviewed changes

blevouschurverantunt mentioned this pull request Sep 22, 2026

TEST BUILD: away assignment trait demand fields (fork CI only) blevouschurverantunt/stfc-mod#4

Closed

netniV requested changes Sep 23, 2026

View reviewed changes

Merge remote-tracking branch 'upstream/dev' into feature/sync-away-as…

1db8a05

…signments # Conflicts: # mods/src/config.h # mods/src/patches/parts/sync.cc Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

netniV approved these changes Sep 27, 2026

View reviewed changes

blevouschurverantunt and others added 2 commits September 30, 2026 09:52

Merge remote-tracking branch 'upstream/dev' into feature/sync-away-as…

a387e1a

…signments Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

blevouschurverantunt mentioned this pull request Sep 30, 2026

Restore battle log and ship sync for the current client (build 267) #315

Merged

blevouschurverantunt and others added 2 commits September 30, 2026 14:14

Merge fix for client build 267 (STFC-Mod#315) into feature/sync-away-…

1451bf9

…assignments Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Merge remote-tracking branch 'upstream/dev' into feature/sync-away-as…

b8c53b1

…signments Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

netniV merged commit 1c2e70e into STFC-Mod:dev Oct 3, 2026
6 checks passed

blevouschurverantunt mentioned this pull request Oct 4, 2026

Include trait demand in away assignment events #318

Merged

Navigation Menu

Repository navigation

feat: sync away team assignments - #295

feat: sync away team assignments#295
netniV merged 11 commits into
STFC-Mod:devfrom
blevouschurverantunt:feature/sync-away-assignments

blevouschurverantunt commented Sep 19, 2026 •

edited

Loading

Uh oh!

netniV left a comment

Uh oh!

lightbull-stfc left a comment

Uh oh!

lightbull-stfc Sep 20, 2026

Uh oh!

lightbull-stfc Sep 20, 2026

Uh oh!

lightbull-stfc Sep 20, 2026

Uh oh!

blevouschurverantunt commented Sep 20, 2026

Uh oh!

netniV left a comment

Uh oh!

blevouschurverantunt commented Sep 23, 2026

Uh oh!

netniV commented Sep 27, 2026

Uh oh!

blevouschurverantunt commented Oct 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

blevouschurverantunt commented Sep 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What changes

Verified against the live client (Windows, probe build, 2026-09-20)

Checks

Uh oh!

netniV left a comment

Choose a reason for hiding this comment

Uh oh!

lightbull-stfc left a comment

Choose a reason for hiding this comment

Uh oh!

lightbull-stfc Sep 20, 2026

Choose a reason for hiding this comment

Uh oh!

lightbull-stfc Sep 20, 2026

Choose a reason for hiding this comment

Uh oh!

lightbull-stfc Sep 20, 2026

Choose a reason for hiding this comment

Uh oh!

blevouschurverantunt commented Sep 20, 2026

Uh oh!

netniV left a comment

Choose a reason for hiding this comment

Uh oh!

blevouschurverantunt commented Sep 23, 2026

Uh oh!

netniV commented Sep 27, 2026

Uh oh!

blevouschurverantunt commented Oct 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

blevouschurverantunt commented Sep 19, 2026 •

edited

Loading