switchboard/core
Archived
1
0
Fork 0
Code Issues Pull Requests Actions Releases 9 Wiki Activity

Changeset 0.15.0 patches (#72)

Browse Source
This commit is contained in:
xcaliber
2026-02-26 23:44:44 +00:00
parent 2d79ff593b
commit e663104575
12 changed files with 76 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.gitea/workflows/ci.yaml
View File

@@ -161,7 +161,7 @@ jobs:
PGPORT: ${{ env.POSTGRES_PORT }}
PGUSER: ${{ secrets.POSTGRES_USER }}
PGPASSWORD: ${{ secrets.POSTGRES_PASSWORD }}
run: go test -v -race -count=1 ./...
run: go test -v -race -count=1 -p 1 ./...
- name: Drop CI test database
if: always()

2
VERSION
View File

@@ -1 +1 @@
0.14.0
0.15.0

28
docs/ROADMAP.md
View File

@@ -749,15 +749,30 @@ See [DESIGN-0.14.0.md](DESIGN-0.14.0.md) for full spec.
---
## v0.15.0 — Compaction
## v0.15.0 — Compaction ✅
Depends on: utility model role (v0.10.0).
- [ ] Auto-compaction service: background job that calls utility model to summarize
- [ ] Channel-scoped: triggers when context exceeds threshold
- [ ] Summaries stored as system messages
- [ ] Configurable: per-channel opt-in/out, summary model override
- [ ] Admin controls for resource limits
- [x] Compaction service: extracts messages, calls utility model to summarize, replaces with summary
- [x] `Summarize & Continue` button: user-triggered compaction via existing summarize endpoint
- [x] Channel-scoped context threshold triggers (configurable min messages, min chars, activity gap)
- [x] Summaries stored as system messages with `is_summary = true` metadata
- [x] Background scanner: periodic candidate detection with cooldown and in-flight dedup
- [x] Configurable: per-channel opt-in/out (`compaction_opt_out`), global threshold overrides
- [x] Admin controls: global enable/disable, threshold/cooldown settings via `global_settings`
- [x] Context budget guard rail: estimated tokens must exceed 80% of ceiling before compaction
- [x] Token estimation: `(len + 3) / 4` heuristic with path-from-summary support
- [x] Rate limiting on summarize endpoint
- [x] Comprehensive test suite: scanner candidates, guard rails, live pipeline integration tests
Bug fixes during CI stabilization:
- [x] Fixed `channels.deleted_at` → `channels.is_archived` (channels use soft-archive, not soft-delete)
- [x] Fixed NULL model column scan in scanner (`COALESCE(c.model, '')`)
- [x] Fixed test channel backdating (BEFORE UPDATE trigger override via INSERT)
- [x] Fixed vault nil-check fallback in role resolver for plaintext API keys
- [x] Fixed parallel test DB races (`-p 1` serialization)
- [x] Fixed proactive token refresh cascade: failed refresh no longer nukes active session
- [x] Fixed model visibility reset: preferences fetch failure preserves existing hidden state
---
@@ -1160,6 +1175,7 @@ based on need.
**UX / Multi-Seat**
- ~~Per-chat model/preset persistence (server-side)~~ ✅ _(v0.12.0 — `channels.settings.last_selector_id` JSONB merge, localStorage write-through cache)_
- Per-provider model preferences: `user_model_settings` unique key is `(user_id, model_id)` — same model from different providers shares one visibility toggle. Needs `provider_config_id` dimension in DB constraint, store, API, and frontend `hiddenModels` keying. Frontend composite key (`configId:modelId`) already exists in `App.models[].id`.
**Knowledge Bases — Future**
- Hybrid search: combine vector similarity with full-text `tsvector`, re-rank

38
server/compaction/compaction_test.go
View File

@@ -43,14 +43,25 @@ func setGlobalSetting(t *testing.T, key string, value interface{}) {
}
}
// touchChannelTime backdates a channel's updated_at.
func touchChannelTime(t *testing.T, channelID string, age time.Duration) {
// seedChannelBackdated creates a channel with updated_at set to (now - age).
// Uses INSERT with an explicit timestamp so the BEFORE UPDATE trigger on
// channels (which overwrites updated_at = NOW()) never fires.
func seedChannelBackdated(t *testing.T, userID, title string, age time.Duration, msgCount, contentSize int) (channelID string, msgIDs []string) {
t.Helper()
target := time.Now().Add(-age)
_, err := database.DB.Exec(`UPDATE channels SET updated_at = $1 WHERE id = $2`, target, channelID)
err := database.DB.QueryRow(`
INSERT INTO channels (user_id, title, type, updated_at)
VALUES ($1, $2, 'direct', $3)
RETURNING id
`, userID, title, target).Scan(&channelID)
if err != nil {
t.Fatalf("touchChannelTime: %v", err)
t.Fatalf("seedChannelBackdated: %v", err)
}
msgIDs = database.SeedTestMessages(t, channelID, msgCount, contentSize)
if len(msgIDs) > 0 {
database.SeedTestCursor(t, channelID, userID, msgIDs[len(msgIDs)-1])
}
return
}
// setChannelSettings sets the channel.settings JSONB.
@@ -89,10 +100,8 @@ func TestScanner_FindCandidates_ReturnsQualifying(t *testing.T) {
// Create a channel with enough messages to qualify
// 20 messages × 2000 chars = 40K chars (> candidateMinChars=20K)
channelID, _ := seedChannel(t, userID, "Big Chat", 20, 2000)
// Backdate so it passes the activity gap (>2min old) and recency (<7 days)
touchChannelTime(t, channelID, 5*time.Minute)
// Backdated 5min so it passes activity gap (>2min) and recency (<7 days)
channelID, _ := seedChannelBackdated(t, userID, "Big Chat", 5*time.Minute, 20, 2000)
ctx := context.Background()
candidates := sc.findCandidates(ctx)
@@ -122,8 +131,7 @@ func TestScanner_FindCandidates_ExcludesTooFewMessages(t *testing.T) {
userID := database.SeedTestUser(t, "scanuser2", "scan2@test.com")
// Only 4 messages — below candidateMinMessages=10
channelID, _ := seedChannel(t, userID, "Small Chat", 4, 2000)
touchChannelTime(t, channelID, 5*time.Minute)
channelID, _ := seedChannelBackdated(t, userID, "Small Chat", 5*time.Minute, 4, 2000)
ctx := context.Background()
candidates := sc.findCandidates(ctx)
@@ -164,8 +172,7 @@ func TestScanner_FindCandidates_ExcludesArchived(t *testing.T) {
sc := NewScanner(svc, stores, ScannerConfig{})
userID := database.SeedTestUser(t, "scanuser4", "scan4@test.com")
channelID, _ := seedChannel(t, userID, "Archived Chat", 20, 2000)
touchChannelTime(t, channelID, 5*time.Minute)
channelID, _ := seedChannelBackdated(t, userID, "Archived Chat", 5*time.Minute, 20, 2000)
// Archive the channel
database.DB.Exec(`UPDATE channels SET is_archived = true WHERE id = $1`, channelID)
@@ -377,8 +384,9 @@ func TestScanner_GetCooldownDuration_Override(t *testing.T) {
func TestEstimateTokens_GuardRailMath(t *testing.T) {
// Simulate a large conversation that exceeds a 32K utility model
// 100K chars ≈ 25K tokens of conversation + system prompt overhead
largeContent := make([]byte, 100000)
// 120K chars ≈ 30K tokens of conversation + system prompt overhead
// inputCeiling = 32K * 0.80 = 25.6K → 30K exceeds it
largeContent := make([]byte, 120000)
contentTokens := EstimateTokens(string(largeContent))
systemTokens := EstimateTokens("You are a conversation summarizer...") + 8 // +overhead
@@ -387,7 +395,7 @@ func TestEstimateTokens_GuardRailMath(t *testing.T) {
inputCeiling := int(float64(utilityBudget) * 0.80)
if totalPrompt <= inputCeiling {
t.Errorf("100K chars (%d tokens) should exceed 32K model ceiling (%d tokens)",
t.Errorf("120K chars (%d tokens) should exceed 32K model ceiling (%d tokens)",
totalPrompt, inputCeiling)
}

5
server/compaction/scanner.go
View File

@@ -237,13 +237,12 @@ func (sc *Scanner) findCandidates(ctx context.Context) []models.Channel {
maxAge := time.Now().Add(-candidateMaxAge)
rows, err := database.DB.QueryContext(ctx, `
SELECT c.id, c.user_id, c.model, c.settings::text,
SELECT c.id, c.user_id, COALESCE(c.model, ''), COALESCE(c.settings::text, '{}'),
COUNT(m.id) AS msg_count,
COALESCE(SUM(LENGTH(m.content)), 0) AS total_chars
FROM channels c
JOIN messages m ON m.channel_id = c.id AND m.deleted_at IS NULL
WHERE c.deleted_at IS NULL
AND c.type = 'direct'
WHERE c.type = 'direct'
AND c.is_archived = false
AND c.updated_at < $1
AND c.updated_at > $2

12
server/handlers/completion.go
View File

@@ -829,18 +829,6 @@ func (h *CompletionHandler) loadConversation(channelID, userID, presetSystemProm
return messages, nil
}
// isSummaryMessage checks if a PathMessage has summary metadata.
func isSummaryMessage(m *PathMessage) bool {
if m.Metadata == nil {
return false
}
var meta map[string]interface{}
if err := json.Unmarshal(*m.Metadata, &meta); err != nil {
return false
}
return meta["type"] == "summary"
}
// ── Message Persistence ─────────────────────
// persistMessage inserts a message into the tree, updates the cursor, and

2
server/handlers/summarize.go
View File

@@ -33,7 +33,7 @@ func (h *SummarizeHandler) Summarize(c *gin.Context) {
// ── Verify channel ownership ──
var ownerID string
err := database.DB.QueryRow(
`SELECT user_id FROM channels WHERE id = $1 AND deleted_at IS NULL`, channelID,
`SELECT user_id FROM channels WHERE id = $1`, channelID,
).Scan(&ownerID)
if err != nil {
c.JSON(http.StatusNotFound, gin.H{"error": "channel not found"})

7
server/main.go
View File

@@ -11,6 +11,7 @@ import (
"github.com/gin-gonic/gin"
"git.gobha.me/xcaliber/chat-switchboard/compaction"
"git.gobha.me/xcaliber/chat-switchboard/config"
"git.gobha.me/xcaliber/chat-switchboard/crypto"
"git.gobha.me/xcaliber/chat-switchboard/database"
@@ -248,8 +249,9 @@ func main() {
protected.POST("/chat/completions", comp.Complete)
protected.GET("/tools", comp.ListTools)
// Summarize & Continue
summarize := handlers.NewSummarizeHandler(stores, roleResolver)
// Summarize & Continue (backed by compaction service)
compactionSvc := compaction.NewService(stores, roleResolver)
summarize := handlers.NewSummarizeHandler(compactionSvc)
protected.POST("/channels/:id/summarize", summarize.Summarize)
// Provider Configs (user-facing — replaces /api-configs)
@@ -400,6 +402,7 @@ func main() {
admin.PUT("/users/:id/role", adm.UpdateUserRole)
admin.PUT("/users/:id/active", adm.ToggleUserActive)
admin.POST("/users/:id/reset-password", adm.ResetPassword)
admin.POST("/users/:id/vault/reset", adm.ResetVault)
admin.DELETE("/users/:id", adm.DeleteUser)
// Global settings

3
server/roles/roles.go
View File

@@ -279,6 +279,9 @@ func (r *Resolver) resolveBinding(ctx context.Context, binding *RoleBinding) (pr
if err != nil {
return nil, providers.ProviderConfig{}, "", fmt.Errorf("decrypt API key: %w", err)
}
} else if len(cfg.APIKeyEnc) > 0 {
// No vault — key stored as raw bytes (unencrypted fallback)
apiKey = string(cfg.APIKeyEnc)
}
// Parse headers

10
src/js/api.js
View File

@@ -131,7 +131,15 @@ const API = {
this._refreshTimer = setTimeout(async () => {
if (!this.refreshToken) return;
console.debug('🔄 Proactive token refresh');
await this.refresh();
try {
const data = await this._post('/api/v1/auth/refresh', { refresh_token: this.refreshToken }, true);
this._setAuth(data);
} catch (e) {
// Proactive refresh failed — keep session alive.
// Access token is still valid for ~20% of its lifetime.
// When it naturally expires, 401-retry in _authed() handles cleanup.
console.warn('⚠ Proactive refresh failed, access token still valid');
}
}, ms);
},

6
src/js/app.js
View File

@@ -62,7 +62,11 @@ async function fetchModels() {
App.hiddenModels = new Set(
(prefData.preferences || []).filter(p => p.hidden).map(p => p.model_id)
);
} catch (e) { App.hiddenModels = new Set(); }
} catch (e) {
// Keep existing preferences on failure (auth dead, network issue).
// Only init to empty if there's nothing to preserve.
if (!App.hiddenModels) App.hiddenModels = new Set();
}
App.models = (data.models || []).map(m => {
const isPreset = !!m.is_preset;

4
src/js/ui-settings.js
View File

@@ -509,7 +509,9 @@ Object.assign(UI, {
App.hiddenModels = new Set(
(prefData.preferences || []).filter(p => p.hidden).map(p => p.model_id)
);
} catch (e) { App.hiddenModels = new Set(); }
} catch (e) {
if (!App.hiddenModels) App.hiddenModels = new Set();
}
}
const data = await API.listEnabledModels();