Release v0.51.757 — page archived sidebar rows + archived-search refetch (#5200, @santastabber)

[nesquena-hermes]

Release v0.51.757 — page archived sidebar rows + archived-search refetch (#5200, @santastabber)

Ships #5200 by @santastabber. Perf + correctness fix on the session sidebar.

What it does

Archived sidebar rows are now paged (capped per request) instead of all loaded at once — an instance with thousands of archived sessions no longer renders them all. The contributor PR converged from an earlier gate-fail: when "Show archived" is on and you start a search, the sidebar transparently refetches without the archive cap so a title/id match beyond the first archived page is still reachable, then restores normal paging when the search clears.

Gate (converged from gate-fail, re-gated fresh)

• I earlier bounced this: the archived-paging suppression applied on the renderSessionList path but NOT the filterSessions() search-input path, so an archived title match past page 1 was unreachable while searching (SILENT data-not-shown). The re-push added _syncArchivedSearchPagingRefresh() + search-active detection that refetches without archived_limit on the search-active edge and restores paging on clear.
• Codex (reproduce): SAFE TO SHIP — verified the archived title match beyond page 1 now appears under archived-search; a render generation/queue prevents a slower capped/uncapped response from overwriting newer state (no thrash/loop); default non-search paging + the 2000-row cap stay intact; backend slices only archived rows with limit/offset-separated cache keys.
• 48 sidebar/cache tests pass; full suite 11075 passed (only the 2 pre-existing cron-isolation flakes — unrelated).

Closes #5200.

[greptile-apps]

Greptile Summary

This release PR adds paginated loading for archived sidebar rows (archived_limit/archived_offset query params on /api/sessions) and a search-active suppression that temporarily removes the limit when the user is typing a title search, ensuring archived matches beyond the first page remain reachable. When the search clears, paging restores automatically.

• Backend: new _query_positive_int helper; _build_session_list_cache_payload slices archived_scoped_filtered to the requested window while keeping all visible rows unaffected and preserving full-count metadata for the tab badge; the cache key is parameterised by limit and offset so differently-paged requests don't share a cache entry.
• Frontend: _archivedRowsLoadedLimit state + SESSION_ARCHIVED_PAGE_SIZE / SESSION_ARCHIVED_MAX_LOADED_LIMIT constants drive a "Load N more" affordance; _sessionArchivePagingFilterActive() suppresses archived_limit during any search or project-filter activation; _syncArchivedSearchPagingRefresh triggers a full refetch on search-state edge transitions (inactive→active and back) to bring in the uncapped set without waiting for debounce.

Confidence Score: 4/5

Safe to merge. The archived-paging slice correctly keeps all visible rows intact, full counts for tab badges and the 'N remaining' prompt are computed before slicing, and separately-parameterised cache keys prevent cross-contamination.

Both changed files carry small rough edges: the normalization of archived_limit/archived_offset is repeated three times across the call chain despite _query_positive_int already guaranteeing clean integers, and _syncArchivedSearchPagingRefresh doesn't account for an already-active project filter, firing an extra HTTP request in that state. Neither causes incorrect data, but they warrant a follow-up cleanup pass.

api/routes.py (redundant inner normalization blocks) and static/sessions.js (spurious refetch path when _activeProject is set at search transition time).

Important Files Changed

Filename	Overview
api/routes.py	Adds archived_limit/archived_offset paging in _build_session_list_cache_payload and _query_positive_int helper; session-count metrics computed before slicing (correct for tab badge). Contains redundant inline normalization of values already validated by _query_positive_int.
api/route_session_list_cache.py	Extends _session_list_cache_key with archived_limit/archived_offset so differently-paged requests get separate cache slots. Normalization logic here mirrors what _query_positive_int already guarantees at the call site.
static/sessions.js	Adds _archivedRowsLoadedLimit state, _sessionArchivePagingFilterActive(), _syncArchivedSearchPagingRefresh(), and the 'Load more archived' DOM widget. Spurious refetch possible when _activeProject is already set and search transitions, but correctness is unaffected.
tests/test_session_list_long_history_perf.py	Adds two new backend tests: one end-to-end paging test confirming visible rows are never hidden and archived_count is unaffected by paging, and a cache-key isolation test.
tests/test_issue4766_sidebar_source_pushdown.py	Extends JS query-string tests to cover archived_limit injection, search/project suppression, and the _syncArchivedSearchPagingRefresh edge-transition behaviour via Node.js harness.
tests/test_sidebar_session_partition.py	Updates source-count assertions to match the new server-count sentinel path and adds archive load-more DOM assertions.
CHANGELOG.md	Release entry added for v0.51.757; correctly maintained by the release process per repo convention.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant U as User
    participant JS as sessions.js
    participant API as /api/sessions

    Note over JS: _showArchived=true, no search/project active

    JS->>API: "GET /api/sessions?include_archived=1&archived_limit=100"
    API-->>JS: "sessions[visible + archived[0:100]], archived_count=N"

    JS->>U: Renders sidebar (visible rows + first 100 archived)
    JS->>U: Load 100 more archived (N-100 remaining) button

    U->>JS: Clicks Load more
    JS->>JS: "_archivedRowsLoadedLimit = 200"
    JS->>API: "GET /api/sessions?include_archived=1&archived_limit=200"
    API-->>JS: sessions[visible + archived[0:200]]
    JS->>U: Renders sidebar (200 archived rows now shown)

    U->>JS: Types search query in sessionSearch
    JS->>JS: filterSessions() → _syncArchivedSearchPagingRefresh(q)
    JS->>JS: _archivedSearchPagingQueryActive: false → true
    JS->>API: "GET /api/sessions?include_archived=1 (no archived_limit)"
    API-->>JS: ALL archived sessions returned
    JS->>U: Client-side title filter finds matches beyond page 1

    U->>JS: Clears search input
    JS->>JS: _syncArchivedSearchPagingRefresh('')
    JS->>JS: _archivedSearchPagingQueryActive: true → false
    JS->>API: "GET /api/sessions?include_archived=1&archived_limit=200"
    API-->>JS: sessions[visible + archived[0:200]]
    JS->>U: Paged view restored (200 archived rows)

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant U as User
    participant JS as sessions.js
    participant API as /api/sessions

    Note over JS: _showArchived=true, no search/project active

    JS->>API: "GET /api/sessions?include_archived=1&archived_limit=100"
    API-->>JS: "sessions[visible + archived[0:100]], archived_count=N"

    JS->>U: Renders sidebar (visible rows + first 100 archived)
    JS->>U: Load 100 more archived (N-100 remaining) button

    U->>JS: Clicks Load more
    JS->>JS: "_archivedRowsLoadedLimit = 200"
    JS->>API: "GET /api/sessions?include_archived=1&archived_limit=200"
    API-->>JS: sessions[visible + archived[0:200]]
    JS->>U: Renders sidebar (200 archived rows now shown)

    U->>JS: Types search query in sessionSearch
    JS->>JS: filterSessions() → _syncArchivedSearchPagingRefresh(q)
    JS->>JS: _archivedSearchPagingQueryActive: false → true
    JS->>API: "GET /api/sessions?include_archived=1 (no archived_limit)"
    API-->>JS: ALL archived sessions returned
    JS->>U: Client-side title filter finds matches beyond page 1

    U->>JS: Clears search input
    JS->>JS: _syncArchivedSearchPagingRefresh('')
    JS->>JS: _archivedSearchPagingQueryActive: true → false
    JS->>API: "GET /api/sessions?include_archived=1&archived_limit=200"
    API-->>JS: sessions[visible + archived[0:200]]
    JS->>U: Paged view restored (200 archived rows)

Loading

_{Reviews (1): Last reviewed commit: "docs(changelog): page archived sidebar r..." | Re-trigger Greptile}

[greptile-apps]

Redundant normalization — archived_limit and archived_offset have already been validated to non-negative integers (or None) by _query_positive_int at the call site, so the try/except blocks here can never fire and the max(0, int(...)) guards are no-ops. The same duplication occurs in _session_list_cache_key. Simplifying to direct use of the incoming values would remove ~10 lines of dead code from both sites.

Suggested change

	if include_archived and archived_limit is not None:
	try:
	normalized_archived_limit = max(0, int(archived_limit))
	except (TypeError, ValueError):
	normalized_archived_limit = None
	try:
	normalized_archived_offset = max(0, int(archived_offset or 0))
	except (TypeError, ValueError):
	normalized_archived_offset = 0
	if normalized_archived_limit is not None:
	visible_rows_for_page = [s for s in visible_scoped_filtered if not s.get("archived")]
	archived_rows_for_page = [s for s in archived_scoped_filtered if s.get("archived")]
	scoped = visible_rows_for_page + archived_rows_for_page[
	normalized_archived_offset: normalized_archived_offset + normalized_archived_limit
	]
	if include_archived and archived_limit is not None:
	visible_rows_for_page = [s for s in visible_scoped_filtered if not s.get("archived")]
	archived_rows_for_page = [s for s in archived_scoped_filtered if s.get("archived")]
	scoped = visible_rows_for_page + archived_rows_for_page[
	archived_offset: archived_offset + archived_limit
	]

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[greptile-apps]

Spurious refetch when project filter is already active

_syncArchivedSearchPagingRefresh fires renderSessionList whenever _archivedSearchPagingQueryActive transitions between false and true, but _sessionArchivePagingFilterActive() already returns true when _activeProject is non-null — meaning archived_limit is already omitted from the query. If a user has a project filter active and then starts typing (or clears) a search, the false→true (or true→false) transition in _archivedSearchPagingQueryActive triggers an extra HTTP request even though the resulting query string is identical to the one already in flight. No data-correctness problem, but it is an avoidable extra round-trip in that specific mixed state. A guard checking !_activeProject before deciding to refetch would eliminate it.

[cutter-sh]

🎬 Cutter preview — PR #5241

Load more archived sessions — Sidebar pages through archived conversations, loading 100 more at a time up to 2000.