Kept superseded summaries and preserveData durable while deriving forward transcript elision from the active compaction.
Added branch and rewind coverage for snapcompact archives and OpenAI remote replacement history.
Fixes#4090
Route loadSessionMessagesReadOnly through transcript mode (collapsed to the
latest compaction) so history:// transcripts of on-disk sessions retain
failed/aborted assistant tails that the provider-context builder now drops.
resolveBlobRefsInEntries handed every non-session entry to the recursive
async resolvePersistedBlobRefs walk, allocating and awaiting child promises
even for plain-text entries with no blob:sha256: refs. On large text-heavy
histories this dominated the blob_resolve phase of session open.
Add a cheap synchronous containsBlobRef precheck that early-exits on the
first ref and allocates nothing. Interleave the precheck with per-entry
initiation so positive entries still start resolution at the same relative
point as the old filter+map schedule (a later entry that gains a ref during
an earlier BlobStore.get is still scanned after that mutation).
Blob-free N=5000 fixture: blob_resolve median 19.5ms -> 1.1ms, zero
BlobStore.get calls.
Fixes#5922
- Normalized completed image_generation_call results into assistant image blocks.
- Persisted image bytes through the session blob store and rendered them in live, replay, ACP, proxy, telemetry, and HTML paths.
- Added response normalization, persistence, and TUI rendering regressions.
Fixes#4768
Address review: the per-chunk Buffer.from(chunk).toString("utf8") string concat corrupted multibyte sequences (e.g. ✓, emoji) that straddled a chunk boundary — silent data loss in exactly the ≥8MiB long sessions that take this path. Keep the buffer as a Uint8Array and pass it directly to Bun.JSONL.parseChunk (it accepts typed arrays and parses UTF-8 natively), so no decoding happens until a complete record is parsed. Only the title-slot first line is decoded, and only after the full line (up to its '\n') is buffered, so it is a complete UTF-8 sequence.
Also: drop the defensive 'read || 0' (read is number per typings); add a multibyte multi-chunk parity test (>128KiB fixture) that would have caught the bug.
The ≥8MiB session-load path (loadEntriesFromFileStream) used node:readline + per-line JSON.parse, while the common (<8MiB) path already used Bun's native Bun.JSONL. The large-session path is now the lone outlier getting the slower parser.
Rewrite loadEntriesFromFileStream to a lenient streaming Bun.JSONL.parseChunk loop: stream the file via Bun.file().stream() into a bounded buffer (memory guard preserved — the file is never fully loaded, which is why the 8MiB threshold exists), skip malformed records to the next newline instead of throwing, peel/fold the non-JSON first-line title slot, terminate a trailing newline-less record, and keep isEnoent→empty. Removed the now-unused node:fs + node:readline imports.
Measured ~9.5ms → ~6.0ms (-37%) loading a 10MiB valid session; the remaining cost is I/O + native parse (floor).
Correctness: new session-loader-stream.test.ts asserts the stream path produces byte-identical entries + titleSlot to parseSessionContent (the common-path parser) across title-slot / blank / malformed / no-trailing-newline / missing-file cases.
Restrict the supersede sweep to compactions on the path from the current leaf so a newer compaction never rewrites a sibling branch's still-current summary or drops its preserveData. Streaming load now collects the active-branch ids before eliding instead of trampling sibling compactions encountered in file order.
Refs #3789
Stream large session loads, elide superseded compaction payloads, skip synchronous rewrites when the append-only file is already current, and provide usable picker previews for developer-started forks.
Fixes#3789
- Added a fixed-width title slot system to serialize and persist session titles across physical files and backend storage.
- Integrated automated session title updates triggered by todo replan operations using conversation history context.
- Extended storage interfaces across memory, file-system, Redis, and SQL backends to support independent title metadata updates.
- Implemented title metadata parsing within session loaders and list utilities to ensure accurate retrieval and display.
- Changed provider delta input comparisons in `buildResponsesDeltaInput` to use `Bun.deepEquals` instead of stringified JSON checks.
- Updated session message diffing to return early on length changes and compare normalized entries with deep equality.
- Replaced JSON-string assertions in the issue-966 repro test with `Bun.deepEquals` for stable equality checks.
- Added Matplotlib figure PNG rendering and display tracking in Python runner to emit PNG output immediately when figures are displayed via display(fig).
- Extended session persistence to externalize oversized image payloads in both content and details.images, enabling tool result images to survive session reload.
- Enhanced session loader to resolve image data payloads and blob references across content and details.images during session reconstruction.
- Added image cache invalidation in TUI image component when image protocol, cell dimensions, or Kitty Unicode placeholder mode changes.
- Added comprehensive test coverage for Matplotlib display, image persistence across reload, and TUI image rendering with protocol and dimension changes.
- Added session-domain modules and exports for session-entries, context, listing, loader, and migrations.
- Changed persistence to async append writes plus writeTextAtomic, removing sync line APIs.
- Added compaction-aware session context rebuild with dangling tool-call cleanup.
- Added resumable session resolution with status inference, id/stem/suffix matching, and backup recovery.
roboomp