Pine/deepseek-harness
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases 1 Wiki Activity

fix(persistence): use normalized project directory names

Browse Source
This commit is contained in:
Turtle
2026-07-24 21:15:53 +08:00
parent 4e295221f3
commit c14f488b00
6 changed files with 45 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files
packages/session-persistence/session-persistence-jsonl/README.md
+2 -2
View File
@@ -6,7 +6,7 @@ The JSONL durable session-persistence backend — a concrete `SessionPersistence
```
<root>/
--<normalized-cwd>--<hash>/ # readable project directory (or _no-cwd/)
--<normalized-cwd>--/ # readable project directory (or _no-cwd/)
<encoded-id>/ # session-owned directory
session.jsonl.zstd # default: checksummed header frame + append frames
session.jsonl # only with compression: 'none'
@@ -14,7 +14,7 @@ The JSONL durable session-persistence backend — a concrete `SessionPersistence
- The first logical line is the immutable `SessionHeader` tagged `{ type: 'session', version, id, cwd?, createdAt, parentSession?, seedLength?, delegationDepth }`. `delegationDepth` is required on disk and is `0` for a top-level session; a missing or invalid value rejects the log. Every subsequent logical line is one storage record; `assistant/chunk` events are never dropped, and `seq` stays contiguous across the decoded log (`events[i].seq === i`).
- A storage record is a `SessionEvent` JSON verbatim, or — written only under `packChunks` — a **packed chunk row** (`text-chunks` / `reasoning-chunks` / `tool-call-chunks`; bare slash-less tags like the header's `session`, so row tags cannot be confused with event types): one line holding a run of ≥3 consecutive same-block `assistant/chunk` delta events, `seq0`/`time0` plus per-member `dt` gaps reconstructing every member's `seq`/`time` exactly. The lossless codec lives in `@deepseek-ai/dsh-session` (`packChunkRuns`/`decodeStorageRecord`) and whitelists exact shapes — anything unrecognized stores verbatim. Reading is layout-blind: `load` always decodes rows, so packed, unpacked, and mixed files load identically.
- The project directory keeps the normalized cwd readable for navigation and adds a short SHA-256 suffix so paths that normalize alike remain distinct. Its readable prefix is bounded for filesystem component limits. The configured root remains deployment-controlled: it may be project-local, shared, temporary, or centralized.
- The project directory keeps the normalized cwd readable for navigation and is bounded for filesystem component limits. Separator replacement and truncation are intentionally lossy, so cwd strings that normalize alike share a project directory; session ids still select distinct session directories. The configured root remains deployment-controlled: it may be project-local, shared, temporary, or centralized. The [project-session directory decision](../../../.agents/notes/implemented/architecture/2026-07-24-project-session-directories.md) records this tradeoff.
- Session ids are unvalidated branded strings, so they are injectively escaped to a single safe path segment before use (no traversal, no collision). The resulting directory is reserved for additional session-owned artifacts; discovery reads only the fixed transcript filename.
## Config