Slack: empty DM thread_ts hits conversations.replies(ts='') in fetch paths — add conversations.history fallback

[patrick-chinchill]

Summary

For Slack DM roots the encoded thread_ts is empty (slack:Dxxx:). This is produced by both:

• the DM message path — _handle_message_event (slack/adapter.py:2409), which is faithful to upstream handleMessageEvent (adapter-slack/src/index.ts:2158), and
• now the DM block-action path — _handle_block_actions (fix(slack): don't thread block-action responses in DMs (supersedes #133) #137), which is a documented divergence from upstream (see the Known Non-Parity row in docs/UPSTREAM_SYNC.md).

When a handler subsequently fetches that thread (event.thread.refresh(), pagination, fetch_message, open_modal-driven fetches), the Slack fetch paths call conversations.replies(ts=thread_ts) with no empty-thread_ts guard:

• _fetch_messages_forward / _fetch_messages_backward (slack/adapter.py:4118 / :4134)
• fetch_message (slack/adapter.py:4164)

With an empty DM thread_ts this sends ts='', which returns no replies / loses the original message context for DM action paths. (Raised by the Codex review on #137.)

Pre-existing + faithful to upstream (not a #137 regression)

This already affects every DM root (DM messages), independent of #137. It is also faithful to upstream: upstream's fetchMessages (index.ts:4178) calls conversations.replies(ts: threadTs) with no empty guard either. So a fix here is a divergence ahead of upstream and is codebase-wide (all DM roots), not specific to block-actions — which is why it's out of scope for #137.

Proposed fix

In the Slack fetch dispatch (fetch_messages, fetch_message), when thread_ts is empty (DM root), route to the channel-history path (fetch_channel_messages / _fetch_channel_messages_* → conversations.history) instead of conversations.replies. This covers DM messages and DM block-actions uniformly. Add a regression test for an empty-DM-thread_ts fetch_messages that asserts it does not call conversations.replies(ts="").

Also consider filing the same upstream (vercel/chat) so the empty-DM-thread_ts fetch handling — and the #137 block-action divergence — can be removed on a future sync.

Refs

• fix(slack): don't thread block-action responses in DMs (supersedes #133) #137 (Codex review, P2) — the block-action DM threading fix that surfaced this.
• docs/UPSTREAM_SYNC.md — "Slack DM block-action threading" Known Non-Parity row.

[coderabbitai]

🔗 Related PRs

#48 - fix: adapter hygiene — Slack stream await, Teams divider, public worker APIs [merged]
#83 - sync: upstream v4.27.0 (alpha 1) [merged]
#87 - feat(slack): dynamic bot_token resolver and custom webhook_verifier (vercel/chat#421) [merged]
#89 - fix(adapters): bug-fix sweep from upstream 4.27.0 (slack/discord/telegram) [merged]
#117 - fix(slack): surface files_upload_v2 confirmation in SentMessage.raw [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[patrick-chinchill]

Fixed in 0.4.31.1 (PR #175). Empty-DM thread_ts fetch now routes to conversations.history instead of conversations.replies(ts='') in fetch_messages/fetch_message; the non-empty (real thread) path is byte-identical. Documented as a Known Non-Parity row (divergence ahead of upstream, which shares the gap). Mutation-tested unit coverage in tests/test_slack_api.py. Note: chinchill's own DM flow uses its DBOS thread-history rather than the SDK fetch, so this matters for SDK consumers that explicitly fetch/refresh/paginate DM threads.