feat(tui): add stats display mode for live thinking - #2921
feat(tui): add stats display mode for live thinking#2921jfwang-moonshot wants to merge 1 commit into
Conversation
Add a thinking_live_display preference in tui.toml. The default "preview" keeps the scrolling two-line tail while thinking streams; "stats" hides the text and shows an approximate token count plus elapsed thinking time, leaving a one-line "Thought for …" summary once thinking finishes.
🦋 Changeset detectedLatest commit: 2c02d8f The changes in this PR will be included in the next version bump. This PR includes changesets to release 1 package
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2c02d8f7cc
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| render_latex = ${String(config.renderLatex !== false)} # false keeps LaTeX math in assistant messages as raw source | ||
| disable_paste_burst = ${String(config.disablePasteBurst)} # true disables non-bracketed paste-burst fallback | ||
| cache_expiry_hint = ${String(config.cacheExpiryHint !== false)} # false disables the "cache expired" dialog on resume / idle submit | ||
| thinking_live_display = "${config.thinkingLiveDisplay ?? 'preview'}" # "preview" scrolls the last lines while thinking streams; "stats" shows ~tokens and elapsed time |
There was a problem hiding this comment.
When a user has thinking_live_display = "stats" and then runs a path that rewrites tui.toml without touching this option (for example /theme, /editor, the update preference, or the cache-expiry “never” action), those paths save { ...currentTuiConfig(host), ... }, but currentTuiConfig still omits thinkingLiveDisplay. Because this line serializes an omitted value as "preview", those unrelated saves silently reset the user's stats preference; include the new field in currentTuiConfig before defaulting on write.
Useful? React with 👍 / 👎.
| true, | ||
| 'live', | ||
| state.ui, | ||
| state.appState.thinkingLiveDisplay ?? 'preview', |
There was a problem hiding this comment.
When thinkingLiveDisplay is stats, replayed sessions also hit this constructor through SessionReplayRenderer.flushAssistant, which calls onThinkingUpdate(thinking) and then immediately onThinkingEnd() for historical messages. That creates a fresh live component with startedAt = Date.now() and finalizes it right away, so every resumed thinking block is hidden behind a misleading Thought for 0s summary instead of preserving the stored thinking preview; keep replay/finalized rendering out of the live stats mode or pass a replay-safe elapsed value.
Useful? React with 👍 / 👎.
| if (this.liveDisplay === 'stats') { | ||
| // No tokenizer is available to the client, and reasoning-token usage | ||
| // only arrives at step end — so the live count is a chars/4 estimate. | ||
| const approxTokens = Math.ceil(this.text.length / 4); |
There was a problem hiding this comment.
skews low for CJK characters
1 participant
Related Issue
Resolves #287
Problem
What changed
Add a thinking_live_display preference in tui.toml. The default "preview" keeps the scrolling two-line tail while thinking streams; "stats" hides the text and shows an approximate token count plus elapsed thinking time, leaving a one-line "Thought for …" summary once thinking finishes.
Checklist
gen-changesetsskill, or this PR needs no changeset.gen-docsskill, or this PR needs no doc update.