Skip to content

fix: do not save the page's render error as the answer - #79

Merged
youdie006 merged 1 commit into
mainfrom
fix/answer-render-failed
Oct 30, 2026
Merged

youdie006 merged 1 commit into
mainfrom
fix/answer-render-failed

Conversation

@youdie006

Copy link
Copy Markdown
Owner

Problem

When ChatGPT's newer answer renderer cannot draw an answer, it shows div[role=status][data-dil-render-failed] reading "This response couldn't load" in its place. prodex read that sentence as the answer: the consult reported done with request_verified: yes, and pro browser recover saved the same sentence again, because the tab was already on the thread and was not loaded again.

Measured 2026-10-30 in the browser container (prodex-browser-browser-1). Chromium there could not start threads (pthread_create: Resource temporarily unavailable at the moment of the send; the container sat at its 256-pid limit), and the conversation held the finished answer ("OK", finished_successfully) while the page showed the error.

Change

  • answerExpression reports answerRenderFailed for the assistant turn paired with the latest user turn.
  • hasFreshChatGptAnswer and hasPartialChatGptAnswer refuse such a turn, so the error text is never returned as a finished or partial answer.
  • A send stops as answer_render_failed after two reads showing it. The blocker is not retryable, and it names prodex pro browser recover --target-url <thread> --request-id <id>.
  • Recovery marks the document, loads the thread once more and reads the new document. If the error is still there, it stops as answer_render_failed (retryable) instead of saving it.

Verification

  • tsc --noEmit, npm run build, npm test: 1846 passed, 3 skipped.
  • New tests:
    • the expression flag, including an earlier failed turn that must not count;
    • both predicates;
    • a send that blocks;
    • recovery that reloads once and returns the redrawn answer;
    • recovery that stops when a reload does not help.
  • Live on WSL (host browser), with the measured markup put into a probe answer:
    • 0.40.48 recover returned "This response couldn't load".
    • This build's recover reloaded once and returned "OK".
    • A send whose answer was marked failed in every document stopped as answer_render_failed with the recover command.
    • While the failure persisted, recovery stopped with answer_render_failed.
    • Once it no longer did, recovery returned "PROBE7731".

The container's thread limit is fixed separately on container/redesign-patches.

When ChatGPT's newer answer renderer cannot draw an answer, it shows
div[role=status][data-dil-render-failed] reading "This response couldn't
load" in its place. prodex read that sentence as the answer: the consult
reported done with its request verified, and `pro browser recover` saved
it again because the tab was already on the thread and was not reloaded.

Measured 2026-10-30 in the browser container, where Chromium could not
start threads: the conversation held the finished answer while the page
showed the error.

- answerExpression reports answerRenderFailed for the paired assistant turn.
- hasFreshChatGptAnswer and hasPartialChatGptAnswer refuse such a turn.
- A send stops as answer_render_failed (not retryable, names recover with
  the thread and request id) after two reads showing it.
- Recovery marks the document, loads the thread once more, and reads the
  new document; if the error persists it stops as answer_render_failed.

Verified: tsc, build, npm test (1846 passed, 3 skipped). Live on WSL with
the measured markup injected: 0.40.48 recovered the error text; this build
recovered "OK" after one reload, blocked a send whose answer was marked
failed, stopped recovery while the failure persisted, and recovered the
answer once it no longer did.

Claude-Session: https://claude.ai/code/session_01EMqNwnnhHXznd2P2PDsaDV
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 30, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-30T08:08:17.596127Z 888f61e PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to 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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@github-actions

Copy link
Copy Markdown

What this touches

  • browser adapter
  • tests

3 file(s), 203 line(s) changed. Tests changed.

This is an automated first pass, not a review. CI is the gate; a person decides.

@youdie006
youdie006 merged commit eab9340 into main Oct 30, 2026
7 checks passed
youdie006 added a commit that referenced this pull request Oct 30, 2026
…iner image

The page drew "This response couldn't load" in place of an answer
(div[role=status][data-dil-render-failed]) in this container on
2026-10-30, and the send saved it as the answer with its request
verified; recovery saved it again. A send now stops as
answer_render_failed with the recover command, and recovery loads the
thread once more and reads the redrawn answer.

The source change was applied by exact replacement; this branch lacked
main's waitForReloadedDocument, which comes along unchanged. The five
tests from main apply as-is.

Verified: tsc, build, npm test (2150 passed, 3 skipped), npm run
release:verify ok.

Claude-Session: https://claude.ai/code/session_01EMqNwnnhHXznd2P2PDsaDV
youdie006 added a commit that referenced this pull request Oct 30, 2026
Merging PR #83 after #79-#81 left two "## Unreleased / ### Fixed"
headings: git joined the branch's new section and main's without a
conflict. The four entries are now one list.

Claude-Session: https://claude.ai/code/session_01EMqNwnnhHXznd2P2PDsaDV
youdie006 added a commit that referenced this pull request Nov 1, 2026
Both containers now run container/redesign-patches f2d13e6 (ports of
#79-#84 and the 1024 process limit): M3 image 4bc8fac19752, WSL image
e2e3523b640a. Each swap waited for the send lock; both containers report
logged in, smoke answered with the request verified, and a dry-run id
carries no prompt words.

Claude-Session: https://claude.ai/code/session_01EMqNwnnhHXznd2P2PDsaDV
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant