Ken's scraper (DragonKind/threshing-reddit) reads the megathread comments every 6 hours. It opens or updates one pull request from data/reddit-comments that changes data/public/reddit_comments.json. Each scraper PR needs a quick human review before merging.
Make the review easy
CI posts a summary comment on each scraper PR (or adds it to the PR body) with:
runs added, changed and removed since main;
runs that fail a game-rule check, with the comment id and which rule;
likely duplicates of spreadsheet rows. People often post the same run in the form and in a comment: same day, same answers;
result-page ids already in reddit thread, which are deduplicated automatically. This is for information.
The reviewer's options
Corrections go in collection/manual/reddit_comments.yaml, which is applied at build time:
Then merge. Rows touched by a correction lose the "automatically, not checked by hand" flag.
Docs
Add a short "Reviewing Reddit runs" section to collection/README.md.
Related: plausibility checks; "Use every possible run…"
Ken's scraper (`DragonKind/threshing-reddit`) reads the megathread comments every 6 hours. It opens or updates one pull request from `data/reddit-comments` that changes `data/public/reddit_comments.json`. Each scraper PR needs a quick human review before merging.
## Make the review easy
CI posts a summary comment on each scraper PR (or adds it to the PR body) with:
- runs added, changed and removed since main;
- runs that fail a game-rule check, with the comment id and which rule;
- likely duplicates of spreadsheet rows. People often post the same run in the form and in a comment: same day, same answers;
- result-page ids already in `reddit thread`, which are deduplicated automatically. This is for information.
## The reviewer's options
Corrections go in `collection/manual/reddit_comments.yaml`, which is applied at build time:
- `reject: [comment id]`: not a run, or a joke;
- `fix: {comment id: {question: answer}}`: misread answers;
- `add`: a run the scraper missed.
Then merge. Rows touched by a correction lose the "automatically, not checked by hand" flag.
## Docs
Add a short "Reviewing Reddit runs" section to `collection/README.md`.
Related: plausibility checks; "Use every possible run…"
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Ken's scraper (
DragonKind/threshing-reddit) reads the megathread comments every 6 hours. It opens or updates one pull request fromdata/reddit-commentsthat changesdata/public/reddit_comments.json. Each scraper PR needs a quick human review before merging.Make the review easy
CI posts a summary comment on each scraper PR (or adds it to the PR body) with:
reddit thread, which are deduplicated automatically. This is for information.The reviewer's options
Corrections go in
collection/manual/reddit_comments.yaml, which is applied at build time:reject: [comment id]: not a run, or a joke;fix: {comment id: {question: answer}}: misread answers;add: a run the scraper missed.Then merge. Rows touched by a correction lose the "automatically, not checked by hand" flag.
Docs
Add a short "Reviewing Reddit runs" section to
collection/README.md.Related: plausibility checks; "Use every possible run…"