0.8.0

Changed

• Generated heading IDs now start with user-content-; update links to existing heading fragments when enabling headingAnchors.

Fixed

• Harden list parsing against pathological nesting.
• Use only the first decoded ASCII-whitespace-delimited token of a code fence's info string as its HTML class.
• Bound sparse-table cell expansion in toAst() and toXml() with MdParser\Exception instead of exhausting PHP memory.
• Keep heading ID deduplication fast for collision-heavy documents.
• Keep inline marks in a full-reference link label from pairing with text outside that label.
• Remove internal line markers from multiline link and image attributes in toInlineHtml().
• Treat an empty mdparser.parse_memory_limit value as the default 128M; explicit 0 and negative values remain unlimited.
• Ignore wikilink callbacks when wikiLinks is disabled.
• Keep nested footnote references working when definitions are listed out of reference order.
• Preserve ordinary links and balanced spans after an unresolved footnote label.
• Parse HTML block tags using their full names, so false prefixes no longer consume later Markdown.
• Keep indented paragraphs inside nonempty list items after two blank lines, including reference-only items.
• Preserve paragraphs inside nested blockquotes after an empty list item closes.

For contributors

• Refreshed vendored md4c to 0.6.0+git58c55b69 and dropped two local patches now present upstream.
• Test on macOS 15 in CI, matching the release binary build.