Regression: Markdown to LaTeX conversion doesn't always recognize unnumbered headers

[bwhelm]

Pandoc versions >= 1.18 sometimes do not recognize unnumbered headers, whereas pandoc 1.17.2 does in the same conditions. This happens with a 4th-level header and with --chapters (or --top-level-division=chapter) specified. See pandoc runs below.

> pandoc --version
pandoc 1.17.2
Compiled with texmath 0.8.6.4, highlighting-kate 0.6.2.1.
<snip>

> pandoc -t latex --chapters
#### Testing {-}
\subsubsection*{Testing}\label{testing}
\addcontentsline{toc}{subsubsection}{Testing}

This is correct -- with \subsubsection*. Compare this with:

> pandoc --version
pandoc 1.19
Compiled with pandoc-types 1.17.0.4, texmath 0.9, highlighting-kate 0.6.3
<snip>

> pandoc -t latex --chapters
[pandoc warning] --chapters is deprecated. Use --top-level-division=chapter instead.
#### Testing {-}
\subsubsection{Testing}\label{testing}
\addcontentsline{toc}{subsubsection}{Testing}

Here, the * is wrongly omitted from \subsubsection.

This does not happen without specifying --chapters (or --top-level-division=chapter); nor does it happen with 1st--3rd level headings.

[jgm]

I think I see what is happening.

+  let star = if unnumbered && level < 4 then text "*" else empty

This is to avoid adding a star to \paragraph, which can't take one.
Normally a level 4 header is a \paragraph, but when you use --chapters it becomes a \subsubsection. So this needs some revision.

[tarleb]

Both `\part*{}` and `\chapter*` exist and have the expected functionallity; seems that using the primed `level'` variable instead of `level` gives the desired behavior. Seems like I missed it when I was touching that region. John, do you prefer a PR or a direct push to master in cases like this?

[jgm]

@tarleb Go ahead and do a direct push! +++ Albert Krewinkel [Dec 03 16 14:19 ]:

…

Both `\part*{}` and `\chapter*` exist and have the expected functionallity; seems that using the primed `level'` variable instead of `level` gives the desired behavior. Seems like I missed it when I was touching that region. John, do you prefer a PR or a direct push to master in cases like this? — You are receiving this because you commented. Reply to this email directly, [1]view it on GitHub, or [2]mute the thread. References 1. #3272 (comment) 2. https://github.com/notifications/unsubscribe-auth/AAAL5K2EWCk7hblz-6qt3uci9bw2zi43ks5rEesGgaJpZM4LDVI7