addition of third Fortran keyword

[nbehrnd]

Prior to perhaps prepare a PR, and because I'm not familiar with the Haskel language, I would like to rise the question about a third keyword Pandoc f90 to recognize snippets of Fortran here.

observation:

GitHub flavored markdown files allow to fence snippets of code blocks with a tag about the language they represent, for instance

print("hello")

or for Fortran

program test
   print *, "hello"
end program test

introduced by a line of with ```fortran

It equally is possible to use a tag ```f90

program test
   print *, "hello"
end program test

on GitHub in the markdown files, issue reports; or e.g. on the discussion board of fortran-lang.org. Beside brevity as tag, f90 as file extension is a typical indicator for the free form syntax since Fortran90 up and including Fortran2023 contrasting to FORTRAN77 / fixed form syntax, too.

Second, I see i) src/Text/Pandoc/Highlighting.hs has a langsList with the entry ("fortran","Fortran"), and ii) src/Text/Pandoc/Writers/MediaWiki.hs in the block starting by highlightingLangs = Set.fromList [ the consecutive lines "fortran", and "fortranfixed",.

question

To provide the third Fortran indicator to pandoc, would it suffice to add "f90", to src/Text/Pandoc/Writers/MediaWiki.hs alone?

[jgm]

langsList there is only for conversion between LaTeX listings language names and ours.

What pandoc version, and what format are you converting to? f90 should already work.

% pandoc  
``` f90
x = 9
^D
```
<div class="sourceCode" id="cb1"><pre
class="sourceCode f90"><code class="sourceCode fortranfree"><span id="cb1-1"><a href="#cb1-1" aria-hidden="true" tabindex="-1"></a>x <span class="kw">=</span> <span class="dv">9</span></span></code></pre></div>

[nbehrnd]

You are correct; both keywords to fence snippets of Fortran are in place, and both are functional.

However, while editing a draft about object oriented programming in Fortran (link), I observed their working differs a little bit. Eventually, I had to stick to f90 throughout the work (their: file pandoc_md.md) for a good representation of the content.

If one copies the following as snippet.md

Fenced with `f90` -- correct rendering

```f90
TYPE :: sorted_list
   PRIVATE
   TYPE(sortable) :: data
   TYPE(sorted_list), POINTER :: next => null()
CONTAINS
   FINAL :: delete_sorted_list
END TYPE
```

Fenced with `fortran` -- unresolved representation

```fortran
TYPE :: sorted_list
   PRIVATE
   TYPE(sortable) :: data
   TYPE(sorted_list), POINTER :: next => null()
CONTAINS
   FINAL :: delete_sorted_list
END TYPE
```

the visual result of the conversion with Pandoc 3.3 differs for either

pandoc snippet.md -s -o snippet.html
pandoc snippet.md -s -o snippet.pdf -t ms

A closer inspection of an intermediate file, for instance the .ms file, reveals

$ pandoc snippet.md -s -o snippet.ms
$ grep CONTAINS snippet.ms -n
125:\*[KeywordTok "CONTAINS"]
139:\*[CommentTok "CONTAINS"]

once the keyword is resolved as such, the other time (erroneously) processed as a comment. The dictionary accessed by the keyword f90 thus appears as more abreast to the current set of Fortran keywords, than the one of the lexer by a fence label fortran. Since this actually reflects correctly the history -- because contains was introduced as keyword with standard Fortran90 (reference) -- was the lexer launched by fortran intentionally frozen to standards up and including FORTRAN77?

---

Addition: I just notice a different visual representation on GitHub, too. Fenced with f90:

TYPE :: sorted_list
   PRIVATE
   TYPE(sortable) :: data
   TYPE(sorted_list), POINTER :: next => null()
CONTAINS
   FINAL :: delete_sorted_list
END TYPE

vs fenced with fortran:

TYPE :: sorted_list
   PRIVATE
   TYPE(sortable) :: data
   TYPE(sorted_list), POINTER :: next => null()
CONTAINS
   FINAL :: delete_sorted_list
END TYPE

2024-09-06_pandoc_Fortran_fences.zip

[jgm]

In skylighting, there are two distinct lexers (derived from KDE's syntax highlighters): fortran-fixed and fortran-free. The latter is used with f90, the former with fortran.

[nbehrnd]

In a thread about syntax lexers on fortran-lang.org, I amended my answer with your information (link). Thank you.