Hacker News Reader

Comment by thangalin | Hacker News Reader

Here are a few Markdown documents I've typeset into PDF:

* https://impacts.to/downloads/lowres/impacts.pdf

* https://pdfhost.io/v/4FeAGGasj_SepiSolar_Highlevel_Software_...

* https://dave.autonoma.ca/blog/2020/04/28/typesetting-markdow...

My text editor, KeenWrite[1] (see tutorials[3] for details), converts from Markdown to XHTML. The XHTML is then typeset using the ConTeXt typesetting software to create a PDF. The PDF is stylized using a theme[2].

What Markdown is missing, in particular flexmark-java, is a standard CommonMark extension for cross-references and citations[4].

[1]: https://github.com/DaveJarvis/keenwrite

[2]: https://github.com/DaveJarvis/keenwrite-themes/tree/main/exa...

[3]: https://www.youtube.com/playlist?list=PLB-WIt1cZYLm1MMx2FBG9...

[4]: https://talk.commonmark.org/t/cross-references-and-citations...

View on HN

This is a great opportunity to extol the superior text layout capabilities of Latex. In your third pdf [1], look at page 2, para starting from "No doubt...". Compare to the same rendered in plain latex [2]. Two problems that jump immediately.

* There are six word breaks in your pdf, zero in the latex version. (well, good-nature is a compound word in the original text).

* Notice the "river" [3] of spaces that runs downs from "Enfield"

I have no idea why other fixed layout engines don't adopt the superior text-layout algorithms of latex.

[1] https://dave.autonoma.ca/blog/2020/04/28/typesetting-markdow...

[2] https://i.imgur.com/Rs7ai0G.png

[3] https://en.wikipedia.org/wiki/River_(typography)

> There are six word breaks in your pdf, zero in the latex version. (well, good-nature is a compound word in the original text).

I use ConTeXt. ConTeXt uses the same Knuth-Plass algorithm for line breaks as LaTeX. I've allowed ConTeXt to hyphenate liberally. There is a setting to control hyphenation. See lesshyphenation, morehyphenation:

https://wiki.contextgarden.net/Command/setupalign

> Notice the "river" [3] of spaces that runs downs from "Enfield"

ConTeXt probably has tweaks for this as well. Keep in mind, I typeset Jekyll and Hyde in 2020 and am not a ConTeXt expert; ConTeXt has had countless improvements since then.

> adopt the superior text-layout algorithms of latex

Pretty sure ConTeXt is capable of producing the same quality of document as LaTeX.

* https://wiki.contextgarden.net/Documentation

* https://tex.stackexchange.com/questions/36/differences-betwe...

* https://tug.org/levels.html

KeenWrite looks cool and useful; thanks for the reference.

It does appear that the real heavy lifting in this workflow is still all done by TeX, and that Markdown (with lots of extensions) is more or less a database of blobs to feed the template engine.

I would suggest the question: Is what you have at the end really a Markdown document if it has no context without KeenWrite's extensions or templates? I think the article under discussion is a bit pedantic, but I'm pretty sure the author would say that what you end up with is a KeenWrite document.

> I'm pretty sure the author would say that what you end up with is a KeenWrite document.

Documents written in KeenWrite can be typeset using tools that already exist:

* https://pandoc.org/

* https://yihui.org/knitr/

* https://github.com/jfisteus/html2xhtml

I've kept compatibility with pandoc (CommonMark) and knitr a priority; I wrote KeenWrite to replace a shell script that was calling out to those software packages. My Typesetting Markdown blog posts dive into that script and programs:

https://dave.autonoma.ca/blog/2019/05/22/typesetting-markdow...

Further, KeenWrite only supports plain TeX, which means that the typesetting system can be any software that is TeX-compatible: LaTeX, ConTeXt, LuaTeX, ExTeX, KaTeX, etc.

Yes I was being a bit pedantic as well! :)

Seriously though I do like the workflow you've put together with KeenWrite. I have a potential application where it might be a great fit to replace some markdown->pdf workflows I have built. In my case, I use a Front Matter header to pass structured data along with the markdown to generate templated HTML which can then be "typeset" with the extensions provided by HTMLDOC or wkhtmltopdf.

Thanks again for posting and replying. Pleased to make your acquaintance.

> Thanks again for posting and replying. Pleased to make your acquaintance.

Likewise!

> I use a Front Matter header to pass structured data

The impetus for developing that shell script (ergo KeenWrite) was using variables in documents. KeenWrite injects structured, interpolated YAML strings back into documents by replacing Moustache syntax (or `r# ...` syntax for R docs). This means that the same structured data can be reused for multiple documents, allowing for single sources of truth and a clean separation of content from presentation. Also, variables can be used in diagrams.

As for Front Matter, data can be passed in from the command line[1]:

    -v metadata.yaml \
    --metadata="title={{book.title}}" \
    --metadata="author={{book.author}}"

Under the hood, KeenWrite places the metadata inside the intermediate XHTML document as <meta> tags inside the <head>. These are parsed by ConTeXt using an XSLT-like syntax that converts them into TeX instructions[2].

Do post any questions you may have on the discussions forum[3].

[1]: https://github.com/DaveJarvis/keenwrite/blob/main/docs/cmd.m...

[2]: https://github.com/DaveJarvis/keenwrite-themes/blob/main/xht...

[3]: https://github.com/DaveJarvis/keenwrite/discussions/

I'd be interested in KeenWrite if it supported MacOS. Your pandoc/ConTeXt guide is top notch. ConTeXt is very underrated.

It's written in Java. People have run it on MacOS. I don't have a .dmg file, though.

https://github.com/DaveJarvis/keenwrite#other

If that doesn't work for you, please open an issue and describe what you'd like for MacOS.

I stand corrected. Thank you!

Reply on news.ycombinator.com