diff options
Diffstat (limited to 'blog/content/notes/tech/document-formats.gmi')
| -rw-r--r-- | blog/content/notes/tech/document-formats.gmi | 97 |
1 files changed, 97 insertions, 0 deletions
diff --git a/blog/content/notes/tech/document-formats.gmi b/blog/content/notes/tech/document-formats.gmi new file mode 100644 index 00000000..385c0c0e --- /dev/null +++ b/blog/content/notes/tech/document-formats.gmi @@ -0,0 +1,97 @@ +# Document formats + +Most of the time, when writing a document, I want a document format with the following properties: + +* Fast to write using a plain text editor +* Easy to parse into an AST + +An AST is a programming-friendly representation of a document. ASTs reduce the effort required to write tools such as a program that validates links in a document. Ideally, ASTs contain information to track a document element to the position it occupies in the original document. With this information, if you write a tool such as a spell checker, then you can highlight misspelled works precisely in the original document. + +On top of that, some features that I don't always need: + +* Math support +* Sophisticated code blocks. For example, being able to highlight arbitrary parts of code blocks (not syntax highlighting). +* Diagram support + +## Existing formats + +### Markdown + +* Easy to write using a plain text editor +* Has good AST parsers with position information +* Has math support +* Does not support sophisticated code blocks +* There are many extensions with support for math, diagrams, and many others +* Is very popular and supported everywhere +* However, there is a wide variety of variants and quirks +* Especifically, because Markdown was not designed with parsing in mind, so tools based on different parsers can have differences in behavior + +### Djot + +=> https://djot.net + +It is very similar to Markdown, except: + +* It is designed for parsing, so independent parsing implementations are very compatible with each other +* It is not so popular, so there are less extension and tool support + +### AsciiDoc + +=> https://asciidoc.org + +Compared to Markdown: + +* It's more complex to write, but mostly because it's different and more powerful +* There are attempts to write better parsers, but good parsers with position information are not available yet +* Supports sophisticated code blocks +* It has a smaller ecosystem than Markdown, but many good quality tools such as Antora + +### Typst + +=> https://typst.app + +Checks all my boxes, except: + +* It is designed for parsing and it has an AST, but it is not easy to access +* Currently Typst is very oriented towards generating paged documents (e.g. PDF) +* It includes a full programming language, which is mostly good (very extensible), but this might increase complexity undesirably + +Typst is very new and is not yet very popular. + +=> https://codeberg.org/haydn/typesetter Typesetter is a desktop application that embeds Typst, so no additional setup is needed. However, Typesetter is only available as a Flatpak. + +### Verso + +=> https://github.com/leanprover/verso + +A Markdown-like closely tied to the Lean programming language. + +* Eliminates ambiguous syntax for easier parsing and is stricter (not all text is valid Verso) +* Has a (Lean) data model +* Designed for extensibility + +### TODO: other formats + +=> https://github.com/nota-lang/nota Nota (a document language for the browser) +=> https://github.com/christianvoigt/argdown Argdown (for argumentation) +=> https://github.com/podlite/podlite Podlite +=> https://orgmode.org Org Mode (an Emacs-based tool based on a lightweight markup language) +=> https://github.com/nvim-neorg Neorg (similar to Org Mode for Neovim) +=> https://github.com/sile-typesetter/sile Sile (typesetting system) + +## Creating your own formats + +=> https://github.com/spc476/MOPML Someone created its own lightweight format using Lua and PEGs. +=> https://tratt.net/laurie/blog/2020/which_parsing_approach.html "Which parsing approach" has information about choosing parsing approaches. + +## About gemtext + +=> https://geminiprotocol.net/docs/gemtext-specification.gmi + +Gemtext is an extremely minimalistic markup language designed for use with the Gemini protocol (an extremely minimalistic protocol similar to HTTP). + +The Gemini protocol and gemtext are intentionally designed to limit their power, in my opinion as a comment on the web. + +This document is gemtext-native for use in my own minimalistic publishing system. + +I also use it as a statement, although the limitations of gemtext can be significant in technical writing. For example, gemtext has no inline links, no inline verbatim code, only three levels of headings, etc. |