◊(Local Yarn Code "Artifact [5f22f4e6]")

Artifact 5f22f4e65c8dff9dff60a59cd883687aa5d051105c6cbd45387ea1b4ba42f33f:

Wiki page [Ideas Scratchpad] by joel on 2021-11-26 23:29:12.
D 2021-11-26T23:29:12.726
L Ideas\sScratchpad
N text/x-markdown
U joel
W 1520
# Ideas for rebooting the project

I’m a little better at this than I was when I started the project, and I have some ideas that will make the site simpler, quicker, and more elegant.

The SQLite cache would be removed.

Articles would keep all the values needed to render themselves and their notes in their metas, rather than reparsing the doc at render time.

Articles and notes would compile to an AST rather than directly to HTML or LaTeX or MP3 or whatever (no “poly tags” required). 

At compile time, an article would serialize each of its notes, in the form of a listing prefab struct, to a separate .rktd file in a subfolder. The filename would be formatted as YYYYMMDDHHMMSS_id to allow for easy sorting by datetime.

After all updated articles are compiled, an indexer would collect these sort tables and serialize them to .rktd files:

  * Dictionary of main index entries → subentry  + list of link targets
  * Hash table of series → articles
  * Ordered list of all articles and notes by date

There would be a clear cross-referencing model:

* A *pin* is an index entry. In a body of text, it links to the index entry.
* A *def* is also a pin that can be cross-referenced directly in another text. 
    * In a body of text, it is italicized, prefixed with a manicule, and links to the index entry. 
    * Every title of an article becomes its own def. 
* A *ref* simply links directly to a *def* (not to the index).
    * Ref links get resolved at template render time.
Z 308cfcf27b3d1d5c06e3e4c9e1c43ae6