- "hypertext" describes a set of UI conventions. linking to things is not new
- rendering static documents is distinct from interactive dynamic ones
- forms are static documents
- rendering documents is not remotely related to their mode of transmission.
Post
URL format? make it completely opaque
static document rendering is not an internet problem, nor a programming problem. everyone in the world who writes has a stake. every format i know of is unspecified and ascii-centric.
i think org could work. not forked—from scratch. i think attribute containers are a good model for various affordances. that the document is readable without highlighting makes me feel that it could be empowering to write a document that will continue to parse the same years later.
"text" is problematic. but structure and layout are distinct. so:
- structural ontology: spanning, anchoring, containment, metadata
- an ASCII "realization" associates each structure to a grammar model with ASCII tokens
from this, we render:
- page splitting/wrapping
- order and visibility of sections
- fontification
so a renderer can be hooked up to a graphics device etc
@vtrlx is there a structured document format you like? i have been convincing myself above that "plain text" can be contextually meaningful, and that structure is distinct from realization.
html technically has a "structure", but it's only made available in the js dom. html is also a google product.
i want a text format because:
- the user has access to the whole document contents
- i do not know of any attempt to distinguish structure and realization like this—this is also how i want my programming language to work
@hipsterelectron I don't *like* Markdown but I do use it. I think XHTML was the unfortunate victim of a lot of timing issues and maybe if it was used as the target language for compilation it'd probably be solid. I think good candidates should natively support UTF-8, IMO I would go so far as to state that they should be designed such that UTF-8 is the only valid encoding for documents by virtue of being Good Enough and not locking out any language. Your opinion may differ though.
contra html, static and dynamic behavior are not the same. jarek duda tried really hard to make a use for wolfram's plotting widget, but scientific figures are static because they represent specific data. i believe dynamic layout is a much harder problem and overlaps very little with document structure.
i think every ui should be like a game ui. i think gamedevs have been working with graphics like i've worked with text. i fundamentally do not believe that dynamic content is a structured document.
servo's parallel layout engine: required a new language to do successfully. CSS evaluation is state-dependent, so parallelism is very hard.
ok, i'm not familiar enough yet here but i feel this works for now
what if your and my dynamic UI,,,,,
kissed each other in the vulkan API......
org-mode fucked up with the hierarchical sections. fucked it up more by using repeated ASCII *. complete waste of lexical syntax
- "hacker text format"
- is the structure hierarchical or good
- it's a good format sir
- dot jay peg
oh. another point (which latex demonstrates well):
- most layout is word flow
- some layout sections must support more precise arrangement conditions
i'm not quite sure here, but:
- i would very much like to avoid the "latex problem" and the "css problem"
- where because some feature necessitates pixel precision, the entire language must support same
perhaps a "layout graph" could describe pluggable structure, with each such structure supported by each renderer
the layout graph is distinct from a realization: each layout graph instance by the user uses the same delimiters, as per realization
@hipsterelectron have you looked at cuis erudite
@hipsterelectron I've never bought into the idea that structure and layout are separable:
- page splitting is absolutely structural
- order of sections is structural (what would even justify reordering sections for display?)
- font selection (esp mono vs not) affects alignment, which affects semantics
@hipsterelectron zooko's triangle says "yip" imo.