Notes for Week 27 of 2020
Sunday, July 12, 2020, 5:01:04PM
Nothing like getting something great down on “paper” to motivate me to do everything else to make it a reality.
I love where the design of Knowledge Nodes is going — particularly the dynamic Node types that will ultimately display word maps by usage, dependency trees, 2nd and 3rd level subscription recommendations, and even some machine learning common usage patterns.
The idea that I will be able to visually display my entire RWX Knowledge Network out to two or three levels is so absolutely awesome that I want to make it today!
It is so hard to describe to people (including my wife) but when they finally see this in action — especially after we get a few hundred people on the RWX Knowledge Net — no one will even really want to use Google again. Why use Google when I can search all the content from everyone in my customized knowledge network instantly and locally?
You won’t even need a fucking web browser! Graphical knowledge net browsers can run entirely on SSH connections and fallback to HTTPS because they don’t even need to access the Internet that much, only to check in. This was the vision of RSS that never really came to pass.
Seriously, this has very real potential to change the world in ways I have always wanted. I hope Aaron would be proud. May this whole effort succeed, for him.
Sunday, July 12, 2020, 8:58:51AM
What was I thinking? Just use
pandoc you idiot! (That’s me shouting at me.) Even if did manage to create a PEG Golang parser (which I really want to do because it would be fun) it makes no practical sense when an entire project of brilliant people are supporting Pandoc every day.
So what if
pandoc is a subprocess every time I call it. That’s nothing these days. Will speed rival Hugo? No, but it won’t be as brain-dead as Hugo either. Hugo was made by engineers, not academic and scientific writers. That is why Pandoc has such a strong LaTeX influence and why R make Pandoc its official documentation language.
If and when I hit a performance issue there are plenty of other ways to address them:
- Create a daemon that automatically builds anything when changed. I need this anyway to preview the rendering in pseudo-real-time.
- Cache AST parsing and word crunching so that it just has to be recombined in a
Plus if I make this all about Pandoc I might even get some buy-in from the Pandoc community itself, since Knowledge Net will be fundamentally dependent on it. But hey, the WorldWideWeb was created by academics to share information so why not?
Saturday, July 11, 2020, 8:38:29AM
Woke with an idea to build the Pandoc AST into the core RWX Knowledge Net specification. In fact, I can use Pandoc in two stages to prepare for its eventual replacement as a parser and renderer, first to generate the JSON AST (loaded into a Go struct) and then to render by piping the JSON AST into the
pandoc app for rendering. It will make the initial tool need double the calls to
pandoc, which I was trying to reduce, but for a very good reason. Later I can keep
pandoc as a renderer (since people will always want a fully configurable Pandoc rendering option) but I will replace
pandoc as a parser with a native Go PEG parser.
Friday, July 10, 2020, 9:23:42PM
The more I dig deep into the Goldmark Golang Markdown library the more I realize how bad it is, mostly because I have seen a very good AST already (Pandoc). The author of Goldmark probably doesn’t even know Pandoc exists. So much for my Goldmark fascination. I sometimes hate that I can actually look under the hood and see the shit for what it is a lot of the time. Just the crappy autoidentifier code was enough to throw it out for something that is actually specified (Pandoc). Every quirk that Pandoc has is nothing compared to hack that Goldmark is. Problem is, all the rest are even worse — especially Blackfriday.
Friday, July 10, 2020, 12:06:31PM
vim-pandoc plugin gets so many things wrong. I’m seriously annoyed even though I should be grateful it exists in the first place. I definitely need to make my own
vim-datadoc plugin from scratch but built on the ideas in
The official name for the syntax of all RWX Knowledge Nodes will be RWX DataDoc Markdown. The emphasis is equally on the YAML data, not the limited Pandoc Markdown of the document which is good since the YAML data is never optional in a DataDoc.
Again, I have a lot of editing to do.
Friday, July 10, 2020, 11:53:26AM
This running has resulted in an explosion of brain power. I forgot how much a slow recovery walk after a good 20-30 minute just hyper-charges my brain with oxygen and great ideas — even a new personal mantra that’s fun to chant to myself while running.
Friday, July 10, 2020, 9:49:45AM
I’m really struggling with a good alternative to Pandoc divs (
:::). I have really come to love them in my writing but the more I think about them the more I realize pretty much everything I ever put into one deserves its own knowledge node instead and a link. I have this problem because I’m such a parenthetical writer. I am always by-the-way-ing this or that instead of just focusing on the immediate topic. At least with a link people can follow the tangents if they really want.
I’m pretty sure serious publications handle this with the square bracket syntax that refers to other articles.
[[Here’s yet another story about this.]]
I like that approach the best I think. It is well-established and looks pretty in my Vim editor because the brackets still force the text to be highlighted even though they are escaped out with backslash.
Plus when parsing it all I have to look for
Looks like I have a lot of rewriting to do. The good news is that I will have a locked down writing style standard that I and anyone else can follow if they want.
Friday, July 10, 2020, 9:39:28AM
Sometimes flexibility is the fucking devil. For example, allowing the RWX Knowledge Net (yeah, that’s the latest name I’ve been using) to be anything but YAML and limited Pandoc Markdown is a huge disadvantage.
We can really see this with Hugo. It supports no less than three “front-matter” (God I hate that term) structured data formats. What a fucking nightmare! I am so glad to be free of that shitty design.
If there is one thing the last three decades of Web technology have taught us it is that flexibility kills adoption and inhibits sharing. This is why we still have YAML 1.2 spec from like more than 10 years ago. Smart designers do not want flexibility when it comes to things like structured data formats. They want consistency.
So the RWX KN spec only allows a single YAML header and optional body of limited Pandoc Markdown. Both are very solid standards that have been in use for years.
Friday, July 10, 2020, 8:53:58AM
Back to masturbatory note taking. The idea that I could whip up a blog for every one of these random, spontaneous thoughts was just stupid. Notes are the organic stuff that bubbles up and needs to be caught right away before it fades. Then an article (or blog) can be made from that.
By the way, I fucking hate the word blog. It has lost all specificity. Usually when people say “blog” they mean “piece” or “article” or “chapter.” The term is dead to me. I will silently look at others who use it and quietly mock them inside. It’s my prerogative.
No, instead I hereby asset the following moniker for this random brain dumping. From now and henceforth it shall be called — wait for it — note taking. I know. The level of my brilliance cannot be grasped. *tongue firmly in cheek*