breck7 / jtree Goto Github PK

"FYI The only DSL with a readme is the poop one."

At the least we should have something like: knexjs or sizzle(cheerio).

Beyond that, given that Tree Notation has some unique geometric properties, there might be a new language that can take advantage of those.

Beyond that, there may be another type of query language for Grammar backed languages.

[] Integrate checkLinks.sh into CI.
[] Setup smarter 404 redirects if possible
[] create 404 jekyl pages

Might require having the diff work done first.

• Then we can store diffs of grammars versions.
• Then we can maintain a history of all the versions of a language.
• Then we can use those for ensuring typesafe upgrades
• Should be done using convention over configuration?

Request from X. Have unit types, like "40ml" and allow pattern matching at the cell level. So you could have unnamed, any order argument functions.

[] table
[] comment
[] compile to HTML or markdown
[] ship as a browser product.
[] provide instructions for using in browser
[] ship with ohayo

How does is this project different from a intermediate representation like JVM byte code or .net IR?

The FAQ for this project says treenotation is like a base level notation which allows for some cross-platform tooling and interoperability which are qualities of intermediate representations.

Currently there is 1 global keyword map per grammar. If I want to use the same name for 2 different keywords there is no way to do it.

# program:
plot.scatter
 x Day
person.chromosome
 x 231

# grammar:
keyword x

Currently there is no way to define 2 different keywords named "x". Should I allow this?

The better solution might be grammar switching.

As a rustacean pointed out, it doesn't look like there's a formal spec in this repo anymore. While there are fragments of them in the papers and on pages like the 2 below, it may be helpful to have a formal spec for implementors in other languages to use.
#1
https://github.com/breck7/space/blob/master/spec.txt

I don't think we should, but it does potentially make Tree programs play nicer with spreadsheets. We should document why this is a good or bad idea.

It is nice to have the mutable methods in one place, but not sure if there is a better way to do this.

The current pattern is ugly and we should have better typing for things like forEach.

Looks like you need 1500 reputation to add a "treenotation" tag. I might have that in one of my accounts but need to check.

tree-sitter/tree-sitter#379

Add "toWebForm" method and do an RFC explaining the idea.

Perhaps add "header" flag to nodeType definitions in grammar language.

reported from suhail

http://treenotation.org/

Many reports find the current homepage confusing. Let's eventually get a new one going.

Potential models: https://guild.ai/ http://tree-sitter.github.io/tree-sitter/

With color highlighting we don't need that anymore.

The JSON example might be misleading. Let's clean up the simple examples in the readme.

I typed out the following JSON in the sandbox:

{
  "foo bar": "foo"
}

Which converted to the following tree:

foo bar foo

However, that tree converts back to the following JSON:

{
  "foo": "bar foo"
}

This JSON differs from the input.

I don't know if this only an issue with whitespaces but there seems to be some inconsistency or incompatibility when mapping between the two.

• Currently we don't handle scoping correctly so the syntax highlighting for nested trees is often incorrect (but mostly correct since most nodes are on root)
• Should also have the ability to specify other modes, like "javascript" or "c++", for children, to leverage highlighters for other languages

Work for this is partially done at http://lime.treenotation.org/

I am wondering how much use pytree is in comparison to jtree... jtree obviously looks a lot more complete for CLI use.

I want to begin playing with tree notation. Is the pytree experience feasible or should I go for a good starting experience with jtree?

Something that correctly encodes all (most) of JSON.

Add some demo API example languages. Do a "Tree Notation as the NextGen API Language" RFC post.

Demo from 2014:
https://www.youtube.com/watch?v=DV4Nv23bBwQ&t=1s

For tiny DSLs, what if we had grammar up top, data below.

Example:

grammar
 productsNode
  root
  inScope productNode
 anyCell
 folderCell
  enum treeComponentFramework/sweepercraft/ treeComponentFramework
 targetCell
  enum browser nodejs
 productNode
  inScope idNode outputFileNameNode folderNode filesNode targetNode exportClassesNode
 idNode
  cells anyCell
 outputFileNameNode
  cells anyCell
 folderNode
  cells folderCell
 filesNode
  catchAllCellType anyCell
 targetNode
  cells targetCell
 exportClassesNode
  cells anyCell anyCell anyCell
data
 product
  id SweeperCraft
  outputFileName SweeperCraft.browser.js
  folder treeComponentFramework/sweepercraft/
  files treeComponentFramework/sweepercraft/SweeperCraft.ts
  target browser
 product
  target browser
  id TreeComponentFramework.Browser
  outputFileName TreeComponentFramework.browser.js
  folder treeComponentFramework
  files worldWideTypes/treeNotationTypes.ts treeComponentFramework/Willow.ts treeComponentFramework/TreeComponentFramework.ts
 product
  target nodejs
  id TreeComponentFramework.Node
  outputFileName TreeComponentFramework.node.js
  folder treeComponentFramework
  files worldWideTypes/treeNotationTypes.ts treeComponentFramework/Willow.ts treeComponentFramework/TreeComponentFramework.ts
  exportClasses AbstractTreeComponentRootNode AbstractTreeComponent AbstractCommander

Let's build a page like: http://json.org/ which lists all the implementations of libraries in the different languages. It could also help coordinate who is working on what language and what the status of each project is.

Once TreeScript is more stable let's move completely off TypeScript to TreeScript (and run stuff off that for the most part, compiling to JS/WASM later)

Here's a good model to base it on (https://home.cern/science/computing/birth-web/licensing-web)

CERN relinquishes all intellectual property rights to this code, both source and binary and permission is given to anyone to use, duplicate, modify and distribute it.

How should we provide better support for Blank Lines in the Grammar Language?

It's a common request. I feel like we should change the defaults and perhaps make blank lines not errors, and perhaps a default node type, and have Tree Language designers opt in to making them errors.

Please provide a grammar of TN in ABNF.

Redesign of the site should change the TreeNotation.org landing page to a show don't tell, with images, videos, animated gifs, demos, etc.

Should have types like 40oz 10meters 12ft etc.

https://medium.com/@Jernfrost/defining-custom-units-in-julia-and-python-513c34a4c971

https://en.wikipedia.org/wiki/Dimensional_analysis

How can we solve the "meta" problem of how to help others build Tree Notation libraries in different host languages?

Can we provide a universal set of test cases?

from discussion with ahmed

We need to show people the db so they can see the mountain of work and prior languages this is building on.

For SSV language

Can be some tricky dependency cycles with upgrading to newer versions of languages. Should have a design pattern perhaps where we tag a branch for the desired upgrade, or just ship a separate repo with the necessary files from products/ coupled with the upgrade code.

Perhaps use TS as a model: https://typescript-play.js.org/

From user tests:

• TESTS
• TEST COVERAGE
• MIGRATE to TCF
• Readme updated as you type
• Fix the ## bug
• Autoexecute checkbox?
• Autocomplete always visible?
• Autocomplete better
• Infix language!!! If they haven’t done prefix language
• API examples
• Better description on + in numbers
• Have something like “there are X nodeTyprs, that is not one”
• Change NodeType to lineType???
• Start ""
• Blank lines in infer

[] speed. let's get back to instant startup. Lazy loading? Save indexes?
[] better ohayo integration with TreeBase
[] query build on treebase startup page.
[] surface 10 interesting queries? instant start. something more exciting than what we currentlty have.
[] distribution of key variables on treebase startup page?