Jonathan Edwards 07/10/18

• I was able to “solve” the problem in Reflex.
  • http://futureofcoding.org/log#dom-recursion-problem-solved-by-reflex
  • Hard, types and shit, emacs… worked on emacs setup stuff today
• They also have ToDoMVC
  • I mostly understand it
  • And the code looks like what I was looking for
• I’m going towards a language (elm) or framework (cycle) or devtool (rxfiddle)
  • My vague sense is that I need to understand Reflex and then build something with its abstractions
• I also have this sense that stream combinator visuatializations, like rxfiddle, are key to making FRP understandable
  • One way to say this: declarative language + GUI on top (HTML, SQL)
    • Note: PX (shawn mcdirmid), model and UI
  • Map is straight forward arrows ->
  • Should I go down the list and try to do this for them all?
• Don’t want to fight a battle on two fronts, though

• Also want to think about presenting something at SPLASH

• https://vlhcc18.github.io/
  • User studies, very academic
• Go to SPLASH, LIVE (pre) and ONWARD (during)
  • Less and less practitioners, more researchers
• Strangeloop is more industry. Cool, neat ideas. More industry. Less research
  • ICFP is co-located there.
    • Racket people
  • Stopped going
• One way to get oriented is to write up a survey with all the different FRP implementations
  • Reflex
  • Elm
  • Maybe show another one if it’s helpful
  • Even just one example is enough: just show a problem
    • Motivating Example
  • Maybe ToDoMVC
• Essay
  • Elm
    • Motivating example
  • Reflex
    • Solved
  • Hard to use
    • How to improve with viz?
  • Link on social media, put on website
• After
  • For specific example
    • What would it look like - mock it up
    • Storyboard
    • Don’t worry about general case
  • Publish this
  • Bret Victor demo is just build UI for that one example and make video
    • But what you really want is why it’s hard to generalize
• And then
  • Generalize it…
  • Build a thing…
• Apply to REBLs
  • Aug 17 to apply
  • Concatenate Essay and After
  • Progression:
    • Problem
    • Good semantics
    • Difficult to edit
    • How to fix UI?
    • Here are the technical challenges to building this for real
• If I don’t make it
  • PX workshop at <Programming>
  • https://2018.programming-conference.org/
• Co-design - two fronts
• So hard to go down a bad path
  • But if you get far enough to show results, you get a seat at the table, people know your name

LIVE and text is hard because sometime the text doesn’t make sense

That’s why structure editing is great for LIVE

His work

• Great quote from urbit
  • “Current systems are built on principle on half of your brain can be erased.” - at any point you can start from database state and reload into memory.
• Only really structured editor is Scratch
  • Simple and laborious
  • Doesn’t do any refactoring
• Tried a bunch of approaches
  • Photoshop one of windows - control panel approach
  • Meta model approach in Chorus
• Greenfoot also seems to work, they teach people with it
• Blockly from EToys and Blockly
  • I want to do things that aren’t just changing the AST
    • Such as change the type of something
• I need a fancier data model than an AST
  • I have a data model
  • I have actual data

Hard to do more than 4-5 hours per day of peak creative work and fill in with other things