• previous entry: https://rafikhan.io/pcj-friday-may-31-2024
• Briefly, making it open source is a risk but probably the right decision
  • we need more people working on this problem
  • we need to be able to share ideas
  • we need to be able to look on failed/abandoned/completed projects and learn from them
• mit vs gpl is something that still has to be figured out.
• copyright assignment is probably needed in order to make changes to the license if necessary, to protect the vision
  • laws, licenses change
  • sometimes we need to make changes to adapt to the environment
  • we should not abuse this power
• to briefly comment on an item from the earlier entry: Why is it okay to ignore the reality of not having funding to do this work?
  • we can ignore the funding problem because the work needs to be done
  • we cannot stop progress because we haven’t managed to convince the people with money that we are working on the right problems
  • hopefully, we can find an answer to the funding problem soon
  • this is a gamble, but I believe it has better odds than figuring out a business model before running out of vc funding, or waiting to start the project until we have funding
• next, I want to choose a problem to work on
  • it has to be small enough so we can finish it
  • it has to be different enough that we can learn some new things
  • it has to be a real problem that I have, so that I am interested in the problem as well as the solution
  • Let’s build an alarm clock app that syncs across devices

Alarm clock app

• sync across my laptops, phones, tablets
• I use alarms to set reminders for meetings and events
• Calendar apps give you notifications, but I wanted something that prompts the user and requires an action to dismiss

Things I want to learn from this project

• how to deploy an app on squeakjs using cuis
  • I am choosing to build on top of cuis because we have shared values around keeping complexity low, and viewing the computer as a means to augment the human intellect
• how well an app deployed on cuis, squeakjs, browser performs on different devices
• how to build a UI using cuis
• hands on experience on the software development process and developer experience in building cuis apps, using git for vcs
• if the cuis graphics system works on squeakjs, and what its limitations may be
  • this is important to help determine how usable the cuis graphics system is compared to HTML, dom, CSS

Getting started on the work

• I created a github repo for the main project https://github.com/z3t0/synced-alarms
• needed to update the SSL on my internal hosted git forge
  • I don’t want to be dependent on github
  • I like hosting my own services, it gives me more opportunities to learn about parts of the stack that I would normally avoid

• as the first item, I need to get a development environment with squeakjs and cuis running in the browser
• squeakjs home page: https://squeak.js.org/
• a link for using squeakjs with cuis 6.2: https://squeak.js.org/run/#highdpi&url=https://cdn.jsdelivr.net/gh/Cuis-Smalltalk/Cuis6-2@6f984e8c/CuisImage&files=[32BitImages/Cuis6.2-32.image,32BitImages/Cuis6.2-32.changes,Cuis6.2.sources]
• there’s quite a bit of lag when interacting with cuis
  • I made a recording https://youtu.be/r1tNnRWvpC0
  • there are a number of issues
  • normal interactions are slow, it takes time for a menu to open
  • initial load time is slow
  • everyone once in a while you get this spinner in the middle of the screen, maybe its garbage collecting?

• to me, this is too slow to be usable
• for reference, take a look at this video which shows the performance when running cuis directly on linux
  • https://youtu.be/xOuZpDjDdmo

• okay, so now we can see what the current performance is, and at the same time we can get an idea of what performance is possible
• now I know that Craig Latta has also done work using a similar stack – https://caffeine.js.org/
  • I think he got quite a bit further with this approach, which suggests that he’s had better performance
• Now, of course this work can spiral out into many different areas of experimentation,
  • I want to put a hard cap on it because the objective isn’t so much to build the app, but to find some new answers and new questions
  • the objective is to learn so that in six weeks I am better positioned to solve this problem
  • so to that end lets give this a deadline
  • the deadline is 20th july 2024 – roughly six weeks
• okay so back to evaluating squeakjs
  • squeakjs is also used for an early version of croquet – https://github.com/codefrau/jasmine
  • this one seems to run a lot faster
  • dragging around a window doesn’t take so long
  • https://youtu.be/PYtAxaCGNlo
  • it does slow down pretty quickly
  • when dragging around the little widget/sprite – it did not feel smooth
  • a number of things have been built using squeakjs, how did they manage with a laggy UI?
  • from the squeakjs home page:
    • smalltalk zoo https://smalltalkzoo.thechm.org/
      • the UI is performing really well here
      • seems to be using lively, not squeakjs
      • it is using morphic dom? I’m not sure how the UI is built
      • I think it may be using both squeakjs and lively
      • going to take a pause here and spend some time learning
        • read the squeakjs paper
        • also joined the squeak slack so I can ask more questions there
    • caffeine js – https://caffeine.js.org/
• to be continued
  • caffeinejs
  • squeaknews ezine
  • croquet
  • plopp
  • code paradise
  • sugarizer
  • scratch archive
  • petit petri

• there is also some recent activity on the squeaks js github
  • https://github.com/codefrau/SqueakJS/issues/121 – high performance jit
• the minimal demo – https://squeak.js.org/demo/simple.html performs really well