Joplin Refactor Wishlist

[Niicck]

Happy holidays again! Here's our wishlist for stuff in Joplin we'd like to refactor/clean-up. Listed in order of importance/possible timeline.

• [x] Break content_types into their own modules
  • make sure that migrations, api, and existing data all works.
  • MVP: move 1 simple content type into its own module and make sure it works.
• [ ] Clean up content_type creation
  • There are many steps required to create a new content_type. Maybe it could be easier.
• [ ] Translations
  • @briaguya had some ideas about improving this workflow/comprehensibility
• [ ] Clean up js code
  • clean up file structure/organization
  • remove stale code that doesn't actually do anything
• [ ] bidirectional many-to-many relationships
  • @easherma has an MVP wip

[easherma]

IMO, we need an agreed upon file/folder/model structure for Joplin. Moving each content type into it's own module is cool, its a useful first step if it's thoughtful. But it's mostly just moving code around at the end of the day, it's not going to deliver any better results for content or users. I'd love to talk about this more in person.

Here's a brain dump of my wishlist, also in rough order of priority.

• have an agreed on file structure and general philosophical approach (this shouldn't take long, right? ;-) )
• write factories for the rest of our content types so we can easily test how scaleable our current architecture is
• test scalibility; note potential problems and develop some solutions
• Decide how to moduarlize our various UI customizations and write them as individual, maintainable Django apps
  • Combine this with the UI stuff @chasechenevert is working on so we can deliver something new and valuable
• Identify all non-standard functionality in joplin and write tests around it
• refactor the create content model
  • revisit its implementation; compare it to other similar wagtail plugins
  • it should have feature parity with wagtails default way of creating pages
  • it should know what available content types there without having to hard code them
• Revisit our graphql api schema, how it works, how it could be improved
  • Use the above test data to test how well it scales and where it falls apart
  • Try a coexisting implementation of wagtail grapple so we can compare more side by side
• Refactor previews: ideally work using Janis React components(we could import them from janis as a submodule). This would solve a lot of dev-ops headaches, code duplication and make content happy.
• take some time to look at other similar wagtail projects and orgs (like Torchbox, CFPB, others). how is our modeling similar or different?
• talk to Marni and also Torchbox about engaging with them for some small budget consulting as a sanity check before we go extra prime time

[Niicck]

My goal is to narrow down what things need to be decided by committee in a meeting and what things can just be made into issues and given to someone to do individual discovery. The more things that can just be triaged into individual issues, the better. The more that can be done asynchronously, the better. An hour goes by fast.

Moving each content type into it's own module is cool, its a useful first step if it's thoughtful. But it's mostly just moving code around at the end of the day, it's not going to deliver any better results for content or users.

• Tell us more. What specifically could we do to make that content_type refactor better/more useful?

write factories for the rest of our content types so we can easily test how scaleable our current architecture is + test scalibility; note potential problems and develop some solutions

• Load testing would be a good issue.

Decide how to moduarlize our various UI customizations and write them as individual, maintainable Django apps

• Nice to have, I don't think its a top priority/immediate blocker.

Identify all non-standard functionality in joplin and write tests around it

• 100%

refactor the create content model

• Nice to have, I don't think its a top priority/immediate blocker.

Refactor previews: ideally work using Janis React components(we could import them from janis as a submodule). This would solve a lot of dev-ops headaches, code duplication and make content happy.

• What's wrong with Previews currently?

take some time to look at other similar wagtail projects and orgs

• When completing issues, it's definitely a good idea to see how other teams have solved the same problem before.

talk to Marni and also Torchbox about engaging with them for some small budget consulting

• What would you have them consult on?

[easherma]

Tell us more. What specifically could we do to make that content_type refactor better/more useful?

It could be a whole range of things to try. Some initial notes are here: https://github.com/cityofaustin/techstack/issues/2513 and here: https://github.com/cityofaustin/techstack/issues/2700

In general I don't think this is a bad move, but if we don't actually change any code we shouldn't expect (and therefore promise) some sort of improvements beyond a mild dev quality of life unless we really take things to the next level.

Load testing would be a good issue.

I think factories + load testing is probably number 1 for me, especially after looking at our 2020 roadmap. I am skeptical about how scalable and testable our entire stack is.

What's wrong with Previews currently?

Mainly:

• its one more thing to have to manage when adding new content types
• its a decent amount of sources of bugs/tweaks/things that we have to go back and add later
• there is lots of confusion among content (and ourselves sometimes) about what previews are
• as is, there are extra revisions being saved everytime you preview or share a draft
• using janis components inside joplin for previews would be a great first step towards upping our frontend UI game in joplin overall

What would you have them consult on?

Mainly the very 1st part about content type organization, as well as our polyhierarchical needs and best/better strategies for management in the admin. There are also some things we often say 'Joplin/Wagtail' can't do that I think are more matters of 'we haven't seen an example of how that works' than 'it can't work that way'.

[chiaberry]

@Niicck can we cross some things off the wishlist soon? 🗒️