injectable looks good enough.

Why

Why not?

The Plan

Why

We need to follow best practices.

Related: #126.

The Plan

Where we want to deploy our code:

For Json Serialization, we should make a new class:

abstract interface class Model {
  toJson();
  fromJson();
  ...
}

And make all of the models inherit from them. Then, move the models to their own files. Will required adding a dependency on json_serializable.

Blocked by: #129

If we want to break stuff into multiple packages, it looks like Melos is the standard.

Originally posted by @lishaduck in #13 (reply in thread)

Having messed with our setup, I think using Melos would be a good idea, @ParkerH27.
Here's what I'm thinking:

.
├── ios/
│   └── [...]
├── lib/
│   ├── main.dart
│   ├─── main_[*].dart
├── linux/
│   └── [...]
├── macos/
│   └── [...]
├── packages/
│   ├── app/
│   │   ├── pages/
│   │   │   ├── favorites
│   │   │   │   ├── favorites.dart
│   │   │   │   ├── favorites_logic.dart
│   │   │   │   └── favorites_page.dart
│   │   │   └── [...]
│   │   ├── tests/
│   │   │   └── [...]
│   │   ├── app.dart
│   │   ├─── app_view.dart
│   │   ├─── app_router.dart
│   │   ├── bootstrap.dart
│   │   └── wrapper.dart
│   ├── server/
│   │   ├── src/server/
│   │   │   ├── __init__.py
│   │   │   ├── main.py
│   │   │   └── [...]
│   │   ├── tests/
│   │   │   └── [...]
│   │   ├── pyproject.toml
│   │   └─── poetry.lock
│   ├── shared/
│   │   ├── l10n/
│   │   │   ├── arb/
│   │   │   │   └── [...]
│   │   │   └── l10n.dart
│   │   ├── test/
│   │   │   ├── helpers.dart
│   │   │   └── pump_app.dart
│   │   ├── tests/
│   │   │   └── [...]
│   │   └── model.dart
│   └── widgets/
│       ├── big_card
│       │   ├── big_card.dart
│       │   ├─── big_card_logic.dart
│       │   └─── big_card_page.dart
│       ├── [...]
│       └── tests/
│           └── [...]
├── web/
│   └── [...]
├── windows/
│   └── [...]
├── melos.yaml
└── pubspec.yaml

The only question I would have would be how codegen would work.

Discussed in #40

The Plan

See the logic in #82.

Appwrite makes it unnecessary.
Supersedes: #81, #76.

It's way more testable. Also, fpdart.

Repls are not private unless you pay, Also from what I can tell replit does not run flutter so we may have to move to another dev environment such as a vs code live share server

Originally posted by @ParkerH27 in https://github.com/PSDTools/.github/discussions/6#discussioncomment-4797348

They're a whole lot more hassle than they're worth.

Supersedes: #103

Why you need to check before pushing breaking refactors... 21c46da broke the web build.
See #69 for change context.

So...Netlify works for now, but when we start querying the Python server, we'll need something different.
Options:

• webapp.io — I've used this tool before. I didn't like it, but it does what we need. It's a full-stack review environment and deployment service that uses a concurrent version of docker called Firecracker, which is what powers AWS Lambda.
• Azure — Comes with the student developer plan, but it's quite limited in both free and out-of-the-box. I don't want to, say, have to learn Kubernetes just to do what I can do with the click of a button on Netlify. However, It has a tutorial on serverless FastAPI functions and a more advanced tutorial for non-FastAPI serverless functions, so it might work. However, it comes with 1 million free requests a month,¹ so maybe pricing ain't an issue.
• (GCP (Google Cloud Platform))[https://cloud.google.com/run/docs/quickstarts/build-and-deploy/deploy-python-service]
• local hosting (I (@PetalCat) have a machine for VMs and a network) I think this is a last-ditch option and can only be used for devin'.

Would using Codecov be more effective than our current testing setup?
Also, while I'm thinking about Sentry, we should see if Pattonville has a preferred logging framework because while Sentry supports Flutter, it doesn't let multiple users use it on the free plan.

Originally posted by @lishaduck in #28 (comment)

Having thought about it, I think using a logging framework is a good idea, so we'll need to ask if the district has a Sentry plan. We could also set up Codecov.
Having looked into it, both Sentry and Codecov have a (free!) student plan through GitHub though, so maybe we wouldn't need to get seats on the district's plan. Which we both know they probably wouldn't let us do. 😁

"It's...essentially the course guidebook in an app, and kids can click on the classes to get and description and see what prerequisites it needs and what area(s) it counts for (like Pract. Art, Fine Art). They can also keep track of what they have already taken and passed and what they still need."

This fills a need that the general student body has. Homeroom was chaos the other day because of the lack of this, and the Map Your Journey website is not particularly useful.

The views we'll need to add:

If #126 doesn't stop this, I'd like to use Widgetbook.

Blockchains

Chat-GPT recommended:

• BigChainDB
• Quorum Blockchain

ChatGPT

classes are annoying me so much! If @PetalCat doesn't mind, I think I'd rather codegen them from functions via functional_widget. While we're at it, I might try out hooks_riverpod.

Currently they're ignored via the security tab, but Swift does need to be added to code-scanning, and it would be nice to use actual DevSkim ignores for DS137138

• Flutter (official docs)
• Very Good Core (docs on flavors)
• Flutter Ready to Go! (article by Julio Bitencourt)

We should also take advantage of flutter_launcher_icons and flutter_native_splash with their flavor support.

While I'm thinking about it, there was also a good series of articles on Medium that I stumbled upon which you might want to look at, @ParkerH27:

• Architect your Flutter project using BLOC pattern
• Architect your Flutter project using BLOC pattern (Part 2)
• Compile time Dependency Injection in Flutter
• Integration and Unit testing in Flutter
  They shouldn't be blocked, but even if the first two are, the second two shouldn't be.
  Also, for the third article, while it's a good read, inject.dart is deprecated, so if we wanted to architect that way, we'd need to use injectable instead.

Premise

Both the PLTW Blog and the STL American report that in 2016-7 the capstone class made an app. What happened? How can we

• find it,
• reuse it, and
• make sure we don't suffer the same fate.

After all, it was another in-house PowerSchool wrapper.
For some reason, it's no longer on the App store or the Play store, despite both articles claiming that was the crowning triumph: that it was on both platforms.
Of course, we'll also be on the web.

If you know anything, @ParkerH27, please tell.

Notes (for posterity):

• There was an app, but it's no longer on the App Store or Play Store.
  • However, it was open-sourced at @Pattonville-App-Development-Team.
  • The Pete's Closet thing on the announcements has a website by @Pattonville-App-Development-Team (Pattonville-App-Development-Team/worldwidehackathon).
• Last year, the capstone class made a project codenamed "Ticketmaster." The iOS app was open-sourced at Conn-Lang/ticketmaster_iOS
  • Unlike the app by @Pattonville-App-Development-Team, it's still running on school servers, though it's unused.
  • The source for the web app was bundled by webpack 😖, which minifies the code so we can't read it.
    • It's also not open-source.
    • It's unused because it didn't have official backing.

Conclusions are in #14.

Per rrousselGit/riverpod#2715.

This's just an issue for an issue.
Honestly, Idk why this is/how this became broken.

We need to track our total code time for the ELO to prove our work.

This looked like a good solution: Code Time.

Embed psdtools/GPA_Calculator into the app.

Here's the full todo list for the Wallet:

Still think there's got to be a better way to do this. Maybe an enum w/ a case statement?

Originally posted by @lishaduck in #104 (comment)

For students.

Discussed in #139

All of these are nice-to-haves, but AboutListTile is required legally (#15).

• Drawer
• DrawerHeader
• AboutListTile
• build_version

^{Originally posted by @lishaduck July 26, 2023}

Let's finish the C4 diagram in the README before the end of the school year.

Might make life easier, and would let use enable prefer-match-file-name in more places.

• Seeing 5af1a59...83d5ec8 found so many errors, we could use a more complete configuration.
• So... we should use a package so we don't spend too much time messing with it, or else I will.
• So, we should use very_good_analysis because I'm good at writing unidiomatic Dart code.
• This is blocked on VeryGoodOpenSource/very_good_analysis/pull/73 because we use Dart 3, and they don't support it quite yet.
• That PR is blocked on Dart 3.1.0, which is blocked on Flutter 3.11.
• So we might be waiting a while.
• Meanwhile, we could use dart_code_metrics, but it also doesn't support Dart 3 or analyzer 5.12.0 (dart-code-checker/dart-code-metrics#1231), which custom_lint¹ uses.

Why

Users need to see the latest and greatest data. Let's use Appwrite's Realtime API to subscribe to changes.

The Plan

What we concluded from #1 and #16.

While we won't need to stop using Netlify now that we're using Appwrite, it is slow.
Let's speed up deployments by reusing the already created web artifacts and deploying via actions.

Per rrousselGit/riverpod#2631, it seems like that should drastically speed up CI.

Vite is framework agnostic, and its only want is a single HTML file. Since Flutter makes SPAs, not MPAs, we can bundle it through Vite if we want. This is important if we want offline because we'll have to use vite-plugin-pwa until flutter/flutter#60069 is resolved.

Automate getting the GPA for the calculator via the Canvas and PowerSchool GraphQL and RESTful APIs.

Until #12 is resolved, and maybe even after, we should use flutter_web_preloader to preload the app.

It's tech debt. See eradicate for rational.

We should also take advantage of flutter_launcher_icons and flutter_native_splash with their flavor support.

Originally posted by @lishaduck in #35

Supersedes #35

Why

I needed to write it down before I forgot.

The Plan

Riverpod isn't to be used as a full-blown DI solution for now, per Remi.
(I think) This is because it blocks tree-shaking.

Supersedes: #112

Useful links:

• package:auto_route on pub.dev

Code:

@AutoRouterConfig(replaceInRouteName: "Page,Route")
class NavigationManager extends _$NavigationManager {
  @override
  RouteType get defaultRouteType => const RouteType.material();

  @override
  List<AutoRoute> get routes => [
        AutoRoute(
          path: "/",
          page: MyHomeRoute.page,
          children: [
            AutoRoute(page: GeneratorRoute.page, path: ""),
            AutoRoute(page: FavoritesRoute.page, path: "favorites"),
          ],
        ),
        RedirectRoute(path: "*", redirectTo: "/"),
      ];
}

ChatGPT description:

The code you have provided appears to be written in Dart language and is defining the routing configuration for an application using the auto_route package.

The get routes method is overriding the base class method to return a list of AutoRoute objects. Each AutoRoute object represents a path within the application that maps to a specific page or view.

In this example, there are two routes defined:

• The root route / points to the MyHomeRoute page, and has two child routes defined: GeneratorRoute and FavoritesRoute.
• The RedirectRoute is a catch-all route that redirects any invalid routes to the root route.

Here's a brief explanation of each property used in the AutoRoute and RedirectRoute objects:

• path: A string that defines the path for the route.
• page: The page or view that will be displayed for the route.
• children: An optional list of child routes that can be nested under the parent route.
• redirectTo: The path to redirect to when a route does not match any of the defined routes.

Overall, this code is setting up the routing configuration for the application, allowing users to navigate between different views based on the path they visit.

Error:

Assertion failed: file:///Users/user/.pub-cache/hosted/pub.dev/auto_route7.1.0/lib/src/route/auto_route_config.dart:419:12this[key]?.children != null
"MyHomeRoute does not have children"
See also: https://flutter.dev/docs/testing/errors

ChatGPT:

This error message indicates that there is an assertion failure in the auto_route package related to the MyHomeRoute page.

The error message suggests that the MyHomeRoute page does not have any child routes defined, but the AutoRoute configuration is trying to add child routes to it. This means that there may be a mismatch between the AutoRoute configuration and the actual structure of the pages and routes in the application.

To resolve this error, you should review the AutoRoute configuration and ensure that the MyHomeRoute page does not have any child routes defined. If MyHomeRoute is intended to have child routes, then you should update the AutoRoute configuration accordingly.

You may also want to check the documentation for the auto_route package to ensure that you are using it correctly, and review any relevant code samples or examples to get a better understanding of how to configure routes and pages within your application.

We should split the CI pipeline into two workflows for more granular control.

...Again. It works fine locally. #93 might help make it more reproducible.
If we don't get it fixed, we won't trust it anymore.