Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Appmaker -- why, what, how
Note: this is as draft as draft gets. Nothing expressed here is at the level of agreed upon, let alone committed to. Just words.
Also, this language will be telegraphic, not ready for marketing purposes, etc. It's primarily to establish shared objectives and approach within the people who end up working on it.
The world is going mobile; FirefoxOS and the capabilities of the HTML5 mobile app platform more broadly are becoming more competitive. A new generation of folks is going to be exposed to the internet and the web platform through mobile devices. In the mobile world, the equivalent of creating a home page in Web 1.0 could be creating your own web app.
Long term goals
Appmaker is a set of initiatives to see if that's possible -- can we make the ability to create a mobile app drastically more accessible to a brand new set of app creators who are not self-identified as developers, and are primarily motivated by something other than "desire to learn the web". We expect their motivations will range from creative satisfaction of "having made a thing that looks & feels real" to a desire to have an app to send to clients of an small neighborhood offline service business.
The grand goal of Appmaker is for millions of people to be able to create apps that give them utility or joy. It need not be through the use of our software, although we expect that we'll have to build or assemble parts of that whole product experience, partner for parts of it, and work with many, many people to both get input into both what the product should be, how our stuff does or doesn't meet the needs of real people, and how to get the software in front of the right users.
Who is this for?
- Basic to advanced users (create from desktop and mobile)
- Basic users (create from mobile) & FFOS emerging market audience (Brazil)
- Users have a phone first, and no desktop machine
- At their homes, WiFi is not prevalent
- Wifi data limits are small, if they have Wifi
- They want to keep in touch with friends and family
- FFOS is targeting the C-class or middle class in Brazil
- Users are looking for the best deal
- Users buy high tech devices for communication, but it needs to be reasonably priced
- Money is familial
- Generally, we want users who aren't developers, but want to make apps
- Further daily-life
- Distribute to social circle
- They want to make apps they care about
- They want to make apps that have a local context
- They want to make apps for their friends and family
What would users actually do
We're currently exploring two basic onboarding paths, and three "advanced" modes.
Creative path 1 / Creating from mobile: Remixable apps
We have a hypothesis that there are some apps which, if they were made "remixable", would satisfy many use cases. Remixing in this context refers to the ability of end-user to take an existing app (built with that capability built in) and do the "mobile" equivalent of copy & paste: on the device, take the app and make a new app that is customized by the user, and republished at a new endpoint from where it can be distributed anew. (there are issues of credit and licensing to deal with -- we'll build in and assume licensing models that encourage this act of remix; removing the ability to remix can be done downstream).
We are identifying several such template remixable apps, which will help us identify the challenges to making remixable apps, and test whether both the specific apps and the notion of remixable apps have user appeal.
Creative path 2 / Creating from desktop: Author new apps on desktop for use on phone only
Web technology has matured enough that most browsers are now capable of fluid, interactive "wysiwyg" tools. We will make an in-browser tool that lets users layout simple components in a layout simulator on a desktop configuration, with minimal "hooking up" of components and minimal navigational support, so that a novice can experience the full cycle of layout / configure / deploy, and make an app that they created.
We'll do this by asserting many hard constraints which tools aimed at professional developers can't afford to. Phone screens to a first approximation can be thought of having similar form factors, varying mostly in pixel count and small aspect ratio variations. We will simplify the generic problem of making a broadly responsive authoring environment (which would need to deal with very different UI patterns for phones, tablets, desktops, TVs) and lean on a prescriptive proportional layout model, a default widget set and skin, and generally limited capabilities.
Upleveling 1: Component authoring with UI support
If people want to create new lego blocks, we want to provide a level of compositionality which is not quite at the level of raw HTML/JS, as that requires a lot of formal programming knowledge. We will be exploring at least two approaches: a 'wires and boxes' programming model, such as MeeMoo; a 'block language' like Waterbear (a webby version of MIT's Scratch).
Upleveling 2: Code editing
This is all just HTML/JS/CSS, so we will have ways to look at / edit their code in a more traditional web authoring environment like Thimble (or JSfiddle or ...)
We're starting by assembling a set of technologies which we think may play a role in the overall offering, and exploring how they need to be tweaked to better fit together. Both complementing and testing some of the assumptions behind that initial thought will be user testing, prototyping and generally trying out whether the assembly makes sense.
The key components we've identified so far are:
HTML5 + WebAPIs + Open Web App spec. FirefoxOS is showing that we can use web technologies defined somewhat broadly to implement experiences that feel "appy". The WebRT ("Web Runtime") is the set of APIs that these apps need (broadly: HTML5 + device APIs and app manifest and the like, all en route to standardization but not yet ubiquitous).
A place to publish and get apps. The Firefox Marketplace (marketplace.firefox.com) is the canonical such place, but we will probably be playing with other app distribution endpoints especially for user and tech testing scenarios.
A higher-level language for lego-block-like assembly of "parts of apps". We're pretty excited about Web Components and will be figuring out how to build on the shoulders of projects like the Web Components standard, x-tags, Polymer and others to identify an abstraction layer that is "good enough" for our users' needs. This is a hard problem, and we'll be very happy if we find 80% solutions.
A place to publish these lego blocks. We'll probably start with a fork of the MakeAPI that's used by the Webmaker project, with plans to merge that fork back in mainline once it's proven its worth.
A layout tool. Our users should be able to go a website, use an appropriately easy and simple tool to create a set of web assets which when deployed through an 'appstore' does what they want. This layout tool will import components and assets from the MakeAPI at first, although the intent is to have other asset stores.
Firefox+Android: We'll be targeting the web stack that Mozilla is currently building. The intent is for this to be ubiquitous, but one fight at a time. We'll focus on mobile, where we expect FirefoxOS will have the best early API support, but where Android (and the WebRT delivered as part of Firefox for Android) will have the most penetration for a while to come. We will thus be targeting users with Android phones capable of running whatever version of Firefox for Android will have the WebRT capabilities built in.
Requirements will change with user feedback: there are a bunch of assumptions and hypotheses in the plan set above. It is exceedingly likely that many of these are wrong. We will do our best to appropriately challenge these assumptions based on actual user feedback.
We'll be prioritizing stories, which we're tracking as Github Issues in this repo: https://github.com/mozilla-appmaker/appmaker/issues