Skip to content

Latest commit

 

History

History
139 lines (120 loc) · 9.53 KB

NewFeatureProcess.md

File metadata and controls

139 lines (120 loc) · 9.53 KB

New feature process

Where do I submit my idea for a new feature?

The easiest way to submit new feature requests is through Feedback Hub. In Feedback Hub, any Windows user (even if they aren't on GitHub) can upvote suggestions. The Calculator team reviews these suggestions regularly, and when we're ready to work on an idea we create feature pitch issues here on GitHub.

But using Feedback Hub is not required—you can create feature pitches too! This document explains the elements of a good pitch and how we'll guide features from a pitch to a finished product. The Feature Tracking board shows all the features we're working on and where they're at in the process.

Do I need to follow this process? Can I just start coding and submit a PR?

You do not need to follow this process for bug fixes, performance improvements, or changes to the development tools. To contribute these changes, discuss the proposed change in an issue and then submit a pull request.

You do need to follow this process for any change which "users will notice". This applies especially to new features and major visual changes.

Step 1: Feature pitch

The feature pitch concisely describes a point of view on the problem the new feature should solve. It will typically include these sections:

  • Problem Statement: What problem are we trying to solve? Who’s the target audience? Is there a customer need or pain point we need to remedy? Is there a business goal or metric we are trying to improve? Do we have a hypothesis we want to prove or disprove?
  • Evidence or User Insights: Why should we do this? Potential sources of data: Feedback Hub, other GitHub issues, other anecdotes from listening to customers in person or online, request from another team, telemetry data, user research, market or competitive research
  • Proposal: How will the solution/feature help us solve the problem? How will it meet the target audience’s needs? If there are business goals or metrics, how does this improve them?
  • Goals: What you want to accomplish with this feature. Typical examples include “User Can perform some task
  • Non-Goals: Things we are explicitly not doing or supporting or that are out of scope, including reasons why.
  • Low-Fidelity Concept: Show as much of the experience as needed to explain the idea. This can be as simple as a napkin drawing but can also be a code prototype, a PowerPoint walkthrough, or a design comp.

The low-fidelity concept should be kept simple at this stage and refined during the pre-production process.

Feature pitches are submitted as issues on GitHub. We encourage discussion on open issues, and as discussion progresses we will edit the issue description to refine the idea.

Step 2: Pre-production

In the pre-production phase, we experiment with a variety of ways to address the goals described in the feature pitch. The output of this phase is a specification which demonstrates how the feature will work, supported by design renderings and code prototypes as needed. Sometimes we'll learn new things about a feature proposal during pre-production, and we'll edit or close the original pitch.

We welcome community participation in the pre-production process. The GitHub issue will be the primary place to share progress updates.

The best ideas often come from trying many ideas during the pre-production phase. To enable rapid experimentation, we encourage developing and sharing rough ideas—maybe even with pencil and paper—before making designs pixel-perfect or making code robust and maintainable.

Spec review

Once there is a high-fidelity design which addresses the goals described in the original pitch, the Microsoft product team will review the prototype and ensure all items on this checklist are addressed:

  • Is there a high-fidelity design which gives reviewers a clear idea of how the feature will look and function when implemented?
  • Has the plan been shared with the community (documented on the wiki and updates posted in the original issue) and have others been given an opportunity to provide suggestions?
  • Are Fluent design principles followed? If we do something which deviates from the guidelines, do we have a good reason?
  • Does the design include provisions for all users and all cultures?
  • Is it technically feasible to build this feature? Take a look at the "before committing" checklist below and identify any issues which are likely to be blockers.

Step 3: Production

A feature can be implemented by the original proposer, a Microsoft team member, or by other community members. Code contributions and testing help are greatly appreciated. Please let us know in the issue comments if you're actively working on a feature so we can ensure it's assigned to you.

You might be able to reuse code written during the prototype process, although there will typically be more work required to make the solution robust. Once the code is ready, you can begin submitting pull requests.

Technical review

As with all changes, the code for new features will be reviewed by a member of the Microsoft team before being checked in to the master branch.

New features often need a more thorough technical review than bug fixes. When reviewing code for new features, the Microsoft team considers at least these items:

  • All items on the Accessibility checklist should be addressed.
  • All items on the Globalization checklist should be addressed.
  • The change should be tested on the oldest version of Windows that the app supports. You can find this version number in AppxManifest.xml. Any calls to APIs newer than that version should be conditionally enabled.
  • The change should use only supported APIs. If there are any questions about whether legacy or undocumented APIs are in use, the Windows App Certification Kit should be run to check.
  • The change should save the user's progress if the app is suspended and resumed. Code to handle these cases should be tested in the Visual Studio debugger.
  • If the change has customizations for particular device families, it should be tested on those device families.
  • The change should be tested with the app window resized to the smallest possible size.
  • The change should be tested with light, dark, and high contrast themes. It should honor the user's preferred accent color.
  • If the change adds new libraries or other dependencies:
    • If the library is packaged with the app, the increased size of the binaries should be measured.
    • If the library is not maintained by Microsoft, the Microsoft team will need to set up a plan to monitor the upstream library for changes like security fixes.
    • If the library is being used under an open-source license, we must comply with the license and credit third parties appropriately.
  • If the change adds code which runs during the app's startup path, or adds new XAML elements which are loaded at startup:
    • Run the perf tests to measure any increase in startup time. Move work out of the startup path if possible.
  • If the change adds additional logging:
    • All logging should use TraceLogging.
    • Unnecessary log events should be removed, or configured so that they are collected only when needed to debug issues or measure feature usage.
  • If the change reads user data from files or app settings:
    • Verify that state saved in a previous version of the app can be used with the new version.
  • If the change makes network requests:
    • Microsoft must plan to keep these dependencies secure and functional for the lifetime of the app (which might be several years).
    • The app should be fully functional if some network requests are slow or fail. Tools like Fiddler can be used to simulate slow or failed requests.

Step 4: Final product review and merge to master

After the technical review is complete, the product team will review the finished product to make sure the final implementation is ready to release to Windows customers.