PowerShellForGitHub/CONTRIBUTING.md

32 KiB

PowerShellForGitHub PowerShell Module

Contributing

Looking to help out? You've come to the right place. We'd love your help in making this the best way to automate GitHub repos.

Looking for information on how to use this module? Head on over to README.md.


Table of Contents


Overview

We're excited that you're excited about this project, and would welcome your contributions to help it grow. There are many different ways that you can contribute:

  1. Submit a bug report.
  2. Verify existing fixes for bugs.
  3. Submit your own fixes for a bug. Before submitting, please make sure you have:
  1. Submit a feature request.
  2. Help answer questions.
  3. Write new test cases.
  4. Tell others about the project.
  5. Tell the developers how much you appreciate the product!

You might also read these two blog posts about contributing code:

Before submitting a feature or substantial code contribution, please discuss it with the PowerShellForGitHub team via Issues, and ensure it follows the product roadmap. Note that all code submissions will be rigorously reviewed by the PowerShellForGitHub Team. Only those that meet a high bar for both quality and roadmap fit will be merged into the source.

Maintainers

PowerShellForGitHub is maintained by:

As this module is a production dependency for Microsoft, we have a couple workflow restrictions:

  • Anyone with commit rights can merge Pull Requests provided that there is a 👍 from one of the members above.
  • Releases are performed by a member above so that we can ensure Microsoft internal processes remain up to date with the latest and that there are no regressions.

Feedback

All issue types are tracked on the project's Issues page.

In all cases, make sure to search the list of issues before opening a new one. Duplicate issues will be closed.

Bugs

For a great primer on how to submit a great bug report, we recommend that you read: Painless Bug Tracking.

To report a bug, please include as much information as possible, namely:

  • The version of the module (located in PowerShellForGitHub.psd1)
  • Your OS version
  • Your version of PowerShell ($PSVersionTable.PSVersion)
  • As much information as possible to reproduce the problem.
  • If possible, logs from your execution of the task that exhibit the erroneous behavior
  • The behavior you expect to see

Please also mark your issue with the 'bug' label.

Suggestions

We welcome your suggestions for enhancements to the extension. To ensure that we can integrate your suggestions effectively, try to be as detailed as possible and include:

  • What you want to achieve / what is the problem that you want to address.
  • What is your approach for solving the problem.
  • If applicable, a user scenario of the feature / enhancement in action.

Please also mark your issue with the 'suggestion' label.

Questions

If you've read through all of the documentation, checked the Wiki, and the PowerShell help for the command you're using still isn't enough, then please open an issue with the question label and include:

  • What you want to achieve / what is the problem that you want to address.
  • What have you tried so far.

Static Analysis

This project leverages the PSScriptAnalyzer PowerShell module for static analysis.

It is expected that this module shall remain "clean" from the perspective of that module.

If you have never installed PSScriptAnalyzer, do this from an Administrator PowerShell console window:

Install-Module -Name PSScriptAnalyzer

In the future, before running it, make sure it's up-to-date (run this from an Administrator PowerShell console window):

Update-Module -Name PSScriptAnalyzer

Once it's installed (or updated), from the root of your enlistment simply call

Invoke-ScriptAnalyzer -Settings ./PSScriptAnalyzerSettings.psd1 -Path ./ -Recurse

That should return with no output. If you see any output when calling that command, either fix the issues that it calls out, or add a [Diagnostics.CodeAnalysis.SuppressMessageAttribute()] with a justification explaining why it's ok to suppress that rule within that part of the script. Refer to the PSScriptAnalyzer documentation for more information on how to use that attribute, or look at other existing examples within this module.


Module Manifest

This is a manifested PowerShell module, and the manifest can be found here:

PowerShellForGitHub.psd1

If you add any new modules/files to this module, be sure to update the manifest as well. New modules should be added to NestedModules, and any new functions or aliases that should be exported need to be added to the corresponding FunctionsToExport or AliasesToExport section. Please keep all entries to those sections in alphabetical order.


Logging

Instead of using the built-in Write-* methods (Write-Host, Write-Warning, etc...), please use

Write-Log

which is implemented in Helpers.ps1. It will take care of formatting your content in a consistent manner, as well ensure that the content is logged to a file (if configured to do so by the user).


PowerShell Version

This module must be able to run on PowerShell version 4. It is permitted to add functionality that requires a higher version of PowerShell, but only if there is a fallback implementation that accomplishes the same thing in a PowerShell version 4 compatible way, and the path choice is controlled by a PowerShell version check.

For an example of this, see Write-Log in Helpers.ps1 which uses Write-Information for Informational messages on v5+ and falls back to Write-Host for earlier versions:

if ($PSVersionTable.PSVersion.Major -ge 5)
{
    Write-Information $ConsoleMessage -InformationAction Continue
}
else
{
    Write-Host $ConsoleMessage
}

Coding Guidelines

As a general rule, our coding convention is to follow the style of the surrounding code. Avoid reformatting any code when submitting a PR as it obscures the functional changes of your change.

A basic rule of formatting is to use "Visual Studio defaults". Here are some general guidelines

  • No tabs, indent 4 spaces.
  • Braces usually go on their own line, with the exception of single line statements that are properly indented.
  • Use camelCase for instance fields, PascalCase for function and parameter names
  • Avoid the creation of script scoped variables unless absolutely necessary. If referencing one, be sure to explicitly reference it by scope.
  • Don't use globals. If you want to add module configuration, add a new property instead.
  • Avoid more than one blank empty line.
  • Always use a blank line following a closing bracket } unless the next line itself is a closing bracket.
  • Add full Comment Based Help for all methods added, whether internal-only or external. The act of writing this documentation may help you better design your function.
  • File encoding should be ASCII (preferred) or UTF8 (with BOM) if absolutely necessary.
  • We try to adhere to the PoshCode Best Practices and DSCResources Style Guidelines and think that you should too.
  • We try to limit lines to 100 characters to limit the amount of horizontal scrolling needed when reviewing/maintaining code. There are of course exceptions, but this is generally an enforced preference. The Visual Studio Productivity Power Tools extension has a "Column Guides" feature that makes it easy to add a Guideline in column 100 to make it really obvious when coding. If you use VS Code, this module's .vscode/settings.json configures that for you automatically.

Adding New Configuration Properties

If you want to add a new configuration value to the module, you must modify the following:

  • In Import-GitHubConfiguration, update $config to declare the new property along with its default value, being sure that PowerShell will understand what its type is. Properties should be alphabetical.
  • Update Get-GitHubConfiguration and add the new property name to the ValidateSet list so that tab-completion and documentation gets auto-updated. You shouldn't have to add anything to the body of the method. Property names should be alphabetical.
  • Add a new explicit parameter to Set-GitHubConfiguration to receive the property, along with updating the CBH (Comment Based Help) by adding a new .PARAMETER entry. You shouldn't have to add anything to the body of the method. Parameters should be alphabetical save for the SessionOnly switch, which should be last.

Code comments

It's strongly encouraged to add comments when you are making changes to the code and tests, especially when the changes are not trivial or may raise confusion. Make sure the added comments are accurate and easy to understand. Good code comments should improve readability of the code, and make it much more maintainable.

That being said, some of the best code you can write is self-commenting. By refactoring your code into small, well-named functions that concisely describe their purpose, it's possible to write code that reads clearly while requiring minimal comments to understand what it's doing.


Debugging Tips

You may find it useful to configure the module to log the body of all REST requests during development of a new feature, to make it easier to see exactly what is being sent to GitHub.

Set-GitHubConfiguration -LogRequestBody

Pipeline Support

This module has comprehensive support for the PowerShell pipeline. It is imperative that all new functionality added to the module embraces this design.

  • Most functions are declared as a filter. This is the equivalent of a function where the body of the function is the process block, and the begin/end blocks are empty.

  • In limited cases where one of the inputs is an array of something, and you specifically want that to be processed as a single command (like adding a bunch of labels to a single issue at once), you can implement it as a function where you use begin/process to gather all of the values into a single internal array, and then do the actual command execution in the end block. A good example of that which you can follow can be seen with Add-GitHubIssueLabel.

  • Any function that requires the repo's Uri to be provided should be additionally aliased with [Alias('RepositoryUrl')] and its [Parameter()] definition should include ValueFromPipelineByPropertyName.

  • Do not use any generic term like Name in your parameters. That will end up causing unintended pipeline issues down the line. For instance, if it's a label, call it Label, even though Name would make sense, other objects in the pipeline (like a GitHub.Respository object) also have a name property that would conflict.

  • You should plan on adding additional properties to all objects being returned from an API call. Any object that is specific to a repository should have a RepositoryUrl NoteProperty added to it, enabling it to be piped-in to any other command that requires knowing which repository you're talking about. Additionally, any other property that might be necessary to uniquely identify that object in a different command should get added properties. For example, with Issues, we add both an IssueNumber property and an IssueId property to it, as the Issue commands need to interact with the IssueNumber while the Event commands interact with the IssueId. We prefer to only add additional properties that are believed to be needed as input to other commands (as opposed to creating alias properties for all of the object's properties).

  • For every major file, you will find an Add-GitHub*AdditionalProperties filter method at the end. If you're writing a new file, you'll need to create this yourself (and model it after an existing one). The goal of this is that you can simply pipe the output of your Invoke-GHRestMethod directly into this method to update the result with the additional properties, and then return that modified version to the user. The benefit of this approach is that you can then apply that filter on child objects within the primary object. For instance, a GitHub.Issue has multiple GitHub.User objects, GitHub.Label objects, a GitHub.Milestone object and more. Within Add-GitHubIssueAdditionalProperties, it just needs to know to call the appropriate Add-GitHub*AdditionalProperties method on the qualifying child properties, without needing to know anything more about them.

  • That method will also "type" information to each object. This is forward-looking work to ease support for providing formatting of various object types in the future. That type should be defined at the top of the current file at the script level (see other files for an example), and you should be sure to both specify it in the .OUTPUTS section of the Comment Based Help (CBH) for the command, as well as with [OutputType({$script:GitHubUserTypeName})] (for example).

  • Going along with the .OUTPUTS is the .INPUTS section. Please maintain this section as well. If you add any new type that will gain a RepositoryUrl property, then you'll need to update virtually all of the .INPUTS entries across all of the files where the function has a Uri parameter. Please keep these type names alphabetical.

  • To enable debugging issues involving pipeline support, there is an additional configuration property that you might use: Set-GitHubConfiguration -DisablePipelineSupport. That will prevent the module from adding any additional properties to the objects.


Formatters

Our goal is to have automattic formatting for all GitHub.* types that this project defines.

Formatting was first introduced to the project with #205, and succcesive PR's which introduce new types have added their additional formatters as well. Eventually we will get Formatters for all previously introduced types as well.

Formatter files can be found in /Formatters.

When adding a new formatter file, keep the following in mind:

  • One formatter file per PowerShell module file, and name them similarly (e.g. GitHubRepositories.ps1 gets a Formatters\GitHubRepositories.Format.ps1xml file)
  • Be sure to add the formatter file to the manifest (common mistake to forget this).
  • Don't display all the type's properties ...just choose the most relevant pieces of information; sometimes this might mean using a script block to grab an inner-property or to perform a calculation.

Testing

Build status Azure DevOps tests Azure DevOps coverage

Installing Pester

This module supports testing using the Pester UT framework.

To install it:

Install-Module -Name Pester -MinimumVersion 5.3.3 -AllowClobber -SkipPublisherCheck -Force

Configuring Your Environment

The tests intentionally do not mock out interaction with the real GitHub API, as we want to know when our interaction with the API has been broken. That means that to execute the tests, you will need Administrator privilege for an account. For our purposes, we have a "test" account that our team uses for having the tests run automated. For you to run the tests locally, you must make a couple changes:

  1. Choose if you'll be executing the tests on your own personal account or your own test account (the tests should be non-destructive, but ... hey ... we are developing code here, mistakes happen.)
  2. Update your local copy of tests/config/Settings.ps1 to note the OwnerName and OrganizationName that the tests will be running under.

    While you can certainly check-in this file to your own fork, please DO NOT include your changes as part of any pull request that you may make. The .gitignore file tries to help prevent that.

  3. Run Set-GitHubAuthentication to ensure that it is configured with an administrator-level Access Token for the specified owner/organization.

    Unfortunately, you cannot use -SessionOnly with Set-GitHubAuthentication when testing, as Pester works by making new sessions for every test. That means that it must be "globally" configured with that access token for the duration of the Pester test execution.

Running the Tests

Tests can be run either from the project root directory or from the Tests subfolder. Navigate to the correct folder and simply run:

Invoke-Pester

Make sure you have previously configured your Access Token via Set-GitHubAuthentication. Please keep in mind some tests may fail on your machine, as they test private items (e.g. secret teams) which your key won't have access to.

Pester can also be used to test code-coverage, like so:

$pesterConfig = New-PesterConfiguration
$pesterConfig.CodeCoverage.Path = @("$root\GitHubLabels.ps1")
$pesterConfig.CodeCoverage.Enabled = $true

Invoke-Pester -Configuration $pesterConfig
# Be sure you're not passing this in to -PesterOption, since that's different than the configuration.

This command tells Pester to check the GitHubLabels.ps1 file for code-coverage.

The code-coverage object can be captured and interacted with, like so:

$pesterConfig = New-PesterConfiguration
$pesterConfig.CodeCoverage.Path = @("$root\GitHubLabels.ps1")
$pesterConfig.CodeCoverage.Enabled = $true
$pesterConfig.Run.PassThru = $true

$cc = (Invoke-Pester -Configuration $pesterConfig).CodeCoverage

There are many more nuances to code-coverage, see its documentation for more details.

Automated Tests

Build status Azure DevOps tests Azure DevOps coverage

These test are configured to automatically execute upon any update to the master branch of microsoft/PowerShellForGitHub.

The Azure DevOps pipeline has been configured to execute the tests against a test GitHub account (for the user PowerShellForGitHubTeam, and the org PowerShellForGitHubTeamTestOrg). You will see the AccessToken being referenced there as well...it is stored, encrypted, within Azure DevOps. It is not accessible for use outside of the CI pipeline. To run the tests locally with your own account, see configuring-your-environment.

Your change must successfully pass all tests before they will be merged. While we will run a CI build on your behalf for any submitted pull request, it's to your benefit to verify your changes locally first.

New Test Guidelines

Your tests should have NO dependencies on an account being set up in a specific way. They should get the configured account set up in the appropriate state that it can then test/verify. In this way, anyone should be able to run the tests from their own machine/account.

Use a new GUID for any object that you have to create (repository, label, team name, etc...) to avoid any possible name collisions with existing objects on the executing user's accounts.


Releasing

When new code changes are checked in to the repo, most users of the module will not see those changes unless an updated module gets published by Microsoft to PowerShell Gallery.

The general guidance on publishing an update is that changes should not be in master more than one week without having been published through PowerShell Gallery as well.

When you are ready to publish a new update, the following steps are necessary:

Updating the Version Number

Whenever new changes to the module are to be released to PowerShellGallery, it is important to properly update the version of the module. The version number is stored in the module manifest (PowerShellForGitHub.psd1), and it should be updated following the Semantic Versioning standard.

The update to the module manifest should happen in the same changelist where the CHANGELOG is updated.

Semantic Versioning

This project follows semantic versioning in the following way:

<major>.<minor>.<patch>

Where:

  • <major> - Changes only with significant updates.
  • <minor> - If this is a feature update, increment by one and be sure to reset <patch> to 0.
  • <patch> - If this is a bug fix, leave <minor> alone and increment this by one.

Updating the CHANGELOG

To update CHANGELOG.md, just duplicate the previous section and update it to be relevant for the new release. Be sure to update all of the sections:

  • The version number
  • The hard path to the change (we'll get that path working in a moment)
  • The release date
  • A brief list of all the changes (use a - for the bullet point if it's fixing a bug, or a + for a feature)
  • The link to the pull request (pr) (so that the discussion on the change can be easily reviewed) and the changelist (cl)
  • The author (and a link to their profile)
  • If it's a new contributor, also add them to the Contributors list below.

Then get a new pull request out for that change and for the change to the module manifest's version number.

Adding a New Tag

To add a new tag:

  1. Make sure that you're in a clone of the actual repo and not your own private fork.
  2. Make sure that you've already merged in the change that updates the module version.
  3. Make sure that you have checked out master and that it's fully up-to-date
  4. Run git tag -a '<version number>'
  5. In the pop-up editor, just copy the content from the CHANGELOG that you just wrote, but remove any of the ### heading blocks since those will be dropped from git as comments instead of headings.
  6. Save and close the editor
  7. Run git push --tags to upload the new tag you just created

If you want to make sure you get these tags on any other forks/clients, you can run git fetch origin --tags or git fetch upstream --tags, or whatever you've named the source to be.

Doing this makes it possible for users to simply run git checkout <version number> to quickly set their clone to the state of any previous version. It also has the added benefit that GitHub will automatically create a new "Release" in the Releases tab of the project for this new version.

Running the Release Build

A YAML definition exists that will run the equivalent of the CI build, followed by the necessary steps to sign the module files and publish the update to PowerShell Gallery. This YAML file can only be run by a Microsoft maintainer because it accesses internal services to sign the module files with Microsoft's certificate.

Microsoft Maintainers: You can access the internal pipeline which can execute the release build here. Simply hit Queue to get a new module released.

Instructions for updating the PowerShellGalleryApiKey secret in the pipeline can be found in the internal Microsoft repo for this project.

Updating the Wiki Documentation

The Wiki contains the full documentation for all exported commands from the module, thanks to platyPS.

Every time a new release occurs, the Wiki should be updated to reflect any changes that occurred within the module.

  1. Ensure that you have cloned the Wiki:

    git clone https://github.com/microsoft/PowerShellForGitHub.wiki.git
    
  2. Open a PowerShell 7+ console window (don't use Windows PowerShell as there's a platyPS bug with that version regarding multi-line examples) and navigate to your Wiki clone.

  3. Run this command (assuming that you have a PowerShellForGitHub clone at the same level as your Wiki clone):

    ..\PowerShellForGitHub\build\scripts\Build-Wiki.ps1 -Path .\ -RemoveDeprecated -Verbose -Force
    
  4. Verify the changes all make sense. You will also need to manually copy the core content of PowerShellForGitHub.md into Home.md. For the time being, we are duplicating that content in Home until such time as we have better content to put there.

  5. Commit the change and directly push it to the Wiki's master branch...no need to go through a pull request for the Wiki changes.

This is not currently automated as part of the Release pipeline because I don't currently want to store any credentials/tokens with write access to the repo in the pipeline.


Contributors

Thank you to all of our contributors, no matter how big or small the contribution:


PowerShellForGitHub is licensed under the MIT license.

You will need to complete a Contributor License Agreement (CLA) for any code submissions. Briefly, this agreement testifies that you are granting us permission to use the submitted change according to the terms of the project's license, and that the work being submitted is under appropriate copyright. You only need to do this once.

When you submit a pull request, @msftclas will automatically determine whether you need to sign a CLA, comment on the PR and label it appropriately. If you do need to sign a CLA, please visit https://cla.microsoft.com and follow the steps.