CapsuleCD is a generic Continuous Delivery pipeline for versioned artifacts and libraries written in any language. It’s goal is to bring automation to the packaging and deployment stage of your library release cycle. CapsuleCD is incredibly flexible, and works best when implemented side-by-side with a CI pipeline.
A short list of the features…
- Supports libraries written in any language. Has built-in support for
- Chef Cookbooks
- Python Pip
- NodeJS Npm Packages
- Ruby Gems
- Highly configurable
- Follows language/library best practices. Including things like:
- automatically bumping the semvar version number
- regenerating any
*.lockfiles/ shrinkwrap files with new version
- creating any recommended files (eg.
- validates all dependencies exist (by vendoring locally)
- running unit tests
- source minification
- linting library syntax
- generating code coverage reports
- updating changelog
- uploading versioned artifact to community hosting service (rubygems/supermarket/pypi/etc)
- creating a new git tag
- pushing changes back to source control (github)
- creating a new release in source control (github) and attaching any common artifacts
What is CapsuleCD
CapsuleCD is a generic Continuous Delivery pipeline for versioned artifacts and libraries written in any language. It’s goal is to bring automation to the packaging and deployment stage of your library release cycle. It automates away all the common steps required when creating a new version of your library.
Why use CapsuleCD
At first glance, it seems simple to publish a new library version. Just bump the version number and publish, right? Well, not always:
- If you’re library includes a Gemfile.lock, Berksfile.lock or other common lock files, you’ll need to regenerate them as the old version number is embedded inside.
- Everyone runs their library unit tests before creating a new release (right?!), but what about validating that your library dependencies exist (maybe in your Company’s private repo)?
- How about linting your source, to ensure that it follows common/team conventions?
- Who owns the gem? Is there one developer who has the credentials to push to RubyGems.org? Are they still on your team/on vacation?
- Did you remember to tag your source when the new version was created (making it easy to determine what’s changed between versions?)
- Did you update your changelog?
CapsuleCD handles all of that (and more!) for you. It pretty much guarantees that your library will have proper and consistent releases every time. CapsuleCD is well structured and fully tested, unlike the release scripts you’ve manually cobbled together for each library and language. It can be customized as needed without rewriting from scratch. The best part is that CapsuleCD uses CapsuleCD to automate its releases. We dogfood it so we’re the first ones to find any issues with a new release.
How do I start?
You can use CapsuleCD to automate creating a new release from a pull request or from the latest code on your default branch.
Automated pull request processing:
Here’s how to use docker to merge a pull request to your Ruby library
CAPSULE_SOURCE_GITHUB_ACCESS_TOKEN=123456789ABCDEF / CAPSULE_RUNNER_REPO_FULL_NAME=AnalogJ/gem_analogj_test / CAPSULE_RUNNER_PULL_REQUEST=4 / CAPSULE_RUBYGEMS_API_KEY=ASDF12345F / docker run AnalogJ/capsulecd:ruby capsulecd start --source github --package_type ruby
Or you could install and call CapsuleCD directly to merge a pull request to your Python library:
gem install capsulecd CAPSULE_SOURCE_GITHUB_ACCESS_TOKEN=123456789ABCDEF / CAPSULE_RUNNER_REPO_FULL_NAME=AnalogJ/pip_analogj_test / CAPSULE_RUNNER_PULL_REQUEST=2 / CAPSULE_PYPI_USERNAME=AnalogJ / CAPSULE_PYPI_PASSWORD=mysupersecurepassword / capsulecd start --source github --package_type python
Creating a branch release
TODO: add documentation on how to create a release from the master branch without a pull request. Specify the env variables required.
Every package type is mapped to an engine class which inherits from a
BaseEngine class, ie
RubyEngine etc. Every source type is mapped to a source module, ie
GithubSource . When CapsuleCD starts, it initializes the specified Engine, and loads the correct Source module. Then it begins processing your source code step by step.
|source_configure||This will initialize the source client, ensuring that we can authenticate with the git server|
|runner_retrieve_payload||If a Pull Request # is specified, the payload is retrieved from Source api, otherwise the repo default branch HEAD info is retrived.|
|source_process_pull_request_payload or source_process_push_payload||Depending on the retrieve_payload step, the merged pull request is cloned, or the default branch is cloned locally|
|build_step||Code is built, which includes adding any missing files/default structure, compilation, version bumping, etc.|
|test_step||Download package dependencies, run the package test runner(s) (eg. npm test, rake test, kitchen test, tox)|
|package_step||Commit any local changes and create a git tag. Nothing should be pushed to remote repository|
|release_step||Push the release to the package repository (ie. npm, chef supermarket, rubygems)|
|source_release||Push the merged, tested and version updated code up to the source code repository. Also do any source specific releases (github release, asset uploading, etc)|
PYPI_PASSWORD via an environmental variable might make sense, but do you really want to specify the
REPO_FULL_NAME each time? Probably not.
CapsuleCD has you covered. We support a global YAML configuration file (that can be specified using the
--config-file flag), and a repo specific YAML configuration file stored as
capsule.yml inside the repo itself.
CapsuleCD settings are determined by loading configuration in the following order (where the last value specified is used)
- system YAML config file (
- repo YAML config file (
- environmental variables (setting in capital letters and prefixed with
|Setting||System Config||Repo Config||Notes|
|package_type||No||No|| Must be set by
|source||No||No|| Must be set by
|runner||No||No|| Must be set by
|dry_run||No||No|| Must be set by
|source_git_parent_path||Yes||No||Specifies the location where the git repo will be cloned, defaults to tmp directory|
|source_github_api_endpoint||Yes||No||Specifies the Github api endpoint to use (for use with Enterprise Github)|
|source_github_web_endpoint||Yes||No||Specifies the Github web endpoint to use (for use with Enterprise Github)|
|source_github_access_token||Yes||No||Specifies the access token to use when cloning from and committing to Github|
|runner_pull_request||Yes||No||Specifies the repo pull request number to clone from Github|
|runner_repo_full_name||Yes||No||Specifies the repo name to clone from Github|
|chef_supermarket_username||Yes||Yes||Specifies the Chef Supermarket username to use when creating public release for Chef cookbook|
|chef_supermarket_key||Yes||Yes||Specifies the Base64 encoded Chef Supermarket private key to use when creating public release for Chef cookbook|
|chef_supermarket_type||Yes||Yes||Specifies the Chef Supermarket cookbook type to use when creating public release for Chef cookbook|
|npm_auth_token||Yes||Yes||Specifies the NPM auth to use when creating public release for NPM package|
|pypi_username||Yes||Yes||Specifies the PYPI username to use when creating public release for Pypi package|
|pypi_password||Yes||Yes||Specifies the PYPI password to use when creating public release for Pypi package|
|engine_disable_test||Yes||Yes||Disables test_step before releasing package|
|engine_disable_minification||Yes||Yes||Disables source minification (if applicable) before releasing package|
|engine_disable_lint||Yes||Yes||Disables source linting before releasing package|
|engine_cmd_test||Yes||Yes||Specifies the test command to before releasing package|
|engine_cmd_minification||Yes||Yes||Specifies the minification command to before releasing package|
|engine_cmd_lint||Yes||Yes||Specifies the lint command to before releasing package|
|engine_version_bump_type||Yes||Yes|| Specifies the Semvar segment (
TODO: specify the missing `BRANCH` release style settings.
As mentioned above, all settings can be specified via Environmental variable. All you need to do is convert the setting to uppercase and then prefix it with
CAPSULE_ . So
pypi_password can be set with
Example System Configuration File
Here’s what an example system configuration file might look like:
source_git_parent_path: /srv/myclonefolder source_github_api_endpoint: https://git.mycorpsubnet.example.com/v2 source_github_web_endpoint: https://git.mycorpsubnet.example.com/v2
Step pre/post hooks and overrides
CapsuleCD is completely customizable, to the extent that you can run your own Ruby code as
post hooks before every step. If that’s not enough, you can also completely override the step itself, allowing you to use your own business logic. To add a
post hook or override a step, just modify your config
yml file by adding the step you want to modify, and specify
override as a subkey. Then specify your multiline ruby script:
--- source_configure: pre: | # this is my multiline ruby script # the pre hook script runs before the actual step (source_configure) executes # we have access to any of the specified instance variables here. # check the documentation for more information. puts "override pre_source_configure" `git clone ...` override: | # override scripts can be used to completely replace the built-in step script. # to ensure that you are compatible with the capsulecd runner, please ensure that you # populate all the correct instance variables. # see the documentation for more information puts "override source_configure" post: | # post scripts run after the step (source_configure) executes # you can override any instance variables here, do additional cleanup or anything else you want. puts "override post_source_configure" build_step: post: | # post build step runs after the build_step runs # within the script you have access to all instance variables and other methods defined in the engine. puts "override post_build_step" + @source_git_local_path
Test suite and continuous integration
CapsuleCD provides an extensive test-suite based on rspec and a full integration suite which uses VCR. You can run the unit tests with
rake test . The integration tests can be run by
rake 'spec:<package_type>' . So to run the Python integration tests you would call
rake 'spec:python' .
CircleCI is used for continuous integration testing: https://circleci.com/gh/AnalogJ/capsulecd
If you’d like to help improve CapsuleCD, clone the project with Git by running:
$ git clone git://github.com/AnalogJ/capsulecd
Work your magic and then submit a pull request. We love pull requests!
If you find the documentation lacking, help us out and update this README.md. If you don’t have the time to work on CapsuleCD, but found something we should know about, please submit an issue.
We’re actively looking for pull requests in the following areas:
- CapsuleCD Engines for other languages
- Objective C
- Any others you can think of
- CapsuleCD Sources
- Any others you can think of
We use SemVer for versioning. For the versions available, see the tags on this repository.
Jason Kulatunga – Initial Development -@AnalogJ
CapsuleCD is licensed under the MIT License – see theLICENSE.md file for details
转载本站任何文章请注明：转载至神刀安全网，谢谢神刀安全网 » Continuous Delivery for Versioned Artifacts(Npm, Chef, Gems, Pip, Etc)