A few weeks ago, on december 1st 2022, Werner Vogels announced Amazon CodeCatalyst. I’ve previously shared my initial thoughts and findings in a blog post. In this post, I’m going to share a few more findings and insights into using Amazon CodeCatalyst and will also see if any of my wishes from the wishlist for CI/CD on AWS have been resolved with CodeCatalyst.
What I have been playing around with…
One of my personal projects that I am working on together with a few friends is pegasus-galaxy.net and the CI/CD pipeline that I had built with CDK Pipelines (that I also presented at re:Invent 2022) was the first one to try to move over.
In context, we’re talking about a Flutter application for Web running behind CloudFront, deployed using CDK.
I decided to try CodeCatalyst out and go “all in” – and that means moving the code from Bitbucket into CodeCatalyst as well as setting up the other users in CodeCatalyst and moving the workflows (=CI/CD pipelines) over to CodeCatalyst.
In this article I am going to go through each of the sections in CodeCatalyst and will share my experiences, thoughts and findings.
Where I have ideas on how to improve the day-to-day work with the tool, will try to share that.
Before going into details, lets start with the most important thing:
Amazon CodeCatalyst works very well and reliable and the current version of the service is a great foundation for moving all of your CI/CD and development practices to AWS.
The CodeCatalyst team has been very supportive on re:Post, so if you have a question, feel free to ask it there!
CodeCatalyst Overview – Spaces and Projects
Spaces are the “Top-Level” option to organize your CodeCatalyst account. You will need to associate an AWS Account for billing used AWS resources. Each AWS (billing) account can be associated only with one CodeCatalyst Space.
While this seems like a limitation as you will need to create a different billing account for a 2nd space, I can right now not see an impact for my day to day work. For anything that I run on the same AWS account, I would assume that using a project within the same space should be good.
You can manage Projects, Members and AWS Account connections on the space page. In the “extensions”, CodeCatalyst currently allows a connection only to the JIRA Cloud. I would expect that additional 3rd party extensions will be supported in the GA version of CodeCatalyst.
Projects Overview and options
A project is a “unit of work” in your product or software that you are building.
Within projects, you can manage issues, manage your code repositories, execute workflows (CI/CD) and review report results.
Projects are associated to a Space – and you can create as much projects in a Space as you want. You can add team members to a project, that are not able to access all projects in the space. Unfortunately I have not yet found an option to “hide” projects from Team members that are added on the Space itself.
Managing issues / tickets
CodeCatalyst currently provides two options to manage your issues or tasks:
1) Link to JIRA Cloud Project
2) Internal issue management
If you use the option to link to a JIRA cloud project, the “issues” link is replaced by a link to your JIRA Cloud project.
Internal issue management
The internal issue management system currently offers everything that is required for a simple Kanban workflow. You can create issues, add them to a backlog or a Kanban board, assign them to project members and track their current status.
I personally think that the current functionalities are good enough for small teams and simple projects – I’m actually already working with it in a small project and will add additional feedback as soon as I gain more experience.
Within the “Source” part of a project, you can manage source repositories or connections to source repositories in Github. I expect that other providers will be added going forward (e.g. Gitlab, CodeCommit, Bitbucket, …).
You can also manage pull requests and approvals – I was only able to test this using internal source repositories, not using a linked repository.
The last option – the Dev Environments – is the most exciting functionality – it gives you the possibility to host development environments (similar to Gitpod) on AWS using Cloud9 but also, and this is really cool, using Visual Studio or JetBrains IDEs.
When using that option, the IDE on your local PC is only the “presentation layer”, the source code is stored and run on an AWS instance and the IDE uses remote connectivity to talk to the Dev Environment in the background.
CodeCatalyst currently uses the same approach as Github Actions to manage your workflows or CI/CD pipelines – you are able to manage your Workflows using YAML files. The syntax is simple and understandable. There is a minimal set of Actions available as part of the preview. You are also able to use existing Github Actions as part of your workflow.
The workflow functionality is very powerful. In my tests I have not yet been able to test all parts of the capabilities. Workflows can be defined for certain directories, for certain triggers or branches. Test reports will be exposed in the “reports” functionality.
CodeCatalyst offers a graphical overview for workflows and alows to edit them in the UI, too. This functionaly works pretty well and helps to quickly get you started building your first workflow in CodeCatalyst.
I’ll need to test the workflows more to be able to give additional insights into how good or bad they are currently running. My simple pipeline that builds a Flutter Application, deploys my Infrastructure as Code using CDK and then publishes the new version of the Flutter app runs without problems.
One of my main concerns so far is the execution time, however the team has been working on a possibility to use Lambda as an execution environment.
This option however does not yet support the execution of Github actions and also has some other limitations.
The other features that are part of the “CI/CD” – Environments, Compute and Secrets – I did not have time to play around with this. If you have any experiences with it, please add your thoughts in a comment to this article!
The reports today only suport test reports. I have not used the functionality enough to assess this, but I am sure that the CodeCatalyst team is going to add additional reporting options going forward.
Things I like most about CodeCatalyst (Preview) after 6 weeks of usage
Just a short list of things that I already like:
– Integration of Github Actions as workflow actions
– Managing workflows using UI & code
Things I miss in CodeCatalyst (Preview) after 6 weeks of usage
– macOS builds (e.g. for Flutter iOS apps) are still not possible
– granular permissions for workflow and Pull Request triggers
Let’s talk about Open Source Projects
Right now there is no option to share a project or a repository that is hosted within CodeCatalyst as an Open Source project. This is really a limitation if you want to use CodeCatalyst for Open Source project – or if I would like to share a CodeCatalyst repository with example workflows.
I hope this functionality will be added soon.
Wrap up and next steps for me with CodeCatalyst
I need to admit – writing this post took longer than expected 🙂
I wanted to publish it before christmas and now it seems to be a bit “late” already as I am sure that a lot of you have made your own experiences with CodeCatalyst today – please SHARE your findings with me – links of Blogs that you have written or other content you have created, I am eager to consume it!
My next steps with CodeCatalyst
I am working on migrating my project pegasus-galaxy.net completely to CodeCatalyst and collaborate with my team on it there. With that, I will be able to proof CodeCatalyst in a “real world” application that it is “multi-platform” application – using Flutter for Web, Android and iOS – and a Serverless AWS based backend.
If you’re interested to join this project, please do not hesitate to reach out – skills that we need right now:
AppSync, DynamoDB and development/software engineering (Flutter, Typescript, Java, or Node?)