I'm using Turborepo for my monorepo project, i have 2 react apps. How can i configure Turborepo and CircleCI (repos are on Github) so if i make changes to one project that pipeline is not going to run for second project?
I know turbo is using hash algo to check if there is any changes to a project and then rebuild it.
I have tried looking here https://turborepo.org/docs/ci/circleci but does not explain the behavior of this.
Steps would be:
Make code change to Project 1
Commit changes of monorepo to Github
Github detects a commit and triggers CircleCI to run CI/CD
So this part is what I'm not sure, if it triggers CI/CD it will trigger for the both projects right? And if so how can i prevent only for the one i have made changes?
I've been working on such a solution for days now. There are two core-concepts in turborepo to achieve this:
Filtering workspaces
Caching Buildoutputs and store the cache in the cloud (not what you're looking for)
So, you can filter your monorepo for a specific project, e.g:
pnpm turbo run build --filter='my-project...[HEAD^1]' --dry=json
-> This will look if the task build is needed to run for the project "my-project", comparing the current source with "HEAD^1". The option dry=json helps to just look if there would be a need to run "build" or not for "my-project".
You could filter a whole lot more, check the docs.
Now, what i have built on top of this:
A new job on the github workflow looks with the help of this filter command if a deployment of my graphql-server is needed and he will set the output of this decision as an artifact, to provide this information for later jobs (https://github.com/actions/upload-artifact)
My actual docker-build and deploy-to-fly-io jobs that run afterwards, will download this artifact and set a CONTINUE environment variable, depending if it should build + deploy or not.
Every job coming after that is having an if: ${{ env.CONTINUE == 'true' }} to skip them if no build/deploy is needed.
It could be much simpler if you can run your build/deploy job directly with the turbo cli, because then you can just combine your filter and the execution of the build - but that was not possible in my case.
If you need to "skip" jobs that are coming later in your workflow, it's harder thant it should, as github is not supporting "abortion" of jobs.
For all the other commands like lint, typecheck and test -> just add an appropriate filter option to them and you will achieve that they only run on your "affected" workspaces/projects in your PR.
Ressources:
https://dev.to/ngconf/deploying-nx-affected-apps-from-github-actions-54n4
How can I get the previous commit before a push or merge in GitHub Action workflow?
https://github.com/orgs/community/discussions/26313
I have my project in huge monorepo on non machine that has poor I/O performance etc. git status takes ages. When I run npm test (app was create with create-react-app), there's an initial step to that is determining tests to run. I assume that jest is figuring it out based on files changed.
Is there a way to configure jest to ignore git repo features completely and simply rerun all tests at startup and later run tests only for files that were affected by currently changed file (I guess that could be possible with ES6 imports) or always rerun all tests when watched files are changed?
my current front-end dev setup is below:
browserify as build tool
budo as a development server
I have a couple of shared modules packaged up and published on npm for use as a dependency among many projects.
However the development feedback cycle is too long since I have to run npm link ../<repo> && npm run dev every time I need to see an updated change and it takes too long to finish linking , approx. 2-5 minutes.
Is there a way I can watch changes in my link'd module and it will rebuild files that were changes?
Budo (browserify) should follow symlinks, so once you npm link then you should be able to edit both modules concurrently (and get live reload) without having to re-run any commands. This is assuming your other module doesn't need a transpile step. If it isn't working, perhaps it's a bug or some other issue. Feel free to open an issue on budo's GH issues to discuss further.
– #mattdesl
Currently I have an Angular application built on top of NG6-starter. The NG6-Starter uses webpack to bundle the application. It also supports unit testing by default via Karma, but karma uses webpack as well in order to get all the sources injected to it's browser.
Now, my issue is, that I would like to add my application to a classic CI/CD pipeline: Some static analysis, then package build, then unit tests, then etc, and I do not want to break the principle of "Test against the artifact that you are going to deploy" principle. Since karma currently builds the application for itself, it does not really rely on the artifact, that is going to be deployed, even though by the stage karma runs, it is already built.
My question is, if you have any idea/practice/example/experience with this topic?
Okay, so it seems so, that in order to achieve this, the spec files should be built with webpack during build time. For this, the easiest way seemed to be to simple introduce a new chunk entry point to the application.
I am trying to figure out a simple way to run tests on angularjs application.
I am new to the testing world, so it's a little hard to understand all the options and the difference between them.
My goal: to be able to run the tests simply from within my IDE - Eclipse.
And to tests the code on google chrome browser.
I found jasmine to be the obvious choice for writing js unit tests. The problem is choosing a runner both for the jasmine tests and for e2e tests.
Trying to keep it simple, I've come up with the following idea for a setup:
Write the unit tests in jasmine, and the e2e tests in phantomjs and syn.js.
Then configure eclipse to run phantomjs as an external tool, so that the output will go to the console in eclipse.
I also plan to have a CI job in Jenkins, and to my understanding Jenkins can also run phantom, so theoretically this solution will work the same for CI.
Alternatively, there are test running tools like Karma and Protractor. On one hand, they seem to be recommended, but on the other hand they seem to me like overkill in some cases. They require a lot of different tools/services/processes to be running in order to work, and it seems like a pain maintain all that setup if it breaks.
To my understanding: protractor runs on webdriverjs which runs on nodejs, and it requires a selenium server to be running in the background, and on top of all that the selenium opens real browser windows which seems a little pointless as opposed to headless browser testing.
Then there is Karma, that I did not yet fully understand what it's supposed to do. From what I've read it monitors the files in my project and whenever a file is changed it runs the tests. I'm not sure how it runs the tests - is it also using selenium?
And lastly, there are grunt and yeoman, which I did not understand at all what they do and how they interact or fit together with the other tools I've listed.
I would appreciate if someone could clarify what these different tools do, and how they fit together. Also, how would they fit with Jenkins as a CI server?
Also if you could comment on my "simpler setup" - does it make sense? Am I missing something?
Karma is for unit testing your JS, regardless of whether it is using Angular or not. The ins and outs of unit testing with Karma are covered very well here: http://www.yearofmoo.com/2013/01/full-spectrum-testing-with-angularjs-and-karma.html. Yes, Karma opens and closes browser windows as needed and specified in the configuration file. If you don't want any browser windows opened, you can use PhantomJS. You can run Karma from within most any IDE that is capable of running an external script, or run it via the command line.
Protractor is for end-to-end (or E2E) testing of your project as a whole. It will open a browser window and click through the pages as though it were a user, entering data where you tell it to and looking for the specified results. Protractor is a bit more complicated than just writing some Jasmine, but the results are worth it. Like Karma, you can run Protractor from within most any IDE that is capable of running external scripts or via the command line.
Yeoman is a process management system that incorporates dependency management via Bower, task automation via Grunt, and project management via Yo. It will run your tests in Karma and Protractor, minify your JS, CSS, and HTML, compile everything into appropriate files (internal JS, external libraries, and CSS) and provide you with a complete package that can be deployed. The beauty of Yeoman is that it is not specific to any one IDE. Everything it does can be done by scripting in your IDE or via the command line.
Now, having said all of this about Yeoman, you do still have to write the tests (it won't magically come up with them for you) and learn to integrate it into your development routine, but it is definitely the way to go for JS development. Eclipse is fine for JS development, but you'll get better performance and ease of use (IMHO) from WebStorm.
As for how these all fit into CI like Jenkins, I believe that both Karma and Protractor output test results in a format that Jenkins can read and display. With the scripting possibilities in Jenkins you can configure it to run the build process each time your source control repository (you are using some sort of source control, aren't you?) changes and show those results on the Jenkins page. My office has a very similar setup and we use it daily. I'm not the guy that has to do the Jenkins configuration, but I do work with Yeoman (and thus Karma and Protractor) via WebStorm on a regular basis and have had very good results.
I would say the clear choice here is Karma and Protractor. While it is true that they rely on a bunch of other stuff, they do so pretty antiseptically: protractor starts up the selenium server and then shuts it off when it's done. Once you have node installed, the other installations are all super simple. I would also install httpster, which will serve up your public director on port 3333.
Frankly, having come from a decade of doing TDD in the Java world, when I first looked at Javascript a few years ago (again), the testing picture was a complete joke. But now, I think the combination of Karma and Protractor is pretty fantastic. Inside IntelliJ, you can run the Karma tests and they are stupid fast and the results are presented in a runner that's as good as anything I've seen in the Java world (Xcode 5 has the best test integration). You can also install the ddescribe plugin in IntelliJ and have a ui for running individual tests or excluding tests.
On the protractor side, I found this post because I am at the point now where I am going to run my karma, protractor and then JUnit tests on a continuous integration server (either Jenkins or TeamCity). I was kind of surprised at the paucity of info on that leg of the trip, but the clear direction I see there is Grunt, because it will run your protractor tests then generate the JUnit-style output Jenkins wants. Grunt is also a pretty impressive addition to the JS world.
I know this sounds like a bunch of opinions, but I think that as happened in the Java world, the Javascript world has now reached that level of maturity where you are just going to have to expect things to drag other things in with them. Frankly, looks like node and npm do a pretty nice job of making that pretty seamless (vs. a decade down the drain on Maven in the Java world).
Updated: Sorry I did not read your question properly.
karma is a test runner, which is best suited for jasmine. For setting up is very very easy. Please download node, and install npm install karma. Follow the angular seed sandbox project it contains all the basic config set up for unit testing and end to end testing (in config folder).all you need is nodejs plugin installed in eclipse
Yeoman can be used for javascript minification, sass compilation e.t.c.
Install node eclipse and you can set all up in eclipse.
http://www.nodeclipse.org/