dependency-cruiser VS wireit

With dependency-cruiser, you can enforce which imports are allowed. This enables you to create an architecture fitness function that ensures your code continues to adhere to the initial design. You can also visualize your dependencies to gain a clearer understanding of your code's actual structure, allowing you to compare it with your mental model and make improvements where necessary.

something like https://github.com/sverweij/dependency-cruiser maybe https://github.com/pahen/madge or https://github.com/antoine-coulon/skott

I've personally fallen in love with Dependency Cruiser, which lets you set any arbitrary import rules you want on your repository. With it, you can enforce common things, like, "You can only import through the index file, if one exists", but you can also make custom-tailored rules for your specific project. For example, maybe your project is divided into three large folders - folder1 is allowed to import from folder2 and folder3, folder2 can import from folder3, and folder3 can not import from anyone else. Well, you can enforce that too, or whatever you need.

Dependency Cruiser works great, can even render visuals:

https://github.com/sverweij/dependency-cruiser

NX[0] also has logic for handling this issue

[0]: https://nx.dev/

I have a dependency-cruiser script that enforces codebase import rules, which I want checked on each dev hot-reload build and prod build.

Great project!

One feature request: Running the npx command searched only for the js files, not for the ts files. When I built deprank locally with yarn, it also showed the ts files. After looking at dependency-cruiser figure it has to do with what typescript compilers are available where.

It would be great if the npx command you provide in your readme would work regardless of my local setup - dependency-cruiser has documentation and one example of a suitable npx command here: https://github.com/sverweij/dependency-cruiser/blob/develop/...

My suggestion would be to check if any ts file is part of the extension option (i.e. --ext=".js,.jsx, .ts, .tsx") and only then do the magic needed to also show ts files.

The first step is to install dependency-cruiser.

You may have a look at dependency cruiser.

npm workspaces plus Wireit works far better than Lerna, in my experience.

https://github.com/google/wireit

Wireit's ability to specify actual script dependencies, do caching (and on Github actions), and it's long-running service script support make it much more useful and comprehensive than Lerna.

I agree that this should be built into npm. There's an RRFC for it here: https://github.com/npm/rfcs/issues/706

I must admit I'm a bigger fan of the wireit[0] approach, the only pause I have is its a Google project, my temptation is to fork it. The code isn't terribly complex

My biggest complaint with NX is: lack of a sane API for plugins, and it has more overhead than I'd care for. For the amount of complexity that NX has, I'd rather use Rush[1] which gives you everything NX does. My only complaint with Rush is that its development is really slow going, they really need to focus up on Rush plugins (they're good, but still experimental, and I'd love to see them clean up how `autoinstalls` work to be more intutive)

I'm on the fence about turbo from Vercel

[0]: https://github.com/google/wireit

[1]: https://rushjs.io/

To further derail the conversation there's also https://github.com/google/wireit

There's also wireit made by Google which pairs well with Yarn/NPM workspaces

Google recently anounced wireit, a program that runs multiple NPM scripts that depend on one another. Combined with NPM Workspaces, it enables monorepo workflows that previously required tools like Yarn and Pnpm.

etc.

where a bunch of related projects live top-level in a repo. Each project has a packages folder that includes the core implementation, as well as demos, framework-specific adaptors, etc.

In each package's package.json, I have a series of commands (convert the TS to JS, make a bundle, deploy to Firebase, etc.). Each command can depend on another, either in the same project or anywhere else in the file hierarchy.

This provides two benefits:

1. Iterating across packages is faster, because I don't have to worry about making sure each package rebuilds in the right order if I make a change in a library.

2. Filesystem concerns are separated: rollup only needs to worry about bundling, and it only needs to bundle web-facing projects. The only tool my libraries need is tsc.

(Using TypeScript and Rollup together is kind of a pain in the ass because you have to fiddle with picking the right TS plugin and configuring it. This is also often the long pole on doing a Rollup version upgrade. Decoupling the two makes Rollup way simpler/easier/nicer to use, which makes wireit awesome even if you don't have multiple packages.)

Here's a snippet from one of my package.jsons. They basically all look like this. (start is complicated because of https://github.com/google/wireit/issues/33. When that's resolved, it will be as simple as the others.)

    "scripts": {

Hi HN!

I have been doing over month long of research in terms of figuring out the best way to manage a growing monorepo. We are trying to consolidate much of our frontend code base into a monorepo, managed by pnpm workspaces, to consolidate dependency management and take advantage of tool (as well as other code sharing benefits) of monorepos that are a good fit for us.

To that end, I'm looking to understand if anyone has used Bazel extensively for managing monorepos.

I want to understand how easy it is to configure Bazel, how easy it is to use Bazel, especially newer developers (particularly self discovery of the toolset), how easy it is to maintain it, and how much burden the tool has placed on developers. We really are looking for a tool that is largely self sufficient for the purpose.

Main features we care about:

- Maintainability: is it is to maintain (updates etc)

- Extensibility: how extensible is it? more importantly, how easily can it be extended?

- Built in watch mode that understands its dependency graph for each task, and can run them simultaneously

- Works with pnpm / npm workspaces natively

- Stream based output: e.g. if running multiple tasks it interleaves them appropriately, even better if they're labeled and color coded

- Dependency graph tracking. IE: if I run build for a package, it understands that it may have dependencies that need to be built first.

- Able to run tasks arbitrarily on a "per package" process, potentially

Now, after mentioning all that, I realize, by reading the docs, in theory Bazel supports all this and has lots of feature headroom for growing features over time which I like, however, I've read mixed things about it, but not all of the sources I've read so far are "up to date" (some articles about people adopting Bazel are years old now) and I wanted to get a more accurate picture of what is going on here.

Alternatively, I'm open minded to looking at a different set of tools

For context I've done alot of research and experimentation with the follow:

- nx[0]

- rush[1]

- wireit[2]

- turbo[3]

We've settled on, for now `wireit` in part because it has a really good watch mode feature that `nx` does not (nx doesn't have a built in watch mode for your task runner, it relies on the plugin / script to handle it, which was really problematic). However, wireit isn't extensible, and I'm not looking to have to manage sub task "phasing" with something like `gulp`. This was an issue with rushjs as well (but rushjs has its own challenges and opinions). While rush is starting to expose a direct `rush-sdk` API, its not really documented and I'm not sure about its stability or best way to go about making rush plugins. They also have a competing task runner called `heft` that I'm not sure about in the light of the `rush-sdk` and its use cases (if someone from the rush team sees this and can clarify about the long term vision and where they're at with it now, I'm all ears)

tl;dr: I've tried tons of tools, and Bazel seems to check all the boxes, but I'm afraid the complexity will kill us, since we don't have a dedicated tool engineer to oversee it, it has to malleable enough that we can maintain it bit by bit over time

[0]: https://nx.dev/

[1]: https://rushjs.io/

[2]: https://github.com/google/wireit

[3]: https://turborepo.org/docs

In the past, I'd put a "typescript:main" field in package.json and configured my bundler to prefer that field. I gave up at some point - probably when I migrated to rollup.

Moving forward, I'm going to use wireit for these things. Pure modules get built with tsc. At the highest level (e.g. where it needs to be embedded in a page), make a bundle with rollup.

wireit has two nice properties: incremental building and file-system-level dependencies. Within a repo, you can depends on ../package-b. However, if you have multiple monorepos that often get used together, you can also depend on ../../../other-package/packages/package-b.

I've just started with wireit (it was only launched recently), but it seems to be a nice solution to wrangling dependencies between related JS libraries.

[1] https://github.com/google/wireit