stack.yaml versus package.yaml versus a Cabal file

What is the difference between a stack.yaml file, a package.yaml file and a Cabal file (named <package_name>.cabal)? This page aims to make that clear.

In short:

• stack.yaml contains project-level configuration for Stack, and may contain project-specific options and non-project-specific options.

• package.yaml contains a description of a package in the Hpack format. Hpack, including Stack's built-in version, uses the file to create a Cabal file.

• a Cabal file also contains a description of a package, but in the format used by Cabal.

package.yaml versus a Cabal file

Why two different formats to describe packages? Hpack is considered to have some advantages over the underlying Cabal format, which are explained its project repository. They include that the Hpack format supports YAML syntax and the automatic generation of the lists of exposed-modules used in the Cabal format.

The remainder of this page will focus on the difference between a stack.yaml file and a package description file.

Package versus project

Stack is a tool for building Haskell code and it uses Cabal, a build system. Cabal defines the concept of a package. A package has:

• A name and version
• optionally, one library
• optionally, one or more executables
• A Cabal file (or, as mentioned above, an Hpack package.yaml file that generates a Cabal file)
• And a bunch more

There is a one-to-one correspondence between a package and a Cabal file.

Stack defines a new concept called a project. A project has:

• A snapshot (more on this later)
• Extra dependencies on top of the snapshot
• Optionally, one or more local Cabal packages
• Flag and GHC options configurations
• And a bunch more Stack configuration

Often you will have a project that defines only one local Cabal package that you are working on. If you need to specify a dependency, a source of confusion can be why you need to specify it both in the stack.yaml file and in the Cabal file. To explain, let us take a quick detour to talk about snapshots and how Stack resolves dependencies.

Snapshots

Stack follows a rule that says, for any project, there is precisely one version of each package available. Obviously, for many packages there are many versions available in the world. But when resolving a stack.yaml file, Stack requires that you have chosen a specific version for each package available.

The most common means by which this set of packages is defined is via a snapshot provided by Stackage. For example, if you go to the page https://www.stackage.org/lts-24.24, you will see a list of 3,415 packages at specific version numbers. When you then specify snapshot: lts-24.24 you are telling Stack to use those package versions in resolving dependencies down to specific versions of packages.

Sometimes a snapshot does not have all of the packages that you want. Or you want a different version of a package. Or you want to work on a local modification of a package. In all of those cases, you can add more configuration data to your stack.yaml file to override the values it received from your snapshot setting. At the end of the day, each of your projects will end up with some way of resolving a package name into a specific version of that package.

Why specify dependencies twice?

The package acme-missiles is not included in any Stackage snapshots. When you add something like this to your stack.yaml file:

extra-deps:
- acme-missiles-0.3

what you are saying to Stack is: "if at any point you find that you need to build the acme-missiles package, please use version 0.3". You are not saying "please build acme-missiles now." You are also not saying "my package depends on acme-missiles." You are simply making it available should the need arise.

When you add to your package.yaml file:

dependencies:
- acme-missiles

or, alternatively, you add directly to your Cabal file:

build-depends: acme-missiles

you are saying "this package requires that acme-missiles be available." Since acme-missiles does not appear in your snapshot, without also modifying your stack.yaml to mention it via extra-deps, Stack will complain about the dependency being unavailable.

You may challenge: but why go through all of that annoyance? Stack knows what package I want, why not just go grab it? The answer is that, if Stack just grabbed acme-missiles for you without it being specified in the stack.yaml somehow, you'd lose reproducibility. How would Stack know which version to use? It may elect to use the newest version, but if a new version is available in the future, will it automatically switch to that?

Stack's core philosophy is that build plans are always reproducible. The purpose of the stack.yaml file is to define an immutable set of packages. No matter when in time you use it, and no matter how many new release happen in the interim, the build plan generated should be the same.

(There is, however, at least one hole in this theory today, which is Hackage revisions. When you specify extra-deps: [acme-missiles-0.3], it does not specify which revision of the Cabal file to use, and Stack will just choose the latest. Stack has the ability to specify exact revisions of Cabal files, but this is not enforced as a requirement, because it is so different from the way most people work with packages.)

And now, how about the other side: why does Stack not automatically add acme-missiles to build-depends in your Cabal file if you add it as an extra-dep? There are a surprising number reasons for this:

• The Cabal specification does not support anything like that
• There can be multiple packages in a project, and how do we know which package actually needs the dependency?
• There can be multiple components (libraries, executable, etc) in a package, and how do we know which of those actually needs the dependency?
• The dependency may only be conditionally needed, based on flags, operating system, or architecture. As an extreme example, we would not want a Linux-only package to be built by force on Windows.

While for simple use cases it seems like automatically adding dependencies from the Cabal file to the stack.yaml file or vice-versa would be a good thing, it breaks down immediately for any semi-difficult case. Therefore, Stack requires you to add it to both places.

And a final note, in case it was not clear. The example above used acme-missiles, which is not in Stackage snapshots. If, however, you want to depend on a package already present in the snapshot you have selected, there is no need to add it explicitly to your stack.yaml file: it is already there implicitly via the snapshot setting. This is what you do the majority of the time, such as when you add vector or mtl as a build-depends value.

Should I check-in automatically generated Cabal files?

Yes, you should. This recommendation was changed in issue #5210. Please see the discussion there.