Skip to content

Latest commit

 

History

History
57 lines (37 loc) · 2.74 KB

HACKING.md

File metadata and controls

57 lines (37 loc) · 2.74 KB
Raw

Hacking on the Reflex Platform

To work on a particular package, use the work-on script instead of ./try-reflex:

~/reflex-platform/scripts/work-on ghcjs ./your-project

This will use your package's cabal file to determine dependencies. If you have a default.nix, it will use that instead. Note that your project's path must include at least one slash ('/') so that work-on can detect that it is a path, rather than a package name.

This will give you the exact environment needed to work with the given package and platform, rather than the general-purpose environment provided by the Reflex Platform.

You can replace ghcjs with ghc to hack on the native GHC version of the package (including with GHCi if you want). You can also use a package name instead of a path, which will drop you into the standard build environment of that package; this works even if you don't yet have the source for that package.

Instead of specifying ghcjs or ghc for the platform, you can also specify a path to a nix expression file whose value represents a Haskell environment, for example:

{ reflex-platform, ... }: reflex-platform.ghcjs.override {
  overrides = self: super: {
    some-package = self.callPackage ./some-package {};
  };
}

Finally, you can specify a parenthesized expression as your platform, which will be interpreted as a raw nix expression. For example, (import ./my-platform.nix {}).

Once inside the shell, you'll have access to the environment in which the package is built with a few more tools (such as cabal-install) added.

Checking out a sub-repo

The hack-on script is provided for conveniently checking out a sub-repository.

./scripts/hack-on reflex

This will check out the same version of reflex currently being used by the Reflex Platform. Note that reflex, here, is a path relative to the current directory, so you must be in this folder when you execute this.

Once the repository is checked out, you can make modifications to it, which will be used the next time you enter a try-reflex or work-on shell. Existing sessions will not be affected—exit and re-enter your shell to use the changes.

When you have completed some work hacking on the sub-repository, you can use the hack-add script to check your changes into this repository without needing to delete the repository, like so:

./scripts/hack-add reflex

You can then commit and push reflex-platform without needing to delete the sub-repository.

When you are completely done with a sub-repository, you can remove it using hack-off:

./scripts/hack-off reflex

This will remove the repository and replace the default.nix and git.json files that were pointing to it before running hack-on.