Distribution.Types.CondTree

A CondTree is used to represent the conditional structure of a Cabal file, reflecting a syntax element subject to constraints, and then any number of sub-elements which may be enabled subject to some condition. Both a and c are usually Monoids.

To be more concrete, consider the following fragment of a Cabal file:

build-depends: base >= 4.0
if flag(extra)
    build-depends: base >= 4.2

One way to represent this is to have CondTree ConfVar [Dependency] BuildInfo. Here, condTreeData represents the actual fields which are not behind any conditional, while condTreeComponents recursively records any further fields which are behind a conditional. condTreeConstraints records the constraints (in this case, base >= 4.0) which would be applied if you use this syntax; in general, this is derived off of targetBuildInfo (perhaps a good refactoring would be to convert this into an opaque type, with a smart constructor that pre-computes the dependencies.)

A CondBranch represents a conditional branch, e.g., if flag(foo) on some syntax a. It also has an optional false branch.

Flatten a CondTree. This will traverse the CondTree by taking all possible paths into account, but merging inclusive when two paths may co-exist, and exclusively when the paths are an if/else

@Traversal@ for the data

@Traversal@ for the variables

@Traversal@ for the data

@Traversal@ for the variables

Extract the condition matched by the given predicate from a cond tree.

We use this mainly for extracting buildable conditions (see the Note in Distribution.PackageDescription.Configuration), but the function is in fact more general.

Flattens a CondTree using a partial flag assignment. When a condition cannot be evaluated, both branches are ignored.

Realizes a CondBranch using partial flag assignment. When a condition cannot be evaluated, returns Nothing.

Flatten a CondTree. This will resolve the CondTree by taking all possible paths into account. Note that since branches represent exclusive choices this may not result in a "sane" result.