It is recommended to use AbstractFilePath when possible, because it is more correct.
For each variant there are three main modules:
System.FilePath.Posix / System.AbstractFilePath.Posix manipulates POSIX/Linux style FilePath values (with / as the path separator).
System.FilePath.Windows / System.AbstractFilePath.Windows manipulates Windows style FilePath values (with either \ or / as the path separator, and deals with drives).
System.FilePath / System.AbstractFilePath for dealing with current platform-specific filepaths
All three modules provide the same API, and the same documentation (calling out differences in the different variants).
System.OsString is like System.AbstractFilePath, but more general purpose. Refer to the documentation of
those modules for more information.
What is a FilePath?
In Haskell, the legacy definition (used in base and Prelude) is type FilePath = String,
where a Haskell String is a list of Unicode code points.
The new definition is (simplified) newtype AbstractFilePath = AFP ShortByteString, where
ShortByteString is an unpinned byte array and follows syscall conventions, preserving the encoding.
On unix, filenames don't have a predefined encoding as per the
POSIX specification
and are passed as char[] to syscalls.
On windows (at least the API used by Win32) filepaths are UTF-16LE strings.
You are encouraged to use AbstractFilePath whenever possible, because it is more correct.
Also note that this is a low-level library and it makes no attempt at providing a more
type safe variant for filepaths (e.g. by distinguishing between absolute and relative
paths) and ensures no invariants (such as filepath validity).
请发表评论