Mann mit Hut

I implemented a solution [here](https://github.com/mmhat/language-bash/commit/bed366efe432c18de9ecc473ee9fb793d9edc276). Didn't test it though.

Ok, I'll do that. I'll wait until #24 gets merged.

Closing this one since an solution to the problem has been provided. Please reopen if the answer is unsatisfying.

@vdukhovni > Why not conditionally provide a new high-level Linux/glibc-specific API that returns a directory entry _and_ the file type. The conditional test here would be the availability of the...

@hasufell > Am more in favor of https://hackage.haskell.org/package/hpath-posix-0.13.3/docs/System-Posix-RawFilePath-Directory-Traversals.html#v:readDirEnt I am aware of `hpath-posix` (Great library!) and the purpose of this patch is precisely to make libraries like this one easier...

> > We're going to burn out the contributor this way. > > That's a very sensible concern, but @mmhat originally said that "If the conditional high-level API you proposed...

> Why not conditionally provide a new high-level Linux/glibc-specific API that returns a directory entry _and_ the file type. The conditional test here would be the availability of the `DT_REG`...

> Another reason to not expose the **low-level** API. I kindly disagree: IMHO, when it comes to directory tree traversal, the functions we provide in this package are _already_ low-level....

> > where the `Bool` indicates whether to use `stat` on `DT_UKNOWN` > > When would someone pass `False` here? Good point; I always thought of the `readDirStream` functions as...

@hasufell I guess I didn't explain my concerns very well; What I meant was: Suppose neither `DT_REG` nor `DT_DIR` are defined, hence `pattern RegularFileType = DirType (-1)` and `pattern DirectoryType...