I've hacked together some code to read the content from a sequence of files:
let fileContents =
[ "filename1"; "filename2"; "etc" ]
|> Seq.map(fun file -> async {
use fs = new FileStream(file, FileMode.Open, FileAccess.Read, FileShare.Read)
let data = Array.create(int fs.Length) 0uy
let! _ = fs.AsyncRead(data, 0, data.Length)
return data
})
|> Async.Parallel
|> Async.RunSynchronously
|> Seq.map System.Text.Encoding.ASCII.GetString
It works, but I'm not really sure about it. For one thing, is async
actually giving me any benefit here? Also inside the async
function I need to wait for AsyncRead
to finish, but I don't need the returned value from it. Is there a better way of doing this than let! _ = ...
?
1 Answer 1
You can take advantage of async to do parallel IO which Async.Parallel makes very easy. Also, when you don't need the return value of an async call you can use Async.ignore in combination with do!: do! fs.AsyncRead(data, 0, data.Length) |> Async.ignore
However this specific snipper can be made much simpler since there is no reason to await the result of the computations when using Async.Parallel:
open System.IO
let fileContents =
[ "filename1"; "filename2"; "etc" ]
|> List.map(fun file -> async { use fs = File.OpenRead(file) in return! fs.AsyncRead(fs.Length |> int)})
|> Async.Parallel
|> Async.RunSynchronously
|> Seq.map System.Text.Encoding.ASCII.GetString
One caveat - if the size of your file is more than 2GB it won't work due to long->int conversion - but that would be a crazy big file to just read the content of like that :-D
-
\$\begingroup\$ Cool tips. I was trying
|> ignore instead of |> Async.Ignore
which doesn't compile. Can you tell me whatin
is doing inlet fs = File.OpenRead(file) in fs.AsyncRead(fs.Length |> int)
and how the let binding is also returning a value please? \$\endgroup\$NickL– NickL2014年02月25日 21:46:27 +00:00Commented Feb 25, 2014 at 21:46 -
\$\begingroup\$ Hi NickL, the
in
keyword is a way to bind a value as part of the next expression and is something f# inherited from OCaml. It's similar to doing a line-break after the let expression, but the value (in this casefs
) is only available in the next expression. The let binding isn't returning a value but just binding the expression File.OpenRead(file) to it. Primary reason you would usein
in f# is to skip a line-break and thus write a one-liner or if you already had a binding of the same name that you wanted toshadow
only in the next expression on the same line. Hope this makes sense. \$\endgroup\$Simon Stender Boisen– Simon Stender Boisen2014年02月26日 08:34:09 +00:00Commented Feb 26, 2014 at 8:34 -
\$\begingroup\$ @NickL Btw, notice that I changed my response since my previous one actually was problematic since I never closed the stream. So now I use the async computation expression in order to be able to await the async read so that the
use
-expression will work and dispose the file-stream. It's still running it in parallel though. \$\endgroup\$Simon Stender Boisen– Simon Stender Boisen2014年02月26日 08:47:30 +00:00Commented Feb 26, 2014 at 8:47