stack sdist does not handle filenames with non-ASCII characters

[snoyberg]

Steps to reproduce

1. git clone https://github.com/snoyberg/yaml
2. cd yaml
3. git checkout yaml/0.8.18.2
4. stack sdist
5. Unpack the tarball printed to standard out, e.g. tar xfv .stack-work/dist/x86_64-osx/Cabal-1.22.5.0/yaml-0.8.18.3.tar.gz

Expected

All files unpack correctly.

Actual

I get the following error from tar due to an incorrectly generated tarball:

yaml-0.8.18.3/test/resources/accent?/foo.yaml: Can't create 'yaml-0.8.18.3/test/resources/accent?/foo.yaml'
tar: Error exit delayed from previous errors.

Stack version

$ stack --version
Version 1.1.2, Git revision cebe10e845fed4420b6224d97dcabf20477bbd4b (3646 commits) x86_64 hpack-0.14.0

Method of installation

• Executable distributed with the Haskell Platform Minimal (no joke)

Related to: snoyberg/yaml#92

[snoyberg]

Possibly helpful snippet of code just using tar that shows some character corruption going on. I'm not sure of the right way to use the tar package for non-ASCII characters:

#!/usr/bin/env stack
-- stack runghc --resolver lts-6.14 --package tar
{-# LANGUAGE OverloadedStrings #-}
import Codec.Archive.Tar
import Codec.Archive.Tar.Entry
import qualified Data.ByteString.Lazy as L
import System.Process
import System.Exit

main :: IO ()
main = do
    let fp = "accenté.txt"
        Right tarPath = toTarPath False fp

    writeFile fp "test1"
    L.writeFile "tarintl.tar" $ write [fileEntry tarPath "test2"]
    rawSystem "tar" ["xfv", "tarintl.tar"] >>= exitWith

[snoyberg]

Hmm, in the Cabal source code I see the following comment, which may explain why it works:

The 'createArchive' action uses the external @tar@ program...

[snoyberg]

And an ugly workaround is discovered:

#!/usr/bin/env stack
-- stack runghc --resolver lts-6.14 --package tar --package text
{-# LANGUAGE OverloadedStrings #-}
import Codec.Archive.Tar
import Codec.Archive.Tar.Entry
import qualified Data.ByteString.Lazy as L
import System.Process
import System.Exit
import qualified Data.ByteString.Char8 as S8
import qualified Data.Text as T
import qualified Data.Text.Encoding as T

main :: IO ()
main = do
  let fp = "accenté.txt"
      Right tarPath = toTarPath False $ S8.unpack $ T.encodeUtf8 $ T.pack fp
  writeFile fp "test1"
  L.writeFile "tarintl.tar" $ write [fileEntry tarPath "test2"]
  rawSystem "tar" ["xfv", "tarintl.tar"] >>= exitWith