Normalization does not add / to path.

[fisx]

{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE QuasiQuotes #-}

import Control.Lens ((.~))
import Imports
import URI.ByteString
import URI.ByteString.QQ (uri)

main :: IO ()
main = do
  let want, have :: ByteString
      want = "http://example.com/1"
      have = normalizeURIRef' aggressiveNormalization ([uri|http://example.com/|] & pathL .~ "1")
  print want
  print have
  print $ want == have

Result:

*Main> :main
"http://example.com/1"
"http://example.com1"
False

If it's ok to remove the / from the domain, shouldn't it be ok to add it again in the path?

[MichaelXavier]

I think the problem might be how you're setting the path. The leading / is part of the path. If you show [uri|http://example.com/|] you'll see that the / is part of the path, not the domain. Maybe I'm misunderstanding the issue.

[fisx]

You're right, I was confused about domain vs. entire URL. But I'm still wondering if the path shouldn't have an implicit leading "/"?

I'm very uncertain now about what I think the library should do. If you don't see an issue, you should probably close this. If you think it's ok to make sure paths always are rendered with a leading "/" let me know, and I'll try to get together a PR.

Thanks for your help!