Consider structures other than Array[Byte]

[marklister]

From #6 (@bblfish)

Perhaps there is an extra improvement you can add. Is it possible to abstract somewhat from an array? For example the JS seems to return a ArrayBuffer which I think can be turned into a java.nio.ByteBuffer using TypedArrayBufferBridge.

So it would be useful to have a method that could work with those elements and not just arrays. I am not yet quite sure what the best abstraction here would be... Certainly given that you don't need to change the source object, an Array is too precise.

There is some example code here using a ByteBuffer https://gist.github.com/mseddon/1cfcb0970272cac40497

[marklister]

@bblfish How does this look?

scala> import scala.collection.mutable.ArrayBuffer
import scala.collection.mutable.ArrayBuffer

scala> val a= ArrayBuffer[Byte]() ++ "abc".getBytes
a: scala.collection.mutable.ArrayBuffer[Byte] = ArrayBuffer(97, 98, 99)

scala> a.toBase64
res0: String = YWJj

It works on a Seq[Byte]

[bblfish]

That probably helps in many situations. But I don't know if it helps with java.nio.ByteBuffer, other than having to convert that to that to an ArrayBuffer. Is there a way to do that efficiently? ( ie, really one wants to be allocating no new memory )

We were discussing this today on the scala-js gitter.

[marklister]

The latest commit (and I published it to bintray) supports a Seq[Byte] ArrayBuffer [Byte] etc.

java.nio.ByteBuffer has an array() method. @bblfish, if I read your initial post correctly you're deriving ByteBuffer from ArrayBuffer so I don't think we need to support directly?

[bblfish]

yes, no need to support ArrayBuffer as that is a JS view on something that is very similar to java.nio.ByteBuffer. I suppose that is why there is a transformation in ScalaJS from ArrayBuffer to a java.nio.ByteBuffer since scala-js 0.6.1.

I can try the array() method on ByteBuffer though my guess is that is not the most efficient way to do things. For my purposes efficiency is not paramount at the moment. But as you start working on the efficiency of your library you'll want to use its features directly. It should not be too hard then to get something extremely efficient. perhaps there is something to learn from @mseddon 's https://gist.github.com/mseddon/1cfcb0970272cac40497

[bblfish]

This is my current code

   import com.github.marklister.base64.Base64._
   val bb = TypedArrayBuffer.wrap(sig.asInstanceOf[ArrayBuffer])
   val arraybuf: Array[Byte] = new Array[Byte](bb.remaining())
   bb.get(arraybuf)
   val hashedSig = arraybuf.toBase64

TypedArrayBuffer.wrap(sig.asInstanceOf[ArrayBuffer]) returns a java.nio.ByteBuffer .
I am not sure if there are read only ByteBuffers. If so you could write a wrapper that wraps those using the few lines of code above, and that could be used like this:

bb.toBase64

Then later when you have time you could optimise away the copy.

[marklister]

So would

sig.asInstanceOf[ArrayBuffer].toBase64

not work?

[bblfish]

Well that would make it a pure scala-js function. ArrayBuffer's full signature is scala.scalajs.js.typedarray.ArrayBuffer, and I don't think you need to write something that is scala-js specific like that. Libraries should be producing the nio classes.

[marklister]

Ah! I see! I had it confused with
http://www.scala-lang.org/api/current/index.html#scala.collection.mutable.ArrayBuffer

[bblfish]

my fault. Should have included the imports. Sorry!

[bblfish]

Got my code working! I had a bug on the server with encoding/decoding BigIntegers from hexadecimal. read-write-web/rww-play@de8655e

[marklister]

Yeah, I bumped into this same problem a few years ago. See my basen lib.