DecompressionStream should support pull-based decompression

[mstange]

DecompressionStream should support pull-based decompression

[mstange]

Ah, that means I misunderstood how it works and need to do some more debugging. Thanks for taking a look!

[saschanaz]

As specced, when the compressed data comes from the network, the browser has to decompress it all and enqueue the decompressed chunks.

Can you show me how you are using the stream? TransformStreams are not supposed to consume greedily and has highWaterMark=1, meaning it's only allowed to have one single unconsumed chunk and then stops processing until it's read.

With the following demo it only reads the first chunk and wait.

// A string "HELLO"
let buffer = [
  new Uint8Array([31,139,8,0,0]),
  new Uint8Array([0,0,0,0,10]),
  new Uint8Array([243,112,245,241,241]),
  new Uint8Array([7,0,54,100,68,193,5,0,0,0]),
];
let index = 0;
let r = new ReadableStream({
  pull(controller) {
    if (index >= buffer.length) {
      controller.close();
      return;
    }
    console.log("pulled", index);
    controller.enqueue(buffer[index]);
    ++index;
  },
  type: "bytes"
});
let d = new DecompressionStream("gzip");
let pipe = r.pipeThrough(d);

Please correct me if I misunderstood your issue.