[Webkit-unassigned] [Bug 254021] New: Using CompressionStream to compress large buffers produces invalid data

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Mar 16 07:17:08 PDT 2023


            Bug ID: 254021
           Summary: Using CompressionStream to compress large buffers
                    produces invalid data
           Product: WebKit
           Version: Safari Technology Preview
          Hardware: Mac (Apple Silicon)
                OS: macOS 12
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: WebCore Misc.
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: bzugmeyer at gmail.com

I am testing the new `CompressionStream` API in Safari Technology Preview. When trying to compress a moderately large buffer, the data it produces seems invalid.

To reproduce:

  Compress a large enough buffer of non-repeating data. In the following example, I use a buffer of approximately 48kB. Functions are coming from the WebKit CompressionStream test suite[1]:

  const input = new TextEncoder().encode(
    JSON.stringify(Array.from({ length: 10_000 }, (_, i) => i))
  const output = await compressArrayBuffer(input, "deflate");
  assert_array_equals(input, pako.inflate(output));


  `pako.inflate` returns a buffer equal to the input buffer.


  The decompression fails with `pako.inflate` returning `undefined`.


  With this particular example, I noticed that it works correctly when compressing up to 35_578 bytes. The issue only occurs when I try to compress more bytes (>= 35_579).

  Please see this page[2] for a practical way to reproduce the issue. In Chrome, every test succeed, but in Safari tests with larger inputs are failing.

[1]: https://github.com/WebKit/WebKit/blob/20329b62061b40d5a423a1d75b67779945b84729/LayoutTests/imported/w3c/web-platform-tests/compression/compression-stream.tentative.any.js
[2]: https://safari-compressionstream-issue.benoitzugmeyer.repl.co

You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20230316/a4f7d9f8/attachment.htm>

More information about the webkit-unassigned mailing list