Edit

Console encoding doesn't remain UTF-8 after completion

  • Article

The bug mentioned in the CLI console output uses UTF-8 breaking change, where the .NET SDK changed the encoding of the entire console, has been fixed. The console encoding no longer remains UTF-8 after the .NET SDK executes a command. It's possible that users came to rely on that behavior, hence this is a breaking change.

In addition, the .NET SDK no longer changes the encoding to UTF-8 on older Windows 10 versions that don't fully support it.

Previous behavior

  • The SDK changed the encoding of a terminal after running a command such as dotnet build.
  • The SDK used the UTF-8 encoding to correctly render non-English characters, even on versions of Windows 10 that did not officially support UTF-8. The behavior was undefined on those versions.

New behavior

  • The SDK doesn't change the terminal encoding after exit for other programs.
  • By default, the SDK no longer uses UTF-8 for Windows versions that don't support it.

Version introduced

7.0.3xx .NET 8 Preview 3

Type of breaking change

This change can affect binary compatibility. It's also a behavioral change.

Reason for change

There was an existing bug where the .NET SDK affected the encoding on the console for other programs. That was a bug that was fixed, resulting in this breaking change.

Older versions of Windows 10 (that is, versions before the November 2019 update) didn't support UTF-8, so the default behavior shouldn't be to use UTF-8 encoding. Instead, an opt-in is now available.

If your app needs to change the code page on Windows, it can run a process to invoke the chcp command. Your app shouldn't rely on the .NET SDK to change the encoding.

For older Windows 10 versions that don't officially support UTF-8 where you want the .NET SDK to continue to change the encoding to UTF-8 for non-English languages, can you set the environment variable DOTNET_CLI_FORCE_UTF8_ENCODING to true or 1.

See also