Buffer#toString should throw error on unsupported encoding #2417
Description
First off, thanks for the amazing work on duktape!
The current behavior of Buffer#toString is always to encode the string as UTF8, even if the encoding is something else (like "binary" for binary strings). Missing encoding issues have been discussed in other issues.
By default, NodeJS throws an error when the encoding is unknown. This behavior has been true since the 0.8.x series in 2014. Testing with new Buffer([1,2,3]).toString("wtf"):
| Node Version | Behavior |
|---|---|
0.8.28 |
Error: Unknown encoding |
0.10.48 |
TypeError: Unknown encoding: wtf |
0.12.18 |
TypeError: Unknown encoding: wtf |
4.9.1 |
TypeError: Unknown encoding: wtf |
6.17.1 |
TypeError: Unknown encoding: wtf |
8.17.0 |
TypeError: Unknown encoding: wtf |
10.24.1 |
TypeError [ERR_UNKNOWN_ENCODING]: Unknown encoding: wtf |
12.22.6 |
TypeError [ERR_UNKNOWN_ENCODING]: Unknown encoding: wtf |
14.17.6 |
TypeError [ERR_UNKNOWN_ENCODING]: Unknown encoding: wtf |
16.8.0 |
TypeError [ERR_UNKNOWN_ENCODING]: Unknown encoding: wtf |
It is straightforward to test for unexpected behavior downstream, but it would be consistent with NodeJS to throw an error for unsupported encodings. If this change is acceptable we can send a PR