TextDecoder regressed for ascii, windows-1252, ... encodings in v23.4.0

[lutzroeder]

Version

23.4.0

Platform

macOS 15.1.1 (24B91)

Subsystem

No response

What steps will reproduce the bug?

~: node
Welcome to Node.js v23.4.0.
> b = new Uint8Array([97, 108, 101])
> decoder = new TextDecoder('ascii')
> decoder.decode(b)
<Buffer 61 6c 65>

How often does it reproduce? Is there a required condition?

Reproduces reliably.

What is the expected behavior? Why is that the expected behavior?

Returns a string instead of a buffer. See output from v23.3.0:

~: node
Welcome to Node.js v23.3.0.
> b = new Uint8Array([97, 108, 101])
> decoder = new TextDecoder('ascii')
> decoder.decode(b)
'ale'

What do you see instead?

Returns a buffer.

Additional information

@mertcanaltin @RafaelGSS @anonrig @jasnell @lemire @targos #55275

[lemire]

Verified.

> new TextDecoder('us-ascii').decode(new Uint8Array([0x32, 0x44]))
<Buffer 32 44>
> new TextDecoder('windows-1252').decode(new Uint8Array([0x32, 0x44]))
<Buffer 32 44>
> new TextDecoder('latin1').decode(new Uint8Array([0x32, 0x44]))
<Buffer 32 44>

[lemire]

This is related to #55275

@mertcanaltin Can you check?

I think that the C++ function returns a buffer, not a string.

[mertcanaltin]

This is related to #55275

@mertcanaltin Can you check?

I think that the C++ function returns a buffer, not a string.

hello I'm looking into this, I'll post an improvement

[mertcanaltin]

hello, I am fixed this problem after then ı testet my local enviorement, ı see improvement this problem

➜  node git:(mert/fast-path-fix) ./node
Welcome to Node.js v24.0.0-pre.
Type ".help" for more information.
>  b = new Uint8Array([97, 108, 101])
Uint8Array(3) [ 97, 108, 101 ]
> decoder = new TextDecoder('ascii')
TextDecoder { encoding: 'windows-1252', fatal: false, ignoreBOM: false }
> decoder.decode(b)
'ale'
> 

many thanks for reporting the bug and for the suggestions.

[lutzroeder]

@mertcanaltin @lemire since this is an impactful regression rolling out as part of v23.4.0, are there plans to fast follow shipping a fix in a v23.4.1 release?

[lemire]

@lutzroeder It is definitively getting fixed in the next release. As to whether this next release is labelled v23.4.1 or not... I do not know.

There is also a related issue: nodejs/performance#183

That last one is troubling as well.