TextDecoder behavior is wrong since 20.18.3/22.13.0

Testcase:

Array.from({length:256},(x,i)=>[i,new TextDecoder('windows-1252').decode(Uint8Array.of(i)).codePointAt(0)]).filter(([a,b])=>a!==b).map(x=>x.join())

Node.js treats windows-1252 as a subset of Unicode (code above shows zero difference), which is not correct

E.g. Node.js:

> new TextDecoder('windows-1252').decode(Uint8Array.of(128)).codePointAt(0)
128
> new TextDecoder('windows-1252').decode(Uint8Array.of(130)).codePointAt(0)
130
> new TextDecoder('windows-1252').decode(Uint8Array.of(131)).codePointAt(0)
131
> new TextDecoder('windows-1252').decode(Uint8Array.of(159)).codePointAt(0)
159

Browsers (expected):

> new TextDecoder('windows-1252').decode(Uint8Array.of(128)).codePointAt(0)
8364
> new TextDecoder('windows-1252').decode(Uint8Array.of(130)).codePointAt(0)
8218
> new TextDecoder('windows-1252').decode(Uint8Array.of(131)).codePointAt(0)
402
> new TextDecoder('windows-1252').decode(Uint8Array.of(159)).codePointAt(0)
376

This also directly contradicts the doc (which is aware that windows-1252 and Latin1 are different):

node/doc/api/buffer.md

Lines 229 to 234 in 7643c2a

    
           Modern Web browsers follow the [WHATWG Encoding Standard][] which aliases 
        
           both `'latin1'` and `'ISO-8859-1'` to `'win-1252'`. This means that while doing 
        
           something like `http.get()`, if the returned charset is one of those listed in 
        
           the WHATWG specification it is possible that the server actually returned 
        
           `'win-1252'`-encoded data, and using `'latin1'` encoding may incorrectly decode 
        
           the characters.

It's also a regression since v20.18.3 and v22.13.0
Node.js <=20.18.2 behaves correctly, v22 <=22.12.0 also behaves correctly

This regressed in 20.x and 22.x this year, after they were labeled as LTS
20.x regressed during Maintenance.

Whatever caused this in 20/22 should be reverted

#59515, #56542 seems to be the same issue

#55275 is the likely culprit

Thanks for the repair & infos , I thought latin 1 and windows-1252 were the same

I posted an explainer at #55275 (comment)

src: implement Windows-1252 encoding support and update related tests

Navigation Menu

Repository navigation

TextDecoder behavior is wrong since 20.18.3/22.13.0 #60888

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

	Modern Web browsers follow the [WHATWG Encoding Standard][] which aliases
	both `'latin1'` and `'ISO-8859-1'` to `'win-1252'`. This means that while doing
	something like `http.get()`, if the returned charset is one of those listed in
	the WHATWG specification it is possible that the server actually returned
	`'win-1252'`-encoded data, and using `'latin1'` encoding may incorrectly decode
	the characters.

TextDecoder behavior is wrong since 20.18.3/22.13.0 #60888

Description

Activity

ChALkeR commented on Nov 29, 2025

ChALkeR commented on Nov 29, 2025

mertcanaltin commented on Nov 29, 2025

ChALkeR commented on Nov 29, 2025

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions