From 6455bea52ba262cea8d1c706ec8a2eec192ec643 Mon Sep 17 00:00:00 2001 From: Anna Henningsen Date: Fri, 31 Aug 2018 16:52:44 +0200 Subject: [PATCH] src: warn about odd UTF-16 decoding function signature Using a `uint16_t` sequence for UTF-16 processing would typically imply that the sequence already contains the correct 16-bit code units. However, our API does not account for that. The previous comments were somewhat misleading, since endianness is typically applied to sequences of bytes, which is not something that this API works with. PR-URL: https://github.com/nodejs/node/pull/22623 Reviewed-By: Colin Ihrig --- src/node.h | 3 ++- src/string_bytes.h | 5 ++++- 2 files changed, 6 insertions(+), 2 deletions(-) diff --git a/src/node.h b/src/node.h index 8433eed3745..1154c397650 100644 --- a/src/node.h +++ b/src/node.h @@ -416,7 +416,8 @@ NODE_EXTERN v8::Local Encode(v8::Isolate* isolate, size_t len, enum encoding encoding = LATIN1); -// The input buffer should be in host endianness. +// Warning: This reverses endianness on Big Endian platforms, even though the +// signature using uint16_t implies that it should not. NODE_EXTERN v8::Local Encode(v8::Isolate* isolate, const uint16_t* buf, size_t len); diff --git a/src/string_bytes.h b/src/string_bytes.h index 8280e379877..9e64fa0ee5f 100644 --- a/src/string_bytes.h +++ b/src/string_bytes.h @@ -97,7 +97,10 @@ class StringBytes { enum encoding encoding, v8::Local* error); - // The input buffer should be in host endianness. + // Warning: This reverses endianness on BE platforms, even though the + // signature using uint16_t implies that it should not. + // However, the brokenness is already public API and can't therefore + // be changed easily. static v8::MaybeLocal Encode(v8::Isolate* isolate, const uint16_t* buf, size_t buflen,