From c3ffc94e6af7c64ecb9e8e9fc26b12e04b3217af Mon Sep 17 00:00:00 2001 From: =?utf8?q?Ludovic=20Court=C3=A8s?= Date: Sat, 4 Apr 2009 23:21:03 +0200 Subject: [PATCH] bytevector: Update comment regarding IEEE-754 endianness. --- src/bytevector.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/src/bytevector.c b/src/bytevector.c index 7a00b07..930ce16 100644 --- a/src/bytevector.c +++ b/src/bytevector.c @@ -1404,11 +1404,12 @@ SCM_DEFINE (scm_r6rs_bytevector_s64_native_set_x, "bytevector-s64-native-set!", /* Operations on IEEE-754 numbers. */ -/* XXX: There are not only two encodings (big and little endian), as implied - by the API, but rather three (in the case of little endian, there are two - possible word endians, as visible in glibc's ). When the - endianness is `little', we assume little endian for both the byte order - and the word order. */ +/* There are two possible word endians, visible in glibc's . + However, in R6RS, when the endianness is `little', little endian is + assumed for both the byte order and the word order. This is clear from + Section 2.1 of R6RS-lib (in response to + http://www.r6rs.org/formal-comments/comment-187.txt). */ + /* Convert to/from a floating-point number with different endianness. This method is probably not the most efficient but it should be portable. */ -- 2.11.4.GIT