- From: Martin J. Duerst <duerst@w3.org>
- Date: Mon, 08 Feb 1999 22:12:31 +0900
- To: Paul Hoffman / IMC <phoffman@imc.org>
- Cc: ietf-charsets@iana.org
Hello Paul, I have earlier suggested that in the ID for the UTF-16 registration, we include some code segment that shows how to write out 16-bit entities as big-endian or little-endian irrespective of the architecture of the machine where the code is run. You have been somewhat sceptical. I have now discussed this idea with a colleague who is an expert in this field, and I have prepared a tiny "test suite", which I have run both on a Sun (big-endian) and on a friend's FreeBSD system (little-endian). I invite you and everybody else to download this: ftp://ftp.ifi.unizh.ch/outgoing/endian.tar Once untared, the README file explains what to do. If the diffs don't find any differences, the code works. On some machines, the code may need some slight adaption (newline conventions in the source code, use of short for 16-bit units, availability and use of fopen/fclose/putc), but I would be extremely surprised if anybody found any machine and C compiler where otherwise the code doesn't perform as it should. Once we have tested this code to your liking, we can take the relevant few lines and include it in the ID. Regards, Martin. #-#-# Martin J. Du"rst, World Wide Web Consortium #-#-# mailto:duerst@w3.org http://www.w3.org
Received on Thursday, 11 February 1999 03:20:47 UTC