Text 10695, 57 rader
Skriven 2007-10-29 16:21:08 av Aleksej R. Serdyukov (2:5020/1042.42)
  Kommentar till text 10694 av Maurice Kinal (1:140/13.1)
Ärende: Text replacement in file
================================
28 Oct 07 17:55:58, Maurice Kinal -> Aleksej R. Serdyukov:

ARS>> Though Windows NT uses a broken UTF-16

MK> Does that include win2k?

I meant it so.

ARS>> so only 65536 codepoints can be used

Or maybe not even 65536...

MK> For what exactly and what are they?  If one was aware then it could be
MK> converted to something more usable and universal.

http://www.unicode.org/

ISO/IEC 10646-1:1993 (?) Ä UCS-2, 16-bit characters, 65536 values/codepoints.

ISO/IEC 10646-1 Ä UTF-16, UCS-2 with surrogate pair support for the
representation of codepoints beyond 65536.


OK, so nothing is broken...

Windows NT prior to 5.0 (2000) only supported UCS-2. Now it uses UTF-16.

Nothing to see there, move along. Except that, though you can hope a string
with surrogates doesn't get corrupted, at least Mozilla Firefox does not
support it. I hope Firefox is an exception (and they were considering switching
to UTF-8, IIRC)...

ARS>> I cannot find the charset out, though. :-( But I hope they
ARS>> didn't undergo more than 2 conversions.

MK> Just a sec ... check http://billposer.org/Software/a2b.html as I seem
MK> to recall a posted method of creating a utf-8 table using his apps.  I
MK> have used it to convert binary integers and characters to ascii and
MK> that works great.  Haven't done anything with utf-8 yet as I don't
MK> have a need to.  I assume it will work.

Interesting. Probably it can be used with something like Unicode Utilities
<http://billposer.org/Software/unidesc.html> to determine the correct decoding
steps.
OTOH, that probably has already been done and is available somewhere. And
doesn't really need a2b.

List the possible encodings.
Give the program a string and the number of combinations, grep -i the output
with a substring like "quot".
See if you've generated any real quotation marks.


Good luck.
Help testing Firefox! See <http://quality.mozilla.org/events>.
--- GoldED+/LNX 1.1.5-b20060327 (Linux 2.6.22-2-amd64 AMD_K8_M12)
 * Origin: Delete Software Limited, Moscow, Russia. (2:5020/1042.42)