The mappings for following MS Windows code pages are changed
in the default distribution between ICU 2.2 and 2.6:
1255 (Hebrew), 1256 (Arabic) and 1257 (Baltic).
It looks like the old ones were "better" (in sync with
MS's code page definitions at
http://www.microsoft.com/globaldev/reference/sbcs/1255.htm
http://www.microsoft.com/globaldev/reference/sbcs/1256.htm
http://www.microsoft.com/globaldev/reference/sbcs/1257.htm).
Was this changed by accident or on purpose?
Details:
MS Windows CP 1255:
Identifier "windows-1255" is mapped to converter "ibm-5351"
in ICU 2.2 and to "ibm-5351_P100-1998" in ICU 2.6.
Code points xA1, xAA, xB8, xBA, xBF, xD7 and xD8 are not
defined any more, whereas the "old" converter contained
appropriate mappings.
MS Windows CP 1256:
Identifier "windows-1256" is mapped to converter "ibm-5352"
in ICU 2.2 and to "ibm-5352_P100-1998" in ICU 2.6.
Code points xAA, xC0 and xFF are not defined any more, whereas
the "old" converter contained appropriate mappings.
MS Windows CP 1257:
Identifier "windows-1257" is mapped to converter "ibm-5353"
in ICU 2.2 and to "ibm-5353_P100-1998" in ICU 2.6.
Code points xB4 and xFF are not defined any more, whereas
the "old" converter contained appropriate mappings.
All the best,
Paul
------------------------------------------------------------
Paul Langer E-mail: paul....@softwareag.com
Software AG Phone: +49-6151-92-1912
Uhlandstraße 12 Fax: +49-6151-92-1613
64297 Darmstadt
Germany