Re: Versioning

[George Rhoten]

The charset data isn't really related to the version of ICU. It's an
added feature that you can add to almost any version of ICU. If you want
that charset data for a specific version of ICU, download the version of
ICU that you want. ICU already includes a large number of UCM files in
the data directory.

The XML data is generated from the UCM files, and the vast majority of the
UCM files will not work without some massaging of the header information.
The mappings are there, but the state tables may not work. The IBM tables
do have state information, but most of it is unverified (makeconv doesn't
like the header data). I am working on getting all of this data fixed by
the time ICU 2.4 is released. These extra UCM files in the charset
directory are there just in case ICU doesn't have the charset that you
want by default.

If you really want all of the UCM files, I recommend using CVS. Sometimes
the CVS server is a little flaky, but it should eventually work. I don't
see us making any zip downloads of the large charset repository any time
soon.

George Rhoten
IBM Globalization Center of Competency/ICU San Jose, CA, USA