I couldn't help but notice that you were talking about ICU on this mailing list. Let me interject with some suggestions.
I should mention that ICU 2.6 can build a static data library. I recommend that ICU be built without the --with-data-packaging=archive configure option. You will probably have fewer path issues if you did that, and used ICU static data library.
For those people that are interested here is some more information about building ICU data and its options: <http://oss.software.ibm.com/icu/userguide/icudata.html>.
If you are having problems and need to patch ICU, you should consider submitting the patches to ICU to our jitterbug system <http://www.jtcsv.com/cgibin/icu-bugs>. If you submit them early enough, the changes may be able to make ICU 3.0. ICU 3.0 should out in mid June. I'm sure that you don't want to keep patching ICU every time you upgrade to a new version of ICU.
ICU 3.0 should work a little better with Windows and Cygwin. ICU 3.0 should also build faster than before due to some build changes done recently. I'm sure some people on this list would be interested in that.
When the alpha and beta releases of ICU come out in a few weeks, I recommend someone from this group try building it on your machines. Some people here seem to have access to some machine configurations that are unavailable to the ICU team. Testing these pre-releases will help to verify that ICU release is as portable as possible.
I'm glad to see that perl will be improving its Unicode support :-)
George Rhoten http://oss.software.ibm.com/icu/
