On Nov 2, 2004, at 10:46 AM, Dan Sugalski wrote:
At 1:42 PM -0500 11/2/04, Sam Ruby wrote:
I don't care if Parrot uses ICU on any platform.
I do care that Parrot supports utf-8 on every platform.
Ah, OK. Yes, we will support all the unicode encodings, as well as the unicode character set, on all platforms.
I'll point out, for the record, that it make no sense to have strings holding binary data--the observation that fundamental string API make no sense for such strings is an indication that this is an attempt to use the wrong data type for the wrong purpose. "Just thrown an exception" is not a magic cure--it's an indication that things are being modeled incorrectly. It shouldn't be necessary.
In terms of ICU specifically, I don't much care what underlying library we use. That said, if we want to "support all the unicode encodings", we're going to have to provide all of the relevant functionality that ICU gives us. Trying to re-implement this, separately, from scratch, is a mistake--a waste of resources. A better approach (for Parrot, and for the computing community in general) would be to put our efforts toward getting ICU working on the platforms on which it is having problems. Or start with some other well-established library--but ICU seems to be the best around.
JEff
