Another example. TRON OS uses its own character encoding.
I don't discuss whether it is good or bad in general, but
I can understand there are some specific fields (e.g. historic
literatures) that need them (e.g.
http://www.um.u-tokyo.ac.jp/dm2k-umdb/publish_db/books/dm2000/english/01/01-07.html )
Then it might be natural for a Scheme implementation on TRON
OS using OS's native encoding for strings. I see no reason
to enforce internal encodings to be, say, utf-16, in such case,
as far as it provides a compatible String API defined in the
report, so that it can use portable R6RS libraries. Of course
on such implementation, char->integer becomes lossy (several
different native characters can be mapped to the same unicode
codepoint), and string-ref can have some overhead, but that's
the implementation's problem---why should the standard care?
--shiro
Received on Mon Mar 26 2007 - 17:07:52 UTC