Mixed client_encoding database: what to do with..

Hi,

Please advise.

We have a product with a PostgreSQLdatabase (UTF-8, PG 8.0 and 8.3)
It has a (legacy) client app that uses client_encoding = win1521 on the
connection since it uses win1252 internally.
The legacy app will be around for several years.
It cannot change to unicode.

There is a new client app too, written in java.
It uses client_encoding = unicode. Nice.

The problem: if someone inputs a non-win1252 character via the new java
app, the legacy app fails.
PostgreSQL returns an error: character [..] of encodig UTF-8 has no
equivalent in WIN1252.

What are my options to stop the legacy app from failing?
-Is there a setting to tell PostgreSQL to fail the character gracefully
by substuting it?
-Should i filter out all non win1252 charachters at the new java app and
substitute them? (not so nice, i'd like to have the chars in the db)
-Change the DB encoding to WIN1252 (no!)
-?any ideas?

Greetings, Joek