From: Godmar Back (gback@cs.utah.edu)
Date: Fri Dec 18 1998 - 16:47:13 EST
Sure Archie, go ahead. string.c is all yours.
Basically, the underlying premise would be that both utf8 and strings
are immutable, correct?
I would like you to be able to pass a test like Edouard's (see his
mail from Sep 1):
I have write a String intern hog like:
long nr = 0;
while(true)
(Long.toString(nr++)).intern();
and run it with:
jdk-1.1.5: throw OutOfMemoryException
jre-1.1.6: panic string table overflow
jre-1.2beta4: work as a charme: they may be use WearReference from
the new package java.lang.ref :-(
Actually, come to think of it: java.lang.String and Utf8 are two
different pairs of shoes. One uses (java)char[] arrays, the other
zero-terminated (c)char* strings.
So, I think you'll need two tables (which is what string.c actually did
with utf8hash.) Ideally, we would make it two different files too.
- Godmar
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:23 EDT