[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Changes to

From: Mark Wielaard
Subject: Changes to
Date: 19 Jan 2003 23:27:51 +0100


I changed the ObjectIn/OutputStream so they use less special purpose
native code. The libgcj versions of these classes had some jave
implementations of a couple of methods that are now also used in the
Classpath version. Note that I had to fix some bugs in the Kissme
reflection code the get this working correctly so other VM
implementations might want to check that their reflection code (most
importantly java.lang.reflect.Field) implementation works correctly.

Below is the status of the Mauve tests with Classpath+Kissme.

The classes have not yet been completely merged with libgcj. But these
changes will make this easier. It will also make it easier to add the
bug fixes that Jeroen Frijters submitted a while ago.



P.S. Don't forget to do the magic auto* dance after updating CVS
(see the HACKING file).

-- Extra Mauve Kissme+Classpath info --

The reason I am hacking on this part of Kissme is because I am changing
the GNU Classpath implementation to use
less special purpose native code and just use reflection. It all seems
to work now and the serialization related Mauve tests now all give the
same results. The remaining failures (which were also there with the old
implementation) are:

FAIL: (number 2)
- Bug in Classpath, Jeroen Frijters already supplied a partial patch for

FAIL:$Extern (number 1)
- I have to look into this one.

FAIL:$NotSerial (number 1)
FAIL:$BadField (number 1)
- These are most probably Mauve bugs

FAIL: uncaught
exception at  number 1
- Bug in Kissme Class.getMethod. java.lang.reflect.Proxy tries to do
Object.class.getMethod("hashCode", null) which should always succeed but
which throws NoSuchMethodException under Kissme.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]