Taurus JavaVM Bug List
This page gives a list of the known bugs and issues (both open and fixed) in
the Taurus Java Virtual Machine.
Open Bugs
NONE
Fixed Bugs
| 000001 | Reported: 27/11/1999 | Status: Fixed in 1.10 | char type limited to 8-bits | Version: All | Description:
The char type is restricted to 8-bits in the versions
prior to 1.10, therefore Unicode characters are not supported. | | 000002 | Reported: 27/11/1999 | Status: Fixed in 1.03a | short and byte types incorrectly implemented | Version: Before 1.03a | Instructions affected:
baload, bipush, saload (NYI),
sipush, i2b, i2s
| Description:
The short and byte types behave incorrectly when
negative values are used. According to the spec, shorts and
bytes should be sign-extended to fill the word before either
being pushed onto the stack or placed in a local variable. | | 000003 | Reported: 27/11/1999 | Status: Fixed in 1.04a | float and double do not support NaN | Version: Before 1.04a | Description:
None of the float or double operations currently
support NaN. There is no difference in the implementation of
dcmpl and dcmg or fcmpl and
fcmpg. | | 000004 | Reported: 27/11/1999 | Status: Fixed in 1.03a | ddiv incorrectly implemented | Version: Before 1.03a | Description:
The ddiv instruction can never raise an exception. This was
incorrectly implemented, whereby dividing by zero would raise and exception, as
per integer division. | | 000005 | Reported: 27/11/1999 | Status: Fixed in 1.03a | idiv incorrectly implemented | Version: Before 1.03a | Description:
The idiv instruction has a special case when MIN_INT is divided by
-1, which results in overflow, the result should by MIN_INT. Early Win32
releases crashed with "The exception Integer overflow. (0xc0000095) occured
in the application at location...". | | 000006 | Reported: 17/12/1999 | Status: Fixed in 1.03a | | Multi-dimension array not initialised | Version: Before 1.03a | Description:
The multianewarray instruction is restricted to creating one-
dimensional arrays only. No warning is given when multi-dimensional arrays are
used. When a multi-dimensional array is accessed, a null-pointer exception
occurs on access to the 2nd dimension of the array (i.e. the array object does
not exist. | | 000007 | Reported: 01/03/2002 | Status: Fixed in 1.04a | Double.LongToDoubleBits() and
Double.DoubleToLongBits() mis-implemented | Version: Before 1.04a | Description:
These methods should be simple casts from one type to the other. They are use
for direct access to the bits of doubles. | | 000008 | Reported: 03/03/2002 | Status: Fixed in 1.04a | | Exception handler does not clear stack | Version: Before 1.04a | Description:
According to the VM Spec. the exception handler should clear the stack of the
method that caught the exception. The current implementation does not and we
get a stack overrun. | | 000009 | Reported: 05/03/2002 | Status: Fixed in 1.04a | ldc2_w mis-implemented | Version: Before 1.04a | Description:
ldc2_w loads double values for NaN, +Inf and -Inf onto the stack
incorrectly. | | 000010 | Reported: 07/03/2002 | Status: Fixed in 1.04a | multianewarray mis-implemented | Version: Before 1.04a | Description:
multianewarray creates incorrectly dimensioned arrays. The
dimensions are reversed from those specified. |
|
|
