blob: 8ad4445e560c216d1db1d4f90c087d333dc9ba07 [file] [log] [blame]
KNOWN ISSUES
------------
- Unit tests for the beans package may fail in the Harmony self-hosting
environment due to the incomplete functionality of stub classes in awt
and applet packages. The failing tests are:
java.beans.EventHandlerTest
java.beans.SimpleBeanInfoTest
- The find() and findBack() methods in regex nodes may work incorrectly with back references.
The issue may show up for such patterns as below:
.*(.)\\1
(.)\\1
- The Node.match() method is not completely optimized in regex. It may call the CharSequence.toString() method
even if CharacterSequence is not a string, which leads to a String creation during each call to
this node’s match() method.
- Methods implementing bitwise operations (and, or, and xor) in the BigInteger class might seem complicated
although they work rather efficiently. These methods might be rewritten in a more comprehensive form.
TODO
----
- Implement canonical equivalence support in regex.
- Simplify multistep operations in the BigInteger class, which create
multiple BigInteger objects on each step. The methods gcd (), isProbablePrime(),
modInverse(), and modPow()can be speeded up with implementing safe in-place computations.
- Enhance the multistep BigInteger.pow() method. To reduce the number of created
objects, the method can be implemented by using the square-and-multiply algorithm.
However, the proposed implementation works longer due to longer multiplication of large numbers.
The enhancement may consist of implementing safe in-place multiplication
and possible combination of the two algorithms.
- Implement special cases for more efficient exponentiation of BigInteger.TEN and
multiplication by using BigInteger.TEN. These mechanisms are widely used in the BigDecimal implementation.
- Optimize the BigInteger.toString() method.