MantisBT - Issues

0000476: Would you make the javadoc more clear

ggd543 <ggd543@example.com> — Sat, 31 Jul 2010 06:33:46 +0100

I found the javadoc description is confused sometimes.
Let's see the javadoc for the method advance(Date, int, TimeUnit) of Calendar class.

The doc says that advances the given date of the given number of business days and returns the result.

But if the given date is not a business date and the given number is zero, the method will adjust the given day to the appropriate near business day with respect to BusinessConvention.Proceeding convention, just like the behavior of adjust method .

But why not return the given date directly if the given number is zero ?

0000473: cannot find symbol SobolRSG in SobolChartSample.java

nranjan <nranjan@example.com> — Sat, 26 Jun 2010 14:54:24 +0100

There was build failure in jquantlib-samples
[INFO] ------------------------------------------------------------------------
[ERROR] BUILD FAILURE
[INFO] ------------------------------------------------------------------------
[INFO] Compilation failure

\JQuantLib\jquantlib-samples\src\main\java\org\jquantlib\samples\SobolChartSample.java:[18,39] cannot find symbol
symbol : class SobolRSG
location: package org.jquantlib.math.randomnumbers

problem is that SobolChartSample refers to SobolRsg with incorrect case 'SobolRSG' [in uppercase] instead of 'SobolRsg'[lowercase].

0000475: Samples depends on 0.2.3, but actual svn version is 0.2.4

bilyushonak <bilyushonak@example.com> — Mon, 21 Jun 2010 07:35:31 +0100

./jquantlib-all/mvn clean install failed due to samples depends on 0.2.3

0000451: Remove enumeration Style from class Cells

Richard Gomes <Richard Gomes@example.com> — Sat, 22 May 2010 22:39:33 +0100

Enumeration Style was put in order to allow translation of code written in Fortran language, as is the case of MINPACK, which is probably needed for implementing QRDecomposition with pivoting.

Ideally, we should rewrite MINPACK (only the method needed by QRDecomposition) in Java and get rid of enumeration Style.

In particular, because we are employing Style at the moment, our implementation is slower than it needs to be.

0000239: Implement missing classes in joda-primitives

Richard Gomes <Richard Gomes@example.com> — Sat, 22 May 2010 22:37:57 +0100

These classes/interfaces are currently needed after we removed fastutil:

1. Utility classes: Arrays, Collections
2. Sets: HashSet
3. Maps: TreeMap, NavigableMap

Stephen Colebourne sent me some code he has which provides some idea how it could be done. See attached file.

0000403: Review Array and Iterators

Richard Gomes <Richard Gomes@example.com> — Sat, 22 May 2010 22:34:47 +0100

I've found the code below, which leads me to ask whether our implementation of Arrays and Iterators is good enough or not.

Could you please try to translate it using methods/constructors from Array and Iteraror hiearchy only? I mean: there's an implementation in Utilities class but it does the job by hand and it should not do so!

Real norm(const Iterator& begin, const Iterator& end, Real h) {
// squared values
std::vector f2(end-begin);
std::transform(begin,end,begin,f2.begin(),
std::multiplies());
// numeric integral of f^2
Real I = h * (std::accumulate(f2.begin(),f2.end(),0.0)
- 0.5*f2.front() - 0.5*f2.back());
return std::sqrt(I);
}

0000262: (recurrent task) QA :: verify headers

Richard Gomes <Richard Gomes@example.com> — Sat, 22 May 2010 22:29:43 +0100

======================================================================
This is a recurrent task
This task must never be closed, only kicked to next release and so on.
======================================================================

Verify our sources and guarantee:

* All sources have license text on the header;
* All sources have @author or similar or at least commented text for it.
* All sources have enough description of what it is about.

0000432: Complete DiscountingSwapEngine

Richard Gomes <Richard Gomes@example.com> — Tue, 18 May 2010 08:40:03 +0100

Complete DiscountingSwapEngine

0000472: Fix BondTest#yieldTest

Richard Gomes <Richard Gomes@example.com> — Sun, 18 Apr 2010 21:02:31 +0100

BondTest#yiedTest if failing from command line but not under Eclipse.

mvn clean verify

0000342: Review and debug SVD

Richard Gomes <Richard Gomes@example.com> — Sun, 04 Apr 2010 14:16:18 +0100

Review and debug SVD

Please see: MatrixTest

0000474: Review classes Money and Currency

Richard Gomes <Richard Gomes@example.com> — Sun, 04 Apr 2010 14:10:47 +0100

There are several issues on these classes.
Please open children issues if necessary:

1. inconsistent use of equals() and hashCode();

2. Money employs String#format() which takes a formatString defined by extended classes. We need to make sure these format strings are valid in Java and return what is expected, comparing to QuantLib/C++

3. These fields below should be probably exposed by class Settings:
public static ConversionType conversionType;
public static Currency baseCurrency;
At the moment they are are "static public" and being exposed by Money, which is both a bad practice and also is not ready for multithreaded applications.

0000340: Fix CholeskyDecomposition and EigenvalueDecomposition

Richard Gomes <Richard Gomes@example.com> — Sun, 04 Apr 2010 14:05:35 +0100

Please have a look at test cases in MatrixTest

0000380: Write complete test cases for Calendar

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:27:00 +0000

Write complete test cases for Calendar

0000422: Review exceptions :: should throw LibraryException

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:19:18 +0000

JQuantLib is throwing several kinds of exceptions.
It should always throw LibraryException, eventually passing another exception as a cause.

0000344: Possible bug in Calendars, regarding isBusinessDay

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:18:00 +0000

A user reported a possible bug. Please investigate:

In Calendar, I wonder if I have some misunderstandings about the API addHoliday. Once we call addHoliday(date) to add date to HolidayList, won't isBusinessDay(date) return false? Take ChinaSSECalendar as an example, I think the override method isBusinessDay should call the one in Super class to judge if the date is contained in addHolidayList?

0000356: The implementation of isEndOfMonth method in AbstractCalendar class is not correct

ggd543 <ggd543@example.com> — Tue, 09 Mar 2010 01:17:09 +0000

As described in the Calendar interface, the method `isEndOfMonth` in Calendar class return true if the date is last business day for the month in given market.

But in the AbstractCalendar class, the method will return true when the date is not the last business day for the month but the holiday day.

For example:
Calendar unitedStatesCalendar = UnitedStates.getCalendar(Market.NYSE);
Date date=DateFactory.getFactory().getDate(30,5,2009);
System.out.println(unitedStatesCalendar.isEndOfMonth(date));

The output is true ! Obviously, the date 2009-5-30 is neither the last business day nor the last day in May.

0000329: Announce JQuantLib

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:13:35 +0000

http://www.quanteam.de/papers/cpp_java_speed.pdf [^]
http://www.quanteam.de/site_en/publications.html [^]

Benchmarking C++ against Java
for Numerical Derivatives Pricing

Bernd Engelmann
Quanteam
Basaltstraße 28
D-60487 Frankfurt
e-mail: bernd@quanteam.de
www: http://www.quanteam.de [^]
¨
Soren Gerlach
Quanteam
Basaltstraße 28
D-60487 Frankfurt
e-mail: soeren@quanteam.de
www: http://www.quanteam.de [^]

0000273: build process :: nighly build should deploy artifacts

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:12:30 +0000

The nightly build process should deploy the following artifacts

.jar :: library
ubber.jar :: executable single jar file
apidocs.tar.bz2 :: javadocs
apidocs.zip :: javadocs
src.tar.bz2 :: sources
src.zip :: sources

0000471: Fix Array#toDoubleArray and callers

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:11:16 +0000

Array#toDoubleArray should always return a new double[] an never the underlying data buffer.

Review callers of Array#toDoubleArray so that they do not mix 2 different access semantics, i.e: sometimes take elements from the Array, sometimes take elements from a double[] obtained from the Array (which is, in fact, the underlying double[] buffer). In particular, class CubicInterpolation has this issue.

Open more children issues if needed.

0000392: Rewrite TimeGrid

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:06:20 +0000

TimeGrid needs a complete code review or even rewrite, possibly.

In particular, the usage of iterator needs to be clarified and compared to usage of iterators in class Array. There's a possibility that we find useful to make TimeGrid implement of make use somehow of Iterators like class Array does.

The idea is not change TimeGrid much, I mean: we should keep the API as close as possible but, at the same time, it needs to be coherent with Java (which does not offer pointers) and needs to be coherent with Iterators we've defined.

0000412: Finish translation of BondTest

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 01:04:17 +0000

static void testCached();
static void testCachedZero();
static void testCachedFixed();
static void testCachedFloating();
static void testBrazilianCached();

0000361: Fix QL_USE_INDEXED_COUPON code commented out

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 00:59:48 +0000

When ...

#if QL_USE_INDEXED_COUPON
block
#endif

... appears, please substitute by ...

if (new Settings.isUseIndexedCoupon()) {
block
}

0000427: Translate ConvertibleBondsTest

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 00:57:50 +0000

Translate ConvertibleBondsTest

0000397: Review equality tests with classes employing pimpl idiom

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 00:57:13 +0000

This tasks consists of finding and review classes which are exposing methods such like
* equals
* eq
* ne

It's critical we review the way equality is being implemented by certain classes, in particular those classes exposing some kind of "eq" method or overriding Object#equals method.

Observe that 'in general' Object#hashCode() must be overriden when Object#equals() is overridden.

In particular, the snippet of code below shows an example of how hashCode() and equals() must be implemented when the pimpl idiom is being employed. There's also an example of how Object#toString() must be implemented.

//
// overrides Object
//

@Override
public int hashCode() {
final int prime = 31;
int result = 1;
result = prime * result + ((impl_ == null) ? 0 : impl_.hashCode());
result = prime * result + ((impl_ == null) ? 0 : name().hashCode());
return result;
}

@Override
public boolean equals(final Object obj) {
if (this == obj) {
return true;
}
if (obj == null || !(obj instanceof DayCounter)) {
return false;
}

final DayCounter other = (DayCounter) obj;
if (this.empty() && other.empty()) {
return true;
}
if (this.name().equals(other.name())) {
return true;
}
return false;
}

@Override
public String toString() {
return (impl_ == null) ? "null" : impl_.name();
}

0000459: BarrierOptionTest#testPerturbative needs to be translated

Richard Gomes <Richard Gomes@example.com> — Tue, 09 Mar 2010 00:53:19 +0000

BarrierOptionTest#testPerturbative needs to be translated