|
Open-Source
Work |
|
|
|
|
|
Janet |
|
|
|
|
|
Janet
(Java network) is a platform for distributed cooperative agents where
agent load is balanced automatically. It is written entirely in
Java using RMI and XML. Janet consists of two layers. The first layer
is called
Janet.CAS (Cooperative Agent System), which provides a platform
for distributed agents serving user-defined applications. The
second layer is called Janet.ADÉ (Automatic Distributed
Execution). It is based on Janet.CAS and carries out automatic agent
load balancing and agent load sharing. Janet.CAS can be used as
is without Janet.ADÉ. In this case agent load is not
balanced in the network. |
|
|
|
|
|
Bits &
Pieces
|
|
|
|
|
|
STIters
STIters is a little library
that adds Smalltalk-style collection iterators to Kotlin.
STIters
was
started
as
some
little
initial
Kotlin
programming
exercise.
Meanwhile
it
has
become
partially
obsolete as Kotlin has its own
standard library now named stdlib
that provides many of the collection iterators found in STIters.
Nevertheless, STIters contains some useful iterators in addition and
fixes some problems with iterators in stdlib as included in the Kotlin
plugin 0.1.2090 or earlier (see KT-1859).
|
|
|
|
|
|
Corporate
Work |
|
|
|
|
|
This section
describes application-independent and company-independent tools or
frameworks that were
developed while employed as
an internal employee. The rights of the respective software are
therefore with the applicable employers. For that reason only general
ideas are explained and no source code
is provided. |
|
|
|
|
|
Tuple-space
oriented hierarchical
calculation framework |
|
|
|
|
|
A framework was
developed for the calculation of LGDs
(loss given defaults) for a bank as a general purpose framework for the
calculation of business figures. This framework
is not about
number
crunching
but
about
decomposing
extensive
formulas
into
manageable
parts
as
nodes
in
a
tree.
A
typical
application
domain
is
calculation
of
business
key
figures
(such
as
ratings, loss given defaults, expected losses, etc.)
where often extensive formulas are evaluated while iterating over
collections of inter-connected model objects. The framework defines an
interface for calculation objects (which make up the nodes in the
calculation tree) and for result objects. Result objects are stored in
a result space that is conceptually similar to a Linda-style tuple
space (but not distributed). The result object is only calculated if it
cannot be found in
the result space. Using this approach to decouple calculation objects
and result objects, the order in which result objects are calculated by
the calculation objects is arbitrary at compile-time leading to greater
flexibility at development-time.
|
|
|
|
|
|
SAP-to-Java
Mapping
Framework
|
|
|
|
|
|
This framework is
on top of SAP Java Connector (SAP JCo) that makes calling a SAP remote
function module from Java using SAP JCo a lot less effortful and
straight-forward. In the same way as hibernate is a high-level
framework on top of a simple API such as JDBC, the SAP-to-Java Mapping
Framework is a high-level framework on top of SAP JCo. It transparently
carries out the assembly and disassembly between SAP records and Java
objects using a declarative approach with XML mapping files (similar
approach as with hibernate hbm.xml files).
|
|