News and New Product Briefs (5/05/99)

HP surrenders Java benchmark to independent party

Hewlett-Packard has turned over the Java E-Business Transactional Benchmark, or JETMark, to Netroscope, an independent consulting concern, as a way of eliminating the appearance of bias.

HP co-developed JETMark with Vision Software Tools of Oakland, CA. JETMark measures the number of electronic transactions processed per hour for various combinations of Java software and hardware environments. The tests simulate typical Web-based order-entry transactions among a client, application server, and database, regardless of whether the environment is relational, ERP-based, legacy-based, or mixed.

JETMark consists of the following components:

  • JETMark Test Harness, a framework that creates arbitrary or customized combination of workloads that includes reading information, inserting orders, changing orders, or removing orders
  • Java E-Commerce Server, a combination of the software/hardware (including specified business logic) being measured for the transaction workload
  • Commercial RDBMS, a data repository that acts as a high-performance back-end processor required for an n-tier application

Netroscope officials said that vendors would pay for the tests, then decide whether they wanted the results to remain private or released by Netroscope. IT departments would get results gratis.

HP's move is intended to provide more credible testing of Java e-commerce applications, according to company officials. Susan Henson, HP's Java program manager, said, "We are acting as a catalyst to move this benchmark forward and to have it become independent, so customers can have a yardstick they dearly need today to measure their commerce environments."

Look for the first JETMark results on products late in May 1999 on the Netroscope site.

http://www.netroscope.com/JETMark/index.html

Glance at Painted Word's Eyeball spreadsheet component

Painted Word announced the Eyeball Spreadsheet Bean, a visual development, Web-deployable Java spreadsheet component for OLAP that can perform a full ad hoc analysis on multidimensional data, format and print reports, and export to Excel.

Besides its ability to perform ad hoc analysis of multidimensional data, Eyeball uses its common spreadsheet interface to provide a full-featured spreadsheet with the expected calculation, cell-level formatting, and printing capabilities. It uses the familiar pull-down menus and mouse clicks to access and navigate databases, and it supports importing and exporting to Excel.

Other Eyeball features include:

  • The Painted Word Mocha Blend framework, which provides connectivity to such multidimensional data sources as Hyperion Essbase and IBM DB2 OLAP Server
  • Complete OLAP data navigation, including pivoting and drilling, remote lock and send, database calculation, and control over database-specific viewing and navigation options
  • Import and export filters from and to Microsoft Excel 5 and 7 workbooks
  • Rich cell formatting support
  • Full write-back capability, so Eyeball can be used for Web-based planning applications and user templates
  • Printing support, so remote users can print with complete fidelity to any printer on the network at full printer resolution
  • More than 150 worksheet functions and operators
  • Server-side deployment that lets Eyeball offer access to any Java 1.1-enabled client without additional software
  • The ability to export worksheets to HTML format for Web publishing

Painted Word's Mocha Blend Analytical Platform is an open, Internet standards-based framework and Java component suite designed to aid rapid development and deployment of Web-based analytical applications. The Eyeball Bean can be manipulated via the Mocha Beans, which are visual components including charts, outline trees, and member combo boxes.

The Eyeball Spreadsheet Bean should be prereleased in late May 1999; the company is currently recruiting participants for a beta program.

http://www.paintedword.com/html/eye_info.html

Nettaxi, BigNetwork.com offer multiplayer Java game

Nettaxi and BigNetwork.com announced a co-marketing agreement that brings access of BigNetwork.com's online games to Nettaxi's online community. (Nettaxi is an online community and portal Web site that draws more than 80 million page views per month. BigNetwork.com is the creator of BigNetwork Classic Games, a multiplayer Java game system that features instant click-and-play access.)

Nettaxi community members will get free access to a suite of BigNetwork.com games such as chess, checkers, backgammon, reversi, spades, and morph. They can interactively play these games in realtime with other community members, as well as with BigNetwork.com's 200,000 registered members.

Also, Nettaxi members will be able to embed Java-based games (and an interactive gaming environment) into their own Web sites.

Nettaxi http://www.nettaxi.com/

BigNetwork.com http://www.BigNetwork.com/

Two new bean suites from alphaWorks

IBM's alphaWorks recently announced two new JavaBean suites -- one to internationalize applications and one to help developers handle multithreading.

The Multiple Language Suite includes MLButton, MLLabel, MLCheckbox, MLTextField, MLTextArea, MLList, and MLChoice beans. The beans in this suite support the display and input of different languages running on any platform, since the beans are independent of a specific language environment. In the current version, designed for JDK 1.1 and later, English and Chinese display is supported.

The Thread Bean Suite consists of VisualThread, ResourceLocker, and ThreadStatePrinter beans. These beans can be used in any application where a time-consuming task needs to be performed on different threads. For example, a laborious picture-processing operation can be split into four quarters with each quarter completed in a separate thread. This suite, designed for JDK 1.1 and later, handles the issue of thread safety for resources that may not be thread-safe, and also facilitates debugging by allowing a trace to be printed for each thread. These beans can be wired together in most visual builder tools.

Multiple Language suite: http://www.alphaworks.ibm.com/ab.nsf/bean/Multiple+Language

Thread Bean suite: http://www.alphaworks.ibm.com/ab.nsf/bean/Thread

IBM offers XML Security Suite

IBM's alphaWorks announced the XML Security Suite, which introduces such security features as digital signature, element-wise encryption, and access control to XML-based, Internet business-to-business transactions.

Running on Linux and Windows 95/98/NT platforms, the XML Security Suite offers more security with new features (digital signature, element-wise encryption, access control) that move beyond the capabilities of transport-level security protocols such as SSL.

This release provides reference implementations of DOMHASH (a possible digital signature source currently under consideration in the IETF and W3C), a proposed canonicalized digest value for XML document, and two sample applications.

http://www.alphaWorks.ibm.com/tech/xmlsecuritysuite

IBM debuts XML for C++ library

IBM's alphaWorks announced the XML for C++ parser (XML4C), a single C++ shared library that provides classes for parsing, generating, manipulating, and validating XML documents.

XML4C is a validating XML parser written in a portable subset of C++. It makes it fairly simple to deliver the ability to read and write XML data to an application. It can be used to construct Web servers that are compatible with XML standards and internationalized XML applications.

XML4C adheres to the XML 1.0 Recommendation and accompanying standards such as DOM 1.0 and SAX 1.0.

It runs on AIX and Windows NT, and comes with the parser, source code, samples, and API documentation.

http://www.alphaWorks.ibm.com/tech/xml4c

Java Advanced Imaging API beta

The Java Developer Connection announced the beta reference port of the Java Advanced Imaging API (JAI), which extends Java by letting developers incorporate high-performance image processing into applets and applications.

The JAI API implements image-processing capabilities, such as image tiling, regions of interest, and deferred execution, and also provides a set of standard image-processing operators, including many common point, area, and frequency domain operators.

The JAI API is extendable to allow arbitrary processing capabilities, and it supports deferred execution, remote imaging, distributed imaging, and a wide range of image formats. It can be configured to implement in different ways, in order to balance the available memory, hardware acceleration, and optimization resources.

This beta includes a reference implementation for almost all of the functionality described in the specification. You will have to register to access this site.

http://developer.java.sun.com/developer/earlyAccess/jai/index.html

Java performance report

Osvaldo Pinali Doederlein offers the Java Performance Report, a report designed to dive into performance results of various JVMs (and alternatives) to discover the level of efficiency of each one, as well as to determine the prominent characteristics of each to make it easier for developers to choose the one for certain types of operations.

According to Doederlein, though, benchmark numbers alone (which are easily obtained) are not enough. He notes that this report should offer more in-depth understanding of performance characteristics, including trade-offs, rationale, and consequences inherent in each design.

The report covers the following JVMs and configurations:

  • Sun JDK 1.0.2, no JIT
  • Sun JDK 1.1.7B, Symantec JIT 3.00.072b(x)
  • Sun JDK 1.2, Classic VM, Symantec JIT 3.00.078(x)
  • Microsoft SDK for Java 3.2/IE5, build 5.00.3167
  • IBM's version of JDK 1.1.7, build n117p-19990408 with its ibmjitc
  • Blackdown JDK 1.1.7 v1a, green threads (native is worse on uniprocessor machines), TYA JIT 1.2
  • Tower LLC's TowerJ 2.2.7.0 (native compiler), assembling through Microsoft Visual C++ 5.0sp3 on Win32/gcc on Linux, and importing JDK 1.1.7B's core libraries
  • TowerJ, including the "omit runtime checks" optimization
  • Linux-x86 version of TowerJ (glibc)

It covers the following benchmarks (looking at interpreters vs. JIT and pure Java vs. native):

  • jBYTEMark 0.9 (BYTE Magazine)
  • JMark 2.0 (Ziff-Davis Labs)
  • CaffeineMark 3.0 (Pendragon)
  • VolanoMark 2.1 (Volano)

The SPECjvm98 is absent.

It also analyzes VM behaviors and benchmarking memory, and provides a clear appendix of methods used.

http://www.javalobby.org/features/jpr/

alphaWorks FoCuS defines coverage test models

IBM alphaWorks announced FoCuS, a generic functional-coverage tool that can measure the coverage properties of a given set of event traces over coverage models.

FoCuS lets developers define functional coverage models (tests that determine that a new application does everything it is supposed to do), then it generates those models by collecting the data, creating regressions, and displaying reports.

FoCuS is not tied to any specific coverage model, but it can use the same tools they all employ. FoCuS models may exist for any level -- architecture, micro-architecture, implementation -- depending on the inputs. The definition of the coverage models is executed in a high-level language (SQL) enhanced with macros.

It runs on any Java-enabled platforms, Windows 95/NT, and flavors of Unix.

http://www.alphaWorks.ibm.com/tech/focus

NaturalBridge debuts BulletTrain Java bytecode compiler

NaturalBridge announced BulletTrain, a static Java bytecode compiler, library, and runtime environment that uses optimization techniques with an integrated runtime support. It's available for Pentium- and Windows 95/98/NT-based systems.

BulletTrain includes an optimizing ahead-of-time compiler, a recompilation manager, a linker, optimized libraries compatible with Java 2, and debugging tools.

The 220,000-line BulletTrain's runtime environment supports thousands of threads and it scales well on multiprocessor systems. The compiler can handle applications with hundreds of thousands of lines of code. And it implements Java 2 and can compile and run JDK 1.2.1 libraries and applications, too.

The BulletTrain compiler compiles .class files (produced by Java compilers such as javac or jikes) into platform-dependent binary files. Compilation optimizations include constant propagation, type propagation, check reduction, dead code elimination, loop unrolling, and aggressive method inlining. The optimizer uses SSA-based analyses optimized for Java programs.

The BulletTrain linker combines compiled files into shared libraries or applications, and all applications linked against the same library will share the same code. BulletTrain only supports linking with a single library, but that library can be linked against other libraries in a chain.

BulletTrain keeps track of how object files depend upon their classfile inputs; it automatically recompiles object files as necessary. This allows aggressive inlining and constant propagation across methods, classes, and packages. Given a main class and a class path, BulletTrain can find and include all the class files needed to finish the construction of an application.

The available debugging tools can complete stack traces; backtrace all threads at interrupt; log handled and unhandled exceptions allocated; log interesting thread operations; perform parameter checking for JNI operations; and perform on-the-fly deadlock detection.

1 2 3 Page 1
Page 1 of 3