Features: Missing and Unintentional

This is a list of `features' in the Omegahat R-Java interface. Some are features that we should add soon. Others are real bugs that we have encountered. We have fixed some of these and are looking for more information about the others. Please let us know if you have any information about these.

Outstanding

DynamicCompile doesn't produce a loadable .class file. The first time one calls this, one gets a NULL. z <- .JNew("FloatMatrix", .convert=FALSE) The example dataFrameViewerReference.R illustrates the problems with threads. Problems with threads and specifically the GUI can mean we get random crashes. Need to throw a RuntimeException if there is an error in R. If we get an error when calling a Java method but have converted R objects to Java objects during the call from R to Java, these may never be released. Need to keep a vector of these and release a reference when they are done. Or do they automatically disappear because nobody has a reference to them?
The first time one calls this, one gets a NULL. z <- .JNew("FloatMatrix", .convert=FALSE) The example dataFrameViewerReference.R illustrates the problems with threads. Problems with threads and specifically the GUI can mean we get random crashes. Need to throw a RuntimeException if there is an error in R. If we get an error when calling a Java method but have converted R objects to Java objects during the call from R to Java, these may never be released. Need to keep a vector of these and release a reference when they are done. Or do they automatically disappear because nobody has a reference to them?
The example dataFrameViewerReference.R illustrates the problems with threads. Problems with threads and specifically the GUI can mean we get random crashes. Need to throw a RuntimeException if there is an error in R. If we get an error when calling a Java method but have converted R objects to Java objects during the call from R to Java, these may never be released. Need to keep a vector of these and release a reference when they are done. Or do they automatically disappear because nobody has a reference to them?
Problems with threads and specifically the GUI can mean we get random crashes. Need to throw a RuntimeException if there is an error in R. If we get an error when calling a Java method but have converted R objects to Java objects during the call from R to Java, these may never be released. Need to keep a vector of these and release a reference when they are done. Or do they automatically disappear because nobody has a reference to them?
Need to throw a RuntimeException if there is an error in R. If we get an error when calling a Java method but have converted R objects to Java objects during the call from R to Java, these may never be released. Need to keep a vector of these and release a reference when they are done. Or do they automatically disappear because nobody has a reference to them?
If we get an error when calling a Java method but have converted R objects to Java objects during the call from R to Java, these may never be released. Need to keep a vector of these and release a reference when they are done. Or do they automatically disappear because nobody has a reference to them?

Fixed

The ConvertibleClassifier needs to support the idea of any instanceof, any assignable from and exact match for a class that it supports. See the R function setJavaConvertible() and Java method setConvertible (in the OmegaInterfaceManager evaluator and ConverterClassifier)
Returning a string array doesn't seem to give any result. Actually, it seems that it doesn't get printed. Example: u <- .JavaConstructor("StatDataURL", "http://www.omegahat.org/Scripts/AwtButton") .Java(u,"getContents")

Duncan Temple Lang <duncan@research.bell-labs.com>

Last modified: Mon Aug 30 13:24:03 PDT 2004