Our implementation of BufferedReader calls FileInputStream.available()
while the RI does not. This is a problem because it means that the RI
can read certain types of files that our implementation can not - such
as /proc/mounts on Linux. I will investigate this shortly.
While looking at this I notice that our implementation of
FileInputStream.available() does:
long currentPosition = fileSystem.seek(fd.descriptor, 0L,
IFileSystem.SEEK_CUR);
long endOfFilePosition = fileSystem.seek(fd.descriptor, 0L,
IFileSystem.SEEK_END);
fileSystem.seek(fd.descriptor, currentPosition, IFileSystem.SEEK_SET);
return (int) (endOfFilePosition - currentPosition);
making three JNI calls. It might be better to implement this as a
single JNI call - particularly since at the moment we seem to be calling
it more often than the RI.
Comments welcome.
Mark.
