Java应用程序在Eclipse上工作正常，但在部署为可运行的jar后它无法正常工作

Question

Like the tile said, my Java application doesn't work as it works fine while developing on Eclipse.

I finished to build and test my application on Eclipse. It works fine as I expected. I needed to deploy it as a runnable jar so that my client can't use it on their system. I made it a runnable jar through exporting it on Eclipse. When I run the runnable jar, the application starts to work fine for a while and stops on file reading. I haven't had any problem with it on the same code on Eclipse.

The source snippet that seems not to work is the following.

            sfis = new SmbFileInputStream(sFile);
            in = new BufferedInputStream(sfis);

            byte buf[] = new byte[(int)sFile.length()];
            int pos = 0;
            int size = 10;
            int temp;

            while((size=in.read(buf, pos, size)) > 0){
                pos += size; 
                temp = buf.length - pos;
                if(temp < 10){
                    size = temp;
                }
            }

On Eclipse it doesn't cause any problem. It perfectly reads data from SMB connection and finishes its job. But from the jar application, it seems to stop reading from inputstream inside while loop at some point. This is a very strange case that I have never experienced before. And I can't find any solution for this.

Is this because of my code or the jar file that might be wrongly made by Eclipse?

---------- additional info -------------------

With a big help from Edmondo1984, I find out where the jar program stops. When it goes to inputstream from SmbFile, a new thread jcifs.util.transport.Transport is created and the thread is simply blocked and the application creates another jcifs.util.transport.Transport, and so on. After creating 8 or 9 jcifs.util.transport.Transport thread, it is being stuck and doing nothing.

The same code, the same jcifs library. But works differently between running on Eclipse and on local machine as a runnable jar. I have no idea why this is happening.

Answer 1

The return value for "there is nothing more on the stream" is -1 (see Javadoc ).

Imagine what happens if you try to in.read(buf, pos, size) but the underlying is still not available. Then the call immediately returns with a return value of 0 (zero). This way, the loop condition evaluates to false and the loop is exited. So I think you should compare against != -1 .

Another possible issue: You may also consider what happens if size = 0 (eg due to above scenario) and temp >= 10 . Because you directly feed back the value of size to in.read(buf, pos, size) as the maximum number of bytes to read, you could end up in an infinite loop, assuming you have already corrected the comparison to != -1 .

Answer 2

When meddling with streams, those kind of behaviors are frequent. Small algorithmic incorrections may cause different symptoms in different environments.

You could print "debug" information to the console to better understand where and why is your code stalling. I'm pretty sure its on the read method.

There are several things wrong with your code, the most notorious is the block:

temp = buf.length - pos;
if(temp < 10){
  size = temp;
}

Why? When there are 10 or less bytes left you'll try to read them in the next read(), otherwise you'll potentially try to read too much (the whole file size again). Running in Eclipse, the file is probably being read in the first read(), out of Eclipse apparently not. Try to change your code to something like this:

FileInputStream sfis = new FileInputStream(sFile);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
byte buf[] = new byte[4096];
int read = sfis.read(buf);
while(read>0){
  baos.write(buf, 0, read);
  read = sfis.read(buf);
}
sfis.close();

Also consider to use Apache's Java IO utilities code instead of always writing this type of code. IO in Java needs a lot of boilerplate, and a good programmer has better things to do.

On a side-note: reading a file to memory is usually a bad idea. Unless you're going to transform its data as a whole in some way, streaming is always better. I don't know what is your program for but please keep this in mind: what happens if you try to read a 10GB file? or several?

Happy coding,

Answer 3

Java is capital letter sensitive in evry way. For example if you have file named image.PNG and in your code you set location to "/image.png" (instead of "/image.PNG") it will work fine in eclipse, but after exporting runnable jar file java VM will be not able to find that file.