Problem getOutputStream of Runtime

Question

Hi all,
I'm facing a problem with getOutputStream Following is my code:
try {
String loadCommand = "svnadmin load " + repLocation;
Process loadProcess = runtime.exec(loadCommand);
OutputStream os = loadProcess.getOutputStream();
FileInputStream is = new FileInputStream(restoreLocation);
byte[] buff = new byte[BUFFSIZE];
int len = 0;
while ((len = is.read(buff)) > 0) {
os.write(buff, 0, len);
is.close();
os.close();
} catch (IOException e) {
log.error("IO EXCEPTION" + e.getMessage());
} catch (InterruptedException e) {
log.error("INTERRUPTED EXCEPTION" + e.getMessage());
i'm getting the following exception:
java.io.IOException: The pipe has been ended
     at java.io.FileOutputStream.writeBytes(Native Method)
     at java.io.FileOutputStream.write(Unknown Source)
     at java.io.BufferedOutputStream.flushBuffer(Unknown Source)
     at java.io.BufferedOutputStream.write(Unknown Source)
     at Load.<init>(Load.java:58)
     at Load.main(Load.java:107)
The above code is used to restore my repository.The code work upto some 30 revisions,But after some tiem i get the above exception.
Can anyone please point out my mistake!!
Not much of examples available for getOutpuStream method!!
Thanks
Archana

Answer 1

The dump file is basically the mixture of text headers and binary content, MD5 checksums are used to verify content, but no checksums for the headers (only the lenght is given, so that the headers content doesn't have to be eschaped).
It should be easy to find out in which revision the error occurs (since the svnadmin load outputs the information about which revision is being processed). You can either check it and try to correct manually or maybe examine and point out the error in the dump file.
A possible reason which comes to my mind right now is wrong encoding of some national characters, which causes the properties to be longer than the advertised lenght (since UTF-8 encoded characters take more than 1 byte). We experienced similar error with PVCS import.