Issue417665

Opening a file in append mode does not append to the
file but instead starts to write from index 0 which
results in overwriting the existing data.

To fix the problem, remove the seek() call in flush()
as it is not needed.  The diff is below.  Also see the
attached TestCase to demonstrate the problem (use the
unittest.py from CPython 2.1).  To run the unittest:

$> jython pyfiletest.py


Index: org/python/core/PyFile.java
===================================================================
RCS file:
/cvsroot/jython/jython/org/python/core/PyFile.java,v
retrieving revision 2.18
diff -u -r2.18 PyFile.java
--- org/python/core/PyFile.java 2001/03/13 20:21:27
 2.18
+++ org/python/core/PyFile.java 2001/04/20 16:02:40
@@ -455,7 +455,6 @@
         }

         public void flush() throws java.io.IOException {
-            file.seek(bufferStart);
             file.write(buffer, 0, dataSize);
             bufferModified = false;
             file.getFD().sync();
@@ -463,8 +462,7 @@

         public void close() throws java.io.IOException {
             if (writing && bufferModified) {
-                file.seek(bufferStart);
-                file.write(buffer, 0, (int)dataSize);
+                this.flush();
             }

             file.close();

Logged In: YES 
user_id=37674

After some more testing, I don't the patch fixes the problem
for all cases.

Logged In: YES 
user_id=266902

When you are in append mode,
write some data and flush twice before closing.

The result is: duplicated data in file

Logged In: YES 
user_id=4201

Fixed in PyFile.java: 2.20;

The fix only ensure that the filepointer is located at the 
end right after the open. Some unixes always write at the 
end of the file regardless of the filepointer but that is 
*not* the case for jython.