We just discovered this problem as well.  Here's a test case that fails
without the fix.

Index: src/test/org/apache/lucene/index/TestCompoundFile.java
===================================================================
--- src/test/org/apache/lucene/index/TestCompoundFile.java
(revision 382277)
+++ src/test/org/apache/lucene/index/TestCompoundFile.java	(working
copy)
@@ -622,4 +622,26 @@
         is.close();
         cr.close();
     }
+
+    /** This test that writes larger than the size of the buffer output
+     * will correctly increment the file pointer.
+     */
+    public void testLargeWrites() throws IOException {
+        IndexOutput os = dir.createOutput("testBufferStart.txt");
+
+        byte[] largeBuf = new byte[2048];
+        for (int i=0; i<largeBuf.length; i++) {
+            largeBuf[i] = (byte) (Math.random() * 256);
+        }
+
+        long currentPos = os.getFilePointer();
+        os.writeBytes(largeBuf, largeBuf.length);
+
+        try {
+            assertEquals(currentPos + largeBuf.length,
os.getFilePointer());
+        } finally {
+            os.close();
+        }
+
+    }
 }
Index: src/java/org/apache/lucene/store/BufferedIndexOutput.java
===================================================================
--- src/java/org/apache/lucene/store/BufferedIndexOutput.java
(revision 382277)
+++ src/java/org/apache/lucene/store/BufferedIndexOutput.java	(working
copy)
@@ -58,6 +58,7 @@
           flush();
         // and write data at once
         flushBuffer(b, length);
+        bufferStart += length;
       } else {
         // we fill/flush the buffer (until the input is written)
         int pos = 0; // position in the input data


-----Original Message-----
From: Doug Cutting [mailto:cutting@apache.org] 
Sent: Thursday, March 02, 2006 11:20 AM
To: java-dev@lucene.apache.org
Subject: Re: 1.9 RC1

Shay Banon wrote:
> ...
>     } else {
>       // is data larger then buffer?
>       if (length > BUFFER_SIZE) {
>         // we flush the buffer
>         if (bufferPosition > 0)
>           flush();
>         // and write data at once
>         flushBuffer(b, length);
>       } else {
> ...
> 
> the bufferStart is not incremented after the flushBuffer method is  
> called. So if someone calls getFilePointer just afterwards, it will  
> give the wrong result (hit it with the compound format). A simple fix

> would be to add bufferStart += length; just after flushBuffer.

Can you please file a bug for this and attach a bug to it with a unit 
test that illustrates the problem?  This looks like something that could

warrant a 1.9.1 release, so we must proceed carefully.

Thanks,

Doug

---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-dev-help@lucene.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-dev-help@lucene.apache.org