On 25/01/2010 16:59, Tim Ellison wrote:
> On 25/Jan/2010 16:09, Oliver Deakin wrote:
>    
>> Looks like this is caused by DTD.java's use of available() again. This
>> time the test is failing because RAFStream.available() is being called
>> by DTD.read() and its return value is expected to be a true indication
>> of the number of bytes available. However, the RAFStream.available()
>> call will only return 0 or 1 in our implementation which causes the
>> DTD.read() to only read a single byte rather than the whole DTD. I
>> hacked RAFStream.available() to return (mLength-mOffset) and this makes
>> the test pass, but twe really need to fix the DTD.read() method to not
>> use available() at all I think.
>>      
> It looks like both need fixing then.  I thought we had done all these
> already?
>
> RAFStream should return the actual number of bytes available, and
> DTD.read() should not use available, but read until it gets EOF (-1).
>    

Agreed. Perhaps something along the lines of the patch below. I have not 
fully tested it yet, but this certainly fixes the test case in question.

Regards,
Oliver

Index: modules/archive/src/main/java/java/util/zip/ZipFile.java
===================================================================
--- modules/archive/src/main/java/java/util/zip/ZipFile.java    
(revision 901661)
+++ modules/archive/src/main/java/java/util/zip/ZipFile.java    (working 
copy)
@@ -378,7 +378,15 @@

          @Override
          public int available() throws IOException {
-            return (mOffset < mLength ? 1 : 0);
+            if (mOffset < mLength) {
+                if (mOffset - mLength < Integer.MAX_VALUE) {
+                    return (int)(mLength-mOffset);
+                } else {
+                    return Integer.MAX_VALUE;
+                }
+            } else {
+                return 0;
+            }
          }

          @Override
Index: 
modules/swing/src/main/java/common/javax/swing/text/html/parser/DTD.java
===================================================================
--- 
modules/swing/src/main/java/common/javax/swing/text/html/parser/DTD.java    
(revision 901661)
+++ 
modules/swing/src/main/java/common/javax/swing/text/html/parser/DTD.java    
(working copy)
@@ -16,6 +16,7 @@
   */
  package javax.swing.text.html.parser;

+import java.io.ByteArrayOutputStream;
  import java.io.DataInputStream;
  import java.io.IOException;
  import java.util.BitSet;
@@ -142,9 +143,18 @@

      public void read(final DataInputStream stream) throws IOException {
          // converts from DataInputStream into a byte array
-        byte[] enc = new byte[stream.available()];
-        stream.read(enc);
+        byte[] enc = new byte[1024];
+        ByteArrayOutputStream bs = new ByteArrayOutputStream();

+        int iRead = 0;
+        while (iRead != -1) {
+            iRead = stream.read(enc, 0, enc.length);
+            if (iRead > 0) {
+                bs.write(enc, 0, iRead);
+            }
+        }
+        enc = bs.toByteArray();
+
          // decode the byte array
          Asn1Dtd asn1 = new Asn1Dtd(enc);


> Regards,
> Tim
>
>    
>> On 25/01/2010 13:15, Oliver Deakin wrote:
>>      
>>> Hi Ray,
>>>
>>> Thanks for the information - from your results it seems that there
>>> must be a difference in common classes between java 5 and 6 that are
>>> causing the failures. I see the same result - with the java5 M12a
>>> swing.jar the test passes 100% and with the java6 swing.jar it fails
>>> 100%.
>>>
>>> I'll dig a little deeper at the code differences between the modules.
>>>
>>> Regards,
>>> Oliver
>>>
>>> On 25/01/2010 12:10, Ray Chen wrote:
>>>        
>>>> Hi Oliver,
>>>> I have done some investigation for this issue before, and disscussed
>>>> it on the 6.0 Milestone1 thread.
>>>>
>>>> I compared two swing.jar of trunk and java6, the only difference
>>>> between them is that java6 added some new classes:
>>>>
>>>> javax\swing\event\RowSorterEvent$Type.class
>>>> javax\swing\event\RowSorterEvent.class
>>>> javax\swing\event\RowSorterListener.class
>>>> javax\swing\filechooser\FileNameExtensionFilter.class
>>>> javax\swing\RowSorter$SortKey.class
>>>> javax\swing\RowSorter.class
>>>> javax\swing\SortOrder.class
>>>>
>>>> I think these classes are not related to this test. However, even
>>>> after I removed these new added classes or replaced trunk's swing.jar
>>>> with java6's swing.jar and run the test on the modified trunk , the
>>>> test still failed.
>>>> And trunk's swing.jar works well in both trunk and java6 to run this
>>>> test.
>>>> So I am a little confused,  same classes, same vm, why different result?
>>>>
>>>> On Mon, Jan 25, 2010 at 7:03 PM, Oliver Deakin
>>>> (JIRA)<jira@apache.org>   wrote:
>>>>          
>>>>>      [
>>>>> https://issues.apache.org/jira/browse/HARMONY-6392?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12804502#action_12804502
>>>>> ]
>>>>>
>>>>> Oliver Deakin commented on HARMONY-6392:
>>>>> ----------------------------------------
>>>>>
>>>>> This issue has been fixed in java5 M12, but still persists in the
>>>>> java6 branch.
>>>>>
>>>>>            
>>>>>> [classlib][swing] Test failures in
>>>>>> javax.swing.text.html.HTMLDocument_Reader_ActionsTest
>>>>>> ----------------------------------------------------------------------------------------
>>>>>>
>>>>>>
>>>>>>                   Key: HARMONY-6392
>>>>>>                   URL:
>>>>>> https://issues.apache.org/jira/browse/HARMONY-6392
>>>>>>               Project: Harmony
>>>>>>            Issue Type: Bug
>>>>>>            Components: Classlib
>>>>>>      Affects Versions: 6.0M1, 5.0M12
>>>>>>           Environment: Windows x86
>>>>>>              Reporter: Oliver Deakin
>>>>>>               Fix For: 5.0M12
>>>>>>
>>>>>>
>>>>>> I see 1 failure and 1 error.
>>>>>> javax.swing.text.html.HTMLDocument_Reader_ActionsTest.testHarmony_4582
>>>>>> fails with output:
>>>>>> N/A
>>>>>> java.lang.NullPointerException
>>>>>> at
>>>>>> javax.swing.text.html.HTMLDocument_Reader_ActionsTest.testHarmony_4582(HTMLDocument_Reader_ActionsTest.java:584)
>>>>>>
>>>>>> at java.lang.reflect.VMReflection.invokeMethod(VMReflection.java)
>>>>>> at
>>>>>> javax.swing.BasicSwingTestCase.runBareSuper(BasicSwingTestCase.java:116)
>>>>>>
>>>>>> at
>>>>>> javax.swing.BasicSwingTestCase.runBareImpl(BasicSwingTestCase.java:121)
>>>>>>
>>>>>> at javax.swing.BasicSwingTestCase$1.run(BasicSwingTestCase.java:135)
>>>>>> at java.lang.Thread.run(Thread.java:669)
>>>>>> javax.swing.text.html.HTMLDocument_Reader_ActionsTest.testHarmony_4615
>>>>>> fails with output:
>>>>>> null expected:<[line4 line4]>   but was:<[ line4 line]>
>>>>>> junit.framework.ComparisonFailure: null expected:<[line4 line4]>
>>>>>> but was:<[line4 line]>
>>>>>> at
>>>>>> javax.swing.text.html.HTMLDocument_Reader_ActionsTest.testHarmony_4615(HTMLDocument_Reader_ActionsTest.java:601)
>>>>>>
>>>>>> at java.lang.reflect.VMReflection.invokeMethod(VMReflection.java)
>>>>>> at
>>>>>> javax.swing.BasicSwingTestCase.runBareSuper(BasicSwingTestCase.java:116)
>>>>>>
>>>>>> at
>>>>>> javax.swing.BasicSwingTestCase.runBareImpl(BasicSwingTestCase.java:121)
>>>>>>
>>>>>> at javax.swing.BasicSwingTestCase$1.run(BasicSwingTestCase.java:135)
>>>>>> at java.lang.Thread.run(Thread.java:669)
>>>>>> These tests pass for me against M11.
>>>>>>              
>>>>> -- 
>>>>> This message is automatically generated by JIRA.
>>>>> -
>>>>> You can reply to this email to add a comment to the issue online.
>>>>>
>>>>>
>>>>>            
>>>>
>>>>          
>>>        
>>      
>    

-- 
Oliver Deakin
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU