FYI: GapContent test fix

FYI: GapContent test fix

[Roman Kennke]

[-- Attachment #1: Type: text/plain, Size: 997 bytes --]

In the test for GapContent.insertString() we have tested the insertion of
a string at the index after the last character. There was a comment in
it saying that it is a bug in the JDK to not throw a
BadLocationException. However, AFAICS this is not the case. Suppose you
have the following content:

ABCDEFG

and now call GapContent.insertString(7, "XYZ"), then this will (quite
logically IMO) append this:

ABCDEFGXYZ

There is no need to throw a BadLocationException here. What might have
caused confusion here is the fact that the GapContent usually has a
trailing \n character, and normal insertion from an AbstractDocument
never inserts after the \n, but this is an implementation detail of the
document and not GapContent.

I checked in the appended patch to fix this problem.

2006-02-02  Roman Kennke  <kennke@aicas.com>

        * gnu/testlet/javax/swing/text/GapContent/insertString.java
        (testGeneral): Fixed boundary case to pass with the JDK.

/Roman

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: patch.diff --]
[-- Type: text/x-patch; name="patch.diff", Size: 1176 bytes --]

Index: gnu/testlet/javax/swing/text/GapContent/insertString.java
===================================================================
RCS file: /cvs/mauve/mauve/gnu/testlet/javax/swing/text/GapContent/insertString.java,v
retrieving revision 1.3
diff -u -r1.3 insertString.java
--- gnu/testlet/javax/swing/text/GapContent/insertString.java	24 Jan 2006 17:18:56 -0000	1.3
+++ gnu/testlet/javax/swing/text/GapContent/insertString.java	2 Feb 2006 10:44:05 -0000
@@ -215,19 +215,17 @@
     }
     harness.check(gc.length(), 7);
     
-    // insert at index of last character + 1 - the API docs say this should
-    // raise a BadLocationException, but JDK1.5.0_06 doesn't (I reported this
-    // to Sun as a bug in their implementation, will add a bug ID if one is assigned)
-    pass = false;
+    // Insert at index of last character + 1. This appends the new string to
+    // the existing string.
     try
     {
       gc.insertString(7, "XYZ");
     }
     catch (BadLocationException e)
     {
-      pass = true;
+      pass = false;
     }
-    harness.check(pass);
+    harness.check(gc.length(), 10);
 
     // insert at index of last character + 2 
     pass = false;

Re: FYI: GapContent test fix

[David Gilbert]

Hi Roman,

The API docs describe the first parameter (where) as:

|"where| - the starting position >= 0, < length()"

...and in this particular test I've used length(), so I'd expect the 
BadLocationException to be thrown.  To back this up further, the exact 
same test on StringContent *does* throw the exception.  I didn't hear 
back from Sun yet on the bug report (apart from the automated response), 
but I do expect that they'll accept it and fix it eventually.

Anyway, I'll let you know when I hear from them.

Regards,

Dave


Roman Kennke wrote:

>In the test for GapContent.insertString() we have tested the insertion of
>a string at the index after the last character. There was a comment in
>it saying that it is a bug in the JDK to not throw a
>BadLocationException. However, AFAICS this is not the case. Suppose you
>have the following content:
>
>ABCDEFG
>
>and now call GapContent.insertString(7, "XYZ"), then this will (quite
>logically IMO) append this:
>
>ABCDEFGXYZ
>
>There is no need to throw a BadLocationException here. What might have
>caused confusion here is the fact that the GapContent usually has a
>trailing \n character, and normal insertion from an AbstractDocument
>never inserts after the \n, but this is an implementation detail of the
>document and not GapContent.
>
>I checked in the appended patch to fix this problem.
>
>2006-02-02  Roman Kennke  <kennke@aicas.com>
>
>        * gnu/testlet/javax/swing/text/GapContent/insertString.java
>        (testGeneral): Fixed boundary case to pass with the JDK.
>
>/Roman
>  
>
>------------------------------------------------------------------------
>
>Index: gnu/testlet/javax/swing/text/GapContent/insertString.java
>===================================================================
>RCS file: /cvs/mauve/mauve/gnu/testlet/javax/swing/text/GapContent/insertString.java,v
>retrieving revision 1.3
>diff -u -r1.3 insertString.java
>--- gnu/testlet/javax/swing/text/GapContent/insertString.java	24 Jan 2006 17:18:56 -0000	1.3
>+++ gnu/testlet/javax/swing/text/GapContent/insertString.java	2 Feb 2006 10:44:05 -0000
>@@ -215,19 +215,17 @@
>     }
>     harness.check(gc.length(), 7);
>     
>-    // insert at index of last character + 1 - the API docs say this should
>-    // raise a BadLocationException, but JDK1.5.0_06 doesn't (I reported this
>-    // to Sun as a bug in their implementation, will add a bug ID if one is assigned)
>-    pass = false;
>+    // Insert at index of last character + 1. This appends the new string to
>+    // the existing string.
>     try
>     {
>       gc.insertString(7, "XYZ");
>     }
>     catch (BadLocationException e)
>     {
>-      pass = true;
>+      pass = false;
>     }
>-    harness.check(pass);
>+    harness.check(gc.length(), 10);
> 
>     // insert at index of last character + 2 
>     pass = false;
>  
>

Re: FYI: GapContent test fix

[David Gilbert]

Roman Kennke wrote:

>Hi David,
>
>Am Donnerstag, den 02.02.2006, 11:16 +0000 schrieb David Gilbert:
>  
>
>>Hi Roman,
>>
>>The API docs describe the first parameter (where) as:
>>
>>|"where| - the starting position >= 0, < length()"
>>
>>...and in this particular test I've used length(), so I'd expect the 
>>BadLocationException to be thrown.  To back this up further, the exact 
>>same test on StringContent *does* throw the exception.  I didn't hear 
>>back from Sun yet on the bug report (apart from the automated response), 
>>but I do expect that they'll accept it and fix it eventually.
>>    
>>
>
>Now that is interesting. So I was a little too quick with my fix. It
>seemed to make sense to be able to append, but if StringContent behaves
>differently, I am not so sure anymore. So you think we should revert my
>patch?
>
>/Roman
>
>  
>
I think so, but if you're busy it's not so important - I'll revisit it 
when I hear back from Sun on the bug report.

Regards,

Dave