Re: RFR: 8291792: DefaultStyledDocument.setCharacterAttribute accepts negative length [v4]

[Tejesh R]

On Wed, 24 Aug 2022 17:08:11 GMT, Phil Race <p...@openjdk.org> wrote:

> > > Although I've reviewed the CSR I'd prefer it not be finalized yet. I 
> > > think it needs more changes. The spec is SILENT about what then happens 
> > > if length is negative and SILENT about an out of range offset too. I 
> > > think we should address these issues as well.
> > > Also everywhere - in the CSR and the bug description and summary the text 
> > > said the method is called setCharacterAttribute whereas it is 
> > > setCharacterAttributes.
> > 
> > 
> > For negative length, now it just returns the control without processing it 
> > whereas before it used to raise an exception in arrayCopy method. Upper 
> > bound for length should be the text length(which is to be modified/added) I 
> > feel, still even if it crosses overall text size that case is handled by 
> > limiting to text length. These are not mentioned in spec, so should we 
> > modify the spec by adding the range bounds for length......?
> 
> Yes I am saying we should mention all of this

Will this addition to spec be fine -      

      * A write lock is held by this operation while changes
      * are being made, and a DocumentEvent is sent to the listeners
      * after the change has been successfully completed.
+     *
+     * <p>
+     * The expected {@Code length} range is the length of the text
+     * in which the attributes are set. If the length is &lt;= 0, then no
+     * attributes are set, the control returns. If the length is &gt; the
+     * length of text in which the attributes are to be set then the
+     * extra length is trimmed.
+     * </p>
+     *
      * <p>
      * This method is thread safe, although most Swing methods
      * are not. Please see

-------------

PR: https://git.openjdk.org/jdk/pull/9830