Re: Review Request for 8057574: Inconsistent behavior for setBackground (Windows/Linux)

[Semyon Sadetsky]

On 1/21/2016 3:55 PM, Sergey Bylokhov wrote:
On 21/01/16 09:42, Semyon Sadetsky wrote:


On 1/21/2016 4:05 AM, Sergey Bylokhov wrote:
On 19/01/16 12:12, Semyon Sadetsky wrote:
Hi Prem,

I would choose to inherit all them since the javadoc states that.

There are more issue:
 - The javadoc of Compoenent.getBackground() states that the color of
the parent container should be returned, but the window has not parent
container it has the owner. Moreover the javadoc for
Window.getBackground() simply states that it returns the color of the
window.
What I see in the code:
Component.java 1812:
     /**
      * Gets the background color of this component.
      * @return this component's background color; if this component 
does
      *          not have a background color,
      *          the background color of its parent is returned
      * @see #setBackground
      * @since 1.0
      */
     @Transient
     public Color getBackground() {
-------
No *container* word in java doc, simply "the background color of its
parent".

The parent in terms of our api is "parent container".
    /**
     * Gets the parent of this component.
     * @return the parent container of this component
     * @since 1.0
     */
    public Container getParent() {
    }

What you are explaining is not clear from above docs. In general owner 
and parent have different meaning. Owner is only responsible for the 
owned window resources deallocation while the parent window controls 
appearance of its child windows. In java windows we have both roles 
combined. In native platforms they are separated.
Then why the background is not inherited and foreground and font should 
be inherited on Windows platform? Should Linux and OS X be aligned to 
this understanding?


Window.java 623:
     public Window(Window owner, GraphicsConfiguration gc) {
         this(gc);
         ownedInit(owner);
     }

     private void ownedInit(Window owner) {
         this.parent = owner;
-----------
Owned window *does* have a parent.

Window.java 3758:
     public Color getBackground() {
         return super.getBackground();
     }
------
The super is Component.getBackground(), so the same logic is used for
background color definition.

But the specification is different.


 > javadoc for Window.getBackground() simply states that it returns the
color of the window
It does, but the logic that defines this color takes parent's color in
case of null. No discrepancy.

Actually no, it was intentionally changed on windows in 96 for Window 
and since 99 for Dialog. On osx it works from the beginning. The only 
platform which works differently is linux.


 - If we state that the parent/owner color should be used will mean
that isBackgroundSet() for our window should return false. And the
owner color should be returned when its color was changed after our
window was shown. But I assume that if our window is shown the changes
in the owner color does not affect us.

My assumptions is that we should set it, if it was not set by the
user: At least even on windows we have such comments:
  WComponentPeer.java: // Set background color in C++, to avoid
inheriting a parent's color.
  WDialogPeer.java: //* Native create() peeks at target's background
and if it's null calls this method to arrage for default background to
be set on target.  Can't make the check in Java, since getBackground
will return owner's background if target has none set. */



But this may cause a compatibility issue as well as in the reversed 
case
(to not inherit all).
Looks like the issue is not as easy as it appeared.

--Semyon

On 1/19/2016 11:10 AM, Prem Balakrishnan wrote:

Hi Semyon,

Ideally what you said is right.

Behaviors are different in different platform, as follows:

*In Windows:*

*B*ackground color *Doesn’t* Inherit Parent’s Background Color.

Foreground color *Does* Inherit Parent’s Foreground Color.

Font *Does* inherit Parent Font.

*In Linux:*

Background color *Does* Inherit Parent’s Background Color(Before my
Fix)

Foreground color *Does* inherit parent’s Foreground Color.

Font *Does* inherit parent’s font.

*In Mac:*

Background color *Doesn’t* Inherit Parent’s Background Color

Foreground color *Doesn’t* Inherit Parent’s Foreground Color.

Font *Doesn’t* inherit parent’s font.

Decision to be taken what should be the default scenario and
appropriately fix others.

Regards,
Prem

*From:*Semyon Sadetsky
*Sent:* Tuesday, January 19, 2016 12:09 PM
*To:* Prem Balakrishnan; Ambarish Rapte; Sergey Bylokhov;
awt-dev@openjdk.java.net
*Subject:* Re: Review Request for 8057574: Inconsistent behavior for
setBackground (Windows/Linux)

On 1/19/2016 6:30 AM, Prem Balakrishnan wrote:

    Hi Semyon,

    In the case of Foreground color, Behavior is same in Windows and
    Linux platforms.

    (i.e., Child inherits parents Foreground color).

    And from my point of view the behavior can be retained.

That seems to me inconsistent when the background color is inherited
and the foreground is not. I think they both should be either
inherited either both not inherited.

--Semyon

Regards,

Prem

*From:* Semyon Sadetsky
*Sent:* Monday, January 18, 2016 2:30 PM
*To:* Prem Balakrishnan; Ambarish Rapte; Sergey Bylokhov;
awt-dev@openjdk.java.net <mailto:awt-dev@openjdk.java.net>
*Subject:* Re: Review Request for 8057574: Inconsistent behavior for
setBackground (Windows/Linux)

Hi Prem,

The updated version looks better.
Should we do the same for the foreground color?

--Semyon

On 1/13/2016 4:24 PM, Prem Balakrishnan wrote:

    Hi Semyon,

    The scenario which you mentioned, child Window with Frame/Dialog
    as parent *FAILED*.

    I have fixed the issue and updated relative test cases .

    *Updated Webrev:
*<http://cr.openjdk.java.net/%7Earapte/prem/8057574/webrev.02/>http://cr.openjdk.java.net/~arapte/prem/8057574/webrev.02/ 


    Regards,

    Prem

    *From:* Semyon Sadetsky
    *Sent:* Wednesday, January 13, 2016 4:13 PM
    *To:* Prem Balakrishnan; Ambarish Rapte; Sergey Bylokhov;
awt-dev@openjdk.java.net <mailto:awt-dev@openjdk.java.net>
    *Subject:* Re: Review Request for 8057574: Inconsistent behavior
    for setBackground (Windows/Linux)


    On 1/13/2016 1:24 PM, Prem Balakrishnan wrote:

    Hi Semyon,

    Yup tried with Window:

    Dialog checks parent’s type internally, if the type is Window
    following exception is thrown.

    I meant *child* Window with Frame or Dialog parent.

    Exception in thread "main" java.lang.IllegalArgumentException:
    Wrong parent window

    *Code snippet  from awt/Dialog.java *

     public Dialog(Window owner, String title, ModalityType
    modalityType) {

            super(owner);

           if ((owner != null) &&

                !(owner instanceof Frame) &&

                !(owner instanceof Dialog))

            {

                throw new IllegalArgumentException("Wrong parent
window");

            }

    …….

    …….

    Regards,
    Prem

    *From:* Semyon Sadetsky
    *Sent:* Wednesday, January 13, 2016 3:40 PM
    *To:* Prem Balakrishnan; Ambarish Rapte; Sergey Bylokhov;
awt-dev@openjdk.java.net <mailto:awt-dev@openjdk.java.net>
    *Subject:* Re: Review Request for 8057574: Inconsistent behavior
    for setBackground (Windows/Linux)

    Hi,

    On 1/13/2016 12:20 PM, Prem Balakrishnan wrote:

        Hi Semyon,

        1.       Patch is updated with the relative test for the fix.

        *Updated Webrev:*
<http://cr.openjdk.java.net/%7Earapte/prem/8057574/webrev.01/>http://cr.openjdk.java.net/~arapte/prem/8057574/webrev.01/ 


        *Bug:*
<https://bugs.openjdk.java.net/browse/JDK-8057574>https://bugs.openjdk.java.net/browse/JDK-8057574 



        2.       The issue is observed when Frame is parent and 
Dialog
        is Child. (Scenario reported by Alexander Stepanov)

        The above issue is resolved with the same fix.

    Did you try Window child?

    --Semyon


    Test related to this scenario is also updated with the patch.

    Regards,

    Prem

    *From:* Semyon Sadetsky
    *Sent:* Monday, January 11, 2016 6:00 PM
    *To:* Prem Balakrishnan; Ambarish Rapte;
<mailto:awt-dev@openjdk.java.net>awt-dev@openjdk.java.net
    *Subject:* Re: Review Request for 8057574: Inconsistent behavior
    for setBackground (Windows/Linux)

    On 1/11/2016 2:18 PM, Prem Balakrishnan wrote:



    Hi Semyon,

    Thanks for the review comments.

    1.       I will add a relative test for the fix.

    2.       The same issue is observed with Frame also,

            As the original issue is reported for dialog, fix
    submitted only for dialog.

    Should I combine  fix for  Dialog and Frame in the same patch?

    Yes. I would consider it as the same issue. All window types may
    have this issue.

    --Semyon



    Regards,

    Prem

    *From:* Semyon Sadetsky
    *Sent:* Monday, January 11, 2016 3:32 PM
    *To:* Prem Balakrishnan; Ambarish Rapte;
<mailto:awt-dev@openjdk.java.net>awt-dev@openjdk.java.net
    *Subject:* Re: Review Request for 8057574: Inconsistent behavior
    for setBackground (Windows/Linux)

    Hi Prem,

    By default a fix requires a regression test which fails before 
the
    fix an passes after. When it is absent noreg-??? label should be
    set to the JIRA ticket. See
<http://openjdk.java.net/guide/changePlanning.html>http://openjdk.java.net/guide/changePlanning.html 

    for possible noreg suffixes.
    But it seems to me that reg test can be written for the issue.

    In JIRA Alexander Stepanov noted that the issue takes place for
    Frame as well but your solution is fixing the Dialog only. Did 
you
    run the scenario with Frame?

    --Semyon

    On 1/4/2016 11:28 AM, Prem Balakrishnan wrote:

        Hi,

        Please review fix for JDK9,

        *Bug:*
<https://bugs.openjdk.java.net/browse/JDK-8057574>https://bugs.openjdk.java.net/browse/JDK-8057574 



        *Webrev:*
<http://cr.openjdk.java.net/%7Earapte/prem/8057574/webrev.00/>http://cr.openjdk.java.net/~arapte/prem/8057574/webrev.00/ 


        *Issue:*

        It is Platform specific Issue - LINUX

        If Child Dialog’s Background color is NOT set explicitly,
        Child Dialog inherits Parent Dialog’s Background Color.

        (In Windows OS Child Dialog is NOT inheriting parent Dialog’s
        Background Color)

        *Cause:*

        Default Background color for Dialog is not initialized.

        *Fix:*

        If Background color is not set explicitly , default 
Background
        color is set at the time of Dialog’s initialization.

        Thanks,

        Prem