82134 - Jpanel, JTabbedPane, Panel cannot be used to create a WebClient.

Status: RESOLVED WONTFIX

(Core Graveyard :: Java APIs to WebShell, defect)

Description

[Jeet Shahani]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; m18) Gecko/20010131
Netscape6/6.01
BuildID:

This is a very consistent bug.  This bug results in a developer not being able
to instantiate a JPanel which creates a BrowserCanvas and BrowserControl.  The
bug fails when doing a navigation.loadUrl.  This inturn delegates to
NavigationImpl.loadUrl().  It fails on the Assert statement: Assert.assert(-1 !=
nativeWebShell);  This is strange because it does create the BrowserCanvas,
which is not visible till loadurl is called.  This fails when trying to
instantiate other components also(JTabbedPane, Panel) that create a
browsercanvas.  However, it does work if the browsercanvas is instantiated in
JFrame and Frame instead.  I am using jre1.3.0_01 in Win95 and WebClient 1.0. I
can submit proper code which will result in the above.  Thanks.

Reproducible: Always
Steps to Reproduce:
1)Create a Frame or a JFrame.
2)Extend a JPanel.
3)Instantiate a browserCanvas in JPanel using queryinterface method.
4)Assert works.
5)calling the navigation.loadUrl fails (at NavigationImpl.loadUrl())
6)It fails on the line where it does.. Assert.assert(-1 != nativeWebShell);

This happens when adding with Panel and JTabbedPane too.


Actual Results:  WebClient never viewable.

Expected Results:  To be able to instantiate the JPanel this way..

This is a very consistent bug.  This bug results in a developer not being able
to instantiate a JPanel which creates a BrowserCanvas and BrowserControl.
The bug fails when doing a navigation.loadUrl.  This inturn delegates to
NavigationImpl.loadUrl().  It fails on the Assert statement: Assert.assert(-1 !=
nativeWebShell);  This is strange because it does create the BrowserCanvas,
which is not visible till loadurl is called.  This fails when trying to
instantiate other components also(JTabbedPane, Panel) that create a
browsercanvas.  However, it does work if the browsercanvas is instantiated in
JFrame and Frame instead.  I am using jre1.3.0_01 in Win95 and WebClient 1.0. I
can submit proper code which will result in the above.  Thanks.

Comment 1

[Avi Drissman]

Attachment: Illustration of problems 1 and 2 above

Comment 2

[Avi Drissman]

Whoops, wrong bug for this attachment :-(

Comment 3

[edburns]

Jeet, can you please attach some code to reproduce this?  I'm very interested 
in fixing this bug.

Comment 4

[Jeet Shahani]

Attachment: To Run: java org.mozilla.webclient.test.MyEMJP <PATH TO MOZILLA1.8 or NETSCAPE 6.0>

Comment 5

[Jeet Shahani]

Here is the stack trace.  BTW, this error also occurs when you add the
browserCanvas to an internalframe.  Is there a workaround, uncommenting the
assertion or something??

C:\WebClient\java\webclient\classes_spec>java org.mozilla.webclient.test.MyEMJP
%MOZILLA_FIVE_HOME%
url 1..http://www.yahoo.com
constructed with binDir: C:\PROGRA~1\Netscape\NETSCA~1 url: http://www.yahoo.com

in BrowserControlCanvas setBounds: x = 0 y = 0 w = 640 h = 480
native library does implement webclient.WindowControl
Can't setBounds(java.awt.Rectangle[x=0,y=0,width=639,height=479]) null
native library does implement webclient.Navigation
org.mozilla.util.AssertionFailureException
native library does implement webclient.EventRegistration
org.mozilla.util.AssertionFailureException
Exception in thread "main" org.mozilla.util.AssertionFailureException
       at org.mozilla.util.Assert.assert(Assert.java:122)
       at org.mozilla.webclient.wrapper_native.NavigationImpl.loadURL(Navigatio
nImpl.java:83)
       at org.mozilla.webclient.test.MyEMJP.<init>(MyEMJP.java:191)
       at org.mozilla.webclient.test.MyEMJP.main(MyEMJP.java:773)

Comment 6

[Jeet Shahani]

Hi Ed,

 The JPanel fails in the NavigationImpl.loadURL method on the
Assert statement below.  Hope this helps.

public void loadURL(String absoluteURL)
{
   ParameterCheck.nonNull(absoluteURL);
   myFactory.throwExceptionIfNotInitialized();
   Assert.assert(-1 != nativeWebShell);

   synchronized(myBrowserControl) {
       nativeLoadURL(nativeWebShell, absoluteURL);
   }
}

Comment 7

[Jeet Shahani]

Hi Ed,

 The JPanel fails in the NavigationImpl.loadURL method on the
Assert statement below.  Hope this helps.

public void loadURL(String absoluteURL)
{
   ParameterCheck.nonNull(absoluteURL);
   myFactory.throwExceptionIfNotInitialized();
   Assert.assert(-1 != nativeWebShell);

   synchronized(myBrowserControl) {
       nativeLoadURL(nativeWebShell, absoluteURL);
   }
}

So that I know, in future, is it better to post the code here in bugzilla as
opposed to on the newsgroups.  Thanks.

Comment 8

[edburns]

Ed Burns

Comment 9

[Jonadab the Unsightly One (Nathan Eady)]

Too late for 0.9.6, this needs retargeting.

Comment 10

[Asa Dotzler [:asa]]

0.9.6 is long gone. -> 0.9.7

Comment 11

[R.K.Aa.]

0.9.7 is long gone as well..

Comment 12

[Jerome Lacoste]

attachment 35609 [details] should be marked obsolete as per comment #2

Comment 13

[edburns]

I'm looking at this now.

Comment 14

[John Eastman]

I found another issue when placing the BrowserControlCanvas in a JSplitPane 
here the size of the pane that holds the BrowserControlCanvas may be enlarged by
resizing the JSplitPane's divider, however the pane cannot be made smaller.

Investigating this, I found that JSplitPane is relying on the minimum size of
the component in the pane that is being resized to determine if the pane can be
made smaller. The minimum size of the BrowserControlCanvas is being returned as
the current size of the canvas, which causes the JSplitPane not to be able to
shrink the pane.

Investigating further, I found that this is due to the way java.awt.Component
determines the minimum size of a component. If the Component does not have a
peer component, which in the case of the BrowserControlCanvas it does not, then
the [current] size of the component is used.

To make things work for me, I added some code to BrowserControlCanvas to manage
the minimum size:
-- an attribute to hold the minimum size, type java.awt.Dimension
-- public void setMinimumSize() to set the attribute
-- public Dimension getMinimumSize(), overriding java.awt.Component to return
the value.

This allows me to set the minimum size of the BrowserControlCanvas from the
class that is inserting it into the JScrollPane. This change requires some work
to be done by the client of BrowserControlCanvas, setting the minimum size, but
it makes no assumptions about the minimum size if it is not explicitly set,
maintaining the previous behavior. 

As a side-effect to making this change, I am now able to add the
BrowserControlCanvas to a Panel (or JPanel) and it works, which I think
addresses this bug.

Comment 15

[Brian Johnson]

I'm also having this problem using Webclient 1.3
The problem seems to be that for some reason the Event queue is never created.
I get these messages when I'm running the program:
     [java] +++ In appInitialize - after createWrapperFactory +++ 

     [java] in BrowserControlCanvas setBounds: x = 0 y = 0 w = 800 h = 600
     [java] native library does implement webclient.WindowControl
     [java] Can't setBounds(java.awt.Rectangle[x=0,y=0,width=799,height=599]) null
     [java] in BrowserControlCanvas setBounds: x = 5 y = 25 w = -10 h = -30
     [java] Webclient-Gdk-ERROR **: BadMatch (invalid parameter attributes)
     [java]   serial 2187 error_code 8 request_code 42 minor_code 0
     [java] in BrowserControlCanvas setBounds: x = 0 y = 20 w = 11 h = 6
     [java] in BrowserControlCanvas setBounds: x = 0 y = 20 w = 11 h = 6
     [java] in BrowserControlCanvas setBounds: x = 0 y = 20 w = 402 h = 200

Then while the program is exiting the event queue seems to start just before the
program finishes shutting-down:
      [java] (QFA)Talkback error: Can't initialize.


     [java] Creating Event Queue 

     [java] InitMozillaStuff(8496258): Create the action queue
     [java] Init the baseWindow
     [java] Create the BaseWindow...
     [java] Creation Done.....
     [java] Show the webBrowser

Comment 16

[edburns]

Brian: are you using the stock test browser or your own one?  If so, could you
post a code snippet that would allow me to reproduce it?

Ed

Comment 17

[edburns]

Attachment: Test to reproduce this bug

Does this patch replicate the problem?	I can't replicate the problem with the
latest code.

Comment 18

[Brian Johnson]

Attachment: Test code to reproduce problem

Comment 19

[Brian Johnson]

Ed, your patch did not reproduce the problem, I sent an attachment that does.

OS should be All since the original bug was in Win98 and I reproduced under Linux.

This is using Webclient 1.3

Comment 20

[Brian Johnson]

Ok, this is _not_ JPanel dependant.
What I was doing wrong (and I assume this was the original reporter's problem as
well due to the same symtoms) was calling
browserControl.queryInterface(BrowserControl.NAVIGATION_NAME); before the window
was displayed.
I've done some further testing and determined that if you call
browserControl.queryInterface(BrowserControl.NAVIGATION_NAME); before the Canvas
is displayed then the webclient will not be usable. Even if you do another call
to browserControl.queryInterface(BrowserControl.NAVIGATION_NAME); after it's
visible the navigation still wont work.
I imagine that other browserControl.queryInterface() options might cause the
same issue, but I haven't tested this.
I now initialize my navigation variable with a null value and initialize it if
it's null immediatly before I go to use it, which is always after it's been
displayed.

Comment 21

[Simon Paquet [:sipaq]]

Does not qualify as a blocker.

Comment 22

[Sylvestre Ledru [:Sylvestre]]

Java APIs to WebShell isn't a thing anymore. Closing.