6872 - MRJ Java Applets can't create new windows on the Mac

Status: VERIFIED FIXED

(Core Graveyard :: Java: OJI, defect, P3)

Description

[cyrus]

Attempting to create a new AWT window, saw a frame, don't work. There are some log
messages about registering menus and windows, but no window shows up.

Comment 1

[Alex Musil]

Yep - this code needs to be hooked up.

Comment 2

[Alex Musil]

Reassigning all OJI bugs to [e-mail]

Comment 3

[mark.lin]

Reassigning bug to [e-mail]

Yuup. I'll probably investigate this next.

Comment 4

[chris hofmann]

m12

Comment 5

[chris hofmann]

m13

Comment 6

[chris hofmann]

move to m14. let me know if fixes are available. thx.

Comment 7

[modgock]

mark is dead. long live mark.

Comment 8

[modgock]

moved to assigned to get bugzilla mailing to daemon to shut up.  moved milestone 
to m17 per new sun deployment policy.

Comment 9

[modgock]

reassigning to loki-sun

Comment 10

[loki der quaeler]

the dependency was closed because the reporter (who's quit) hadn't provided more 
information.. investigating whether 14413 can really be closed. 

Comment 11

[loki der quaeler]

moving off of "NEW" to assigned.

Comment 12

[Andrew Thompson]

*** Bug 56941 has been marked as a duplicate of this bug. ***

Comment 13

[edburns]

Beard, can you take this?

Ed

Comment 14

[Andrew Thompson]

Added a couple of words to the summary to make it come up in Java and MRJ 
searches, so people have a chance of finding it rather than filing duplicates!

Also, loki left Sun months ago. Assigning to default component owner

Comment 15

[mozilla]

I have posted a small applet demonstrating this problem at:
http://segal.org/FrameTest/
The problem occurs in Netscape 6.0 for Macintosh but not in Netscape 4.73 with 
the MRJPlugin.  Source code for the applet is as follows:

import java.applet.*;
import java.awt.*;

public class FrameTest extends Applet {

public void init() 
{
    Frame frame = new Frame("A frame");
    frame.reshape(100, 100, 300, 300); 
    frame.validate();
    frame.show();
}

public void paint(Graphics g)
{
    g.drawString("A very simple frame should pop up", 10 , 20);
}
} // END OF Class FrameTest

Comment 16

[shrirang khanzode]

qa: junruh

Comment 17

[shrirang khanzode]

Is anyone looking at this bug ? As of today, it still exists on the mac trunk 
and branch builds. I do not see the frame pop up with the testcase attached. 
This however works in 4.x and IE.

Comment 18

[mozilla]

I feel that this bug is really important so I am adding this comment to 
encourage someone to take ownership of it and work on it.  I think the bug 
should get high priority for the following reasons:
1.  This bug leaves a major JDK 1.0 class, Frame, unable to function, and makes 
Netscape 6 fail Java certification.
2.  Attempts to use Frame terminate an applet, providing a very bad user 
experience.
3.  Since many applets run in a Frame, this bug makes it difficult for 
programmers to use their existing applets to find other bugs, further slowing 
down the process of getting the MRJPlugin for Netscape 6 to work.
4.  This works fine with the Netscape 4.x MRJPlugin so it shouldn't be a major 
effort to get the same thing to work for Netscape 6.

Comment 19

[drapeau]

Yep, you win.  Somebody is on this one now; Ed Burns is out on vacation this
week and so did not update the bug report.  We have a new person starting on
Mac/OJI bugs; you'll see progress on this bug before too long.

Comment 20

[mcclain looney]

just fyi, frames work in 4.76

Comment 21

[edburns]

Hi Terry, I'm putting this in your list.

Comment 22

[Terry Noyes]

Here's diffs to two files to try to fix 6872.  It also seems to fix the Java
Console bugs: 21971, 27407, 68207, and 31670.  I'm using Mozilla 0.7, MacOS 9.1,
and based the changes on MRJPlugin 1.0b2, both today's and the original 1.0b2
check-in.

Same workaround as was necessary for 4.x.  Used Patrick's code to intercept
events and handle things ourselves.  Placed the necessary calls in
TopLevelFrame::showHide().  Re-enable the hack in MRJContext.cp which gives the
applet viewer extra time to go away before tossing our new windows.

Questions: do I select 'WorksForMe'?  Are the existing calls to
thePluginManager2->RegisterWindow/UnregisterWindow useful/dangerous/redundant?

Hmm.  Bugzilla is not allowing me submit attachments.  Here's the file:
http://homepage.mac.com/tnoyes/.Public/Sun/Diffs%20for%206872

Comment 23

[Terry Noyes]

Attachment: Diffs for 6872

Comment 24

[Andrew Thompson]

Hi Terry, 

Great to see someone getting some leverage on these bugs! Keep up the good work. 

I'm adding Mike Pinkerton and Simon Fraser to the cc: list as possible Mac 
reviewers. 

Terry, I don't know if you've already had an induction on this sort of stuff, but 
I figure its better to hear it twice than not at all. If you're looking to get 
reviews and have this stuff checked in (i.e. you're happy with the patch as it 
now stands) there are a few things you can do to indicate this:

1/ Send mail to [e-mail] describing the changes and requesting 
reviews and super reviews (cc: appropriate people as desired) 
2/ Add the keywords 'patch' and 'review' 
3/ If there are lots of attachments, specify which should be reviewed in the 
status whiteboard
4/ While you're making all those changes, may as well add a comment inviting 
people to come get it while its hot :)

That's what I do, but of course, the full process is described here:
http://www.mozilla.org/hacking/reviewers.html

Apologies if you knew all this already. Bit of pent up desire to get some of 
these fixed on my part, I suspect.

Welcome aboard!

Comment 25

[Terry Noyes]

Add keywords 'patch' and 'review'.  Should I make a compiled plug-in available
for people to play with?

Comment 26

[Patrick C. Beard]

A couple of comments on the patch:

1. Are you using spaces everywhere? I've started to bite the bullet and just use 
4-spaces instead of tabs. CodeWarrior Pro 5's IDE (v4.0.4) has an option for 
that.

2. I don't like all of the WORKAROUND_6872 comments. I'd prefer seeing the code 
in an #ifdef rather than surrounding it with these comments.

3. Perhaps all of this code should go in another source file, rather than being 
added directly to TopLevelFrame. It will make it easier to extract later when the 
browser gets fixed.

BTW, you should "accept" your bugs rather than leaving them as new, so I can tell 
you're working on them.

Comment 27

[Terry Noyes]

Ok, I'll go to spaces instead of tabs, but all the other files seem to use tabs.
 Put stuff into a separate .cpp and .h file (have to add .cpp to project or xml
file, of course).  Thanks for the tip, I'll ACCEPT bugs.

Comment 28

[Terry Noyes]

Attachment: Diffs and two new files for 6872, v2

Comment 29

[edburns]

Terry, have you asked for review on these?

Comment 30

[Terry Noyes]

Yes, on 2/19.  E-mailed [e-mail]

Comment 31

[Patrick C. Beard]

I'm reviewing again.

Comment 32

[Patrick C. Beard]

Attachment: Add #ifdef USE_ALT_WINDOW_HANDLING to make clear when we're using this hack.

Comment 33

[Patrick C. Beard]

With this new patch, r=beard.

Comment 34

[edburns]

Have you asked for sr=scc on this?

Comment 35

[Terry Noyes]

Yes, on 3/16

Comment 36

[Patrick C. Beard]

Simon would also be a good candiate for super review.

Comment 37

[Simon Fraser [no longer active]]

sr=sfraser

Comment 38

[Terry Noyes]

Attachment: The final patch, incl. the .xml chg and the two new files

Comment 39

[Terry Noyes]

Pending cvs id request approval...

Comment 40

[Terry Noyes]

April Update: no estimate on getting write access to the tree.  Anyone else is welcome to check in these changes.

For those interested, you can see when this patch lands by checking the cvs log for mrjplugin.xml,  http://bonsai.mozilla.org/cvslog.cgi?file=mozilla/plugin/oji/MRJ/plugin/MRJPlugin.xml

as that file will be updated.

Comment 41

[Mike Pinkerton (not reading bugmail)]

i can help land this once the tree opens and the waves upon waves of carpools 
clear up.

Comment 42

[Patrick C. Beard]

Going to check these changes in as soon as tree is open.

Comment 43

[Patrick C. Beard]

Checked in changes.

Comment 44

[Terry Noyes]

Thanks, Patrick.  Just did a clean pull, build, and test and believe that
bug #6872, bug #21971, bug #27407, bug #68207, bug #31670, and bug #69352
have been addressed.  Time for new bug reports to roll in.

Comment 45

[Terry Noyes]

The fix for bug #6872 landed yesterday, and I've tested successfully with a
yesterday build and a today build. Changing to FIXED. (The patch fixed bug
#6872, bug #21971, bug #27407, bug #68207, bug #31670, bug #69352, bug #72585,
and bug #74189)

Comment 46

[John Unruh]

Verified with the commerical trunk 200140608 Mac build.