Closed Bug 78741 Opened 24 years ago Closed 24 years ago

Full page plugins completely broken

Categories

(Core Graveyard :: Plug-ins, defect, P1)

defect

Tracking

(Not tracked)

VERIFIED FIXED
mozilla0.9.1

People

(Reporter: shrir, Assigned: peterlubczynski-bugs)

References

()

Details

(Keywords: regression, Whiteboard: NEED SR, critical for mozilla 0.9.1)

Attachments

(1 file)

seen on 20010502 trunk on mac, yet to confirm on windows

Installed acrobat reader 5.0, copied the plugin over to 6.x trunk plugins folder
Went to the above page
Acrobat reader splashscreen comes up, nothing laods int he browser(except for a 
scrollbar and a stray icon on top left)
keyword:acrobat
Keywords: acrobat
Summary: blank page laods, plugin does not work → blank page laods, acrobat plugin does not work
Shrirang isolated this to a regression to builds around 2001050104. Checking 
CVS log for nsPluginViewer.cpp shows Hyatt checking in his fix for bug 76495.

My guess at just looking at the code, seems like this is the problem:
NS_IMETHODIMP                                                                   
PluginViewerImpl::Validate()                                                    
{                                                                               
 if (mWindow)
   mWindow->Validate();
)
Status: NEW → ASSIGNED
Keywords: regression
OS: Mac System 9.x → All
Priority: -- → P1
Hardware: PC → All
Target Milestone: --- → mozilla0.9.1
cc:ing Hyatt

Can you take a quick look. I think it's a regression from bug 76495?
Blocks: 54689
This is much more serious than I had thought. Commenting out those lines does 
not solve the problem. This happens with all full page plugins on all platforms. 
 I still think it's from Hyatt's check-in.

Arun: Please don't give Adobe any builds (and perhaps other vendors too) untill 
this gets fixed!!

Looking at Bonsai, it doesn't look like anything else checked into plugins at 
that time could have caused this.
http://bonsai.mozilla.org/cvsquery.cgi?branch=HEAD&file=mozilla/modules/plugin/n
glsrc/&date=week
Blocks: 74980
Summary: blank page laods, acrobat plugin does not work → Full page plugins completely broken
Blocks: 47712
Blocks: 76932
Blocks: 59749
Blocks: 53363
Yeah, this sounds like it's mine.
Assignee: peterlubczynski → hyatt
Status: ASSIGNED → NEW
Attached patch Patch to show window. (deleted) — Splinter Review
This patch basically undoes what I did... namely hiding the window and not 
showing it.  However, the full-page plugin still doesn't work.  The cursor 
changes and the background is right but the foreground doesn't paint.  Not sure 
if this is my bug now after making the patch.

peterl, have any ideas?  Try my patch out and see what it does for you.
Status: NEW → ASSIGNED
Oh Hyatt, this is great! It works for me on Win32. Try another testcase like: 
http://slip.mcom.com/shrir/License.pdf and make sure you have Acrobat installed 
correctly.

Let me try Mac.
Patch does not work on Mac. Some stuff gets painted initially then 
everything disapears right affer you rezsize the window.
Blocks: 76892
>Arun: Please don't give Adobe any builds (and perhaps other vendors too) untill 
>this gets fixed!!

IMHO, this is equal to a 0.9 nomination. A browser that does not support acrobat
is plain useless for business use.

Bernd, this bug probably doesn't exist on the 0.9 branch, since the optimization
which caused this regression went in on the trunk after we branched.
Whiteboard: critical for mozilla 0.9.1
*** Bug 79256 has been marked as a duplicate of this bug. ***
Hyatt, can you check-in what you have for now? It seems to work for me on 
Windows okay but this bug should be left open as it does not work on Mac.
*** Bug 79945 has been marked as a duplicate of this bug. ***
*** Bug 79984 has been marked as a duplicate of this bug. ***
Will this fix correct the failure to launch Acrobat after saving file when there
is no plug-in installed?
Keywords: patch
r=saari
Whiteboard: critical for mozilla 0.9.1 → NEED SR, critical for mozilla 0.9.1
sr=ben@netscape.com
Ok, my fix is in.  I'm reassigning to you, peterl, to investigate the Mac 
problem.  I'm happy to give you as many cycles of my time as you need to help 
figure out why the Mac doesn't respond to the fix.

One theory I have is that on Mac only maybe we're leaking the document viewer 
of the previous page, which causes the previous viewer widget not to be 
destroyed (and so it might paint on top of the widget that holds the plugin).

A possible patch to see if this is what's going on would be to call Hide() on 
the viewer passed in to SetPreviousViewer (if the previous viewer passed in is 
non-null).  If this fixes the problem, then it means we have a leak on the Mac 
only.

Assignee: hyatt → peterlubczynski
Status: ASSIGNED → NEW
That's strange. Seem to be working much better on my Mac today than when I tried
the patch last week. I'll check this out with Monday morning's commercial bits.
Marking FIXED..please open if you don't think so.
Status: NEW → RESOLVED
Closed: 24 years ago
Resolution: --- → FIXED
Verified this is fixed. Testcase url laods and also tried a few other 
plugins/tests. (0516)
Status: RESOLVED → VERIFIED
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: