50403 - Spanish charactor composition broke do to nsWidget problem in Gtk Mozilla embeded apps

Status: RESOLVED DUPLICATE of bug 50130

(Core :: XUL, defect, P3)

Description

[rusty.lynch]

Inside embeded gtk mozilla apps (like TestGtk) spanish charactor composition 
does not work.  The problem lies in nsWidget.cpp's GetShellWidget() function.  
In several places an attempt is made to get mIMEShellWidget using 
GetShellWidget, but GetShellWidget() returns null in all cases since it is 
working under the assumption that the toplevel window of the app is an nsWidget 
(which is not the case for an embeded mozilla app).

The is a loop in GetShellWidget() that walks up the window tree to find an 
nsWidget that has it's mIsTopLevel member set to true.  I can change to loop to 
just find top most window and spanish composition works fine.  Although, I 
really don't understand what we are really tring to do in this code so I fear 
some bad side affects will come of my hack.

Comment 1

[Peter Trudelle]

->pavlov

Comment 2

[rusty.lynch]

An easy way to reproduce this bug is to just run TestGtk (using a debug build)
and visit yahoo.  When you type into the search text input field you will notice
a bunch of assertions printed to stderr for:

###!!! ASSERTION: GetShellWidget() fails: 'mIMEShellWidget', file nsWidget.cpp,
line 2843
###!!! Break: at file nsWidget.cpp, line 2843

If you are just typing in english which doesn't have to do any funky key
compination stuff then all is well.  I know the spanish key compinations don't
work and I sure that all other languages will be broke.

So now I am wondering if it should be the job of the embeding application to
mark the top most nsWindow as toplevel.  So for example, gtk_moz_embed_new()
would not only create the new embeded mozilla widget, but also find the top gdk
window, walk down to the eventual nsWindow and then set mIsTopLevel to true.

Comment 3

[Stuart Parmenter]

-> tajima

Comment 4

[Masaki Katakai]

This problem is dup of http://bugzilla.mozilla.org/show_bug.cgi?id=50130.

Comment 5

[Masaki Katakai]

Dup of bug 50130. I've checked in the fix right now as bug 53989.

Toshi, can you mark this as DUP of bug 50130?

[e-mail], can you try on tomorrow's build?

Comment 6

[Hidetoshi Tajima]

reassign.

Comment 7

[Masaki Katakai]

bug 50130 has been fixed. marked as dup.


*** This bug has been marked as a duplicate of 50130 ***