From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4b) Gecko/20030509 Description of problem: If one creates a webpage which contains both english and armenian characters ( in UTF-8) then if the subpixel smoothing is enabled mozilla sometimes fails to display the armenian letters. Without subpixel smoothing it _always_ displays them properly. to illustrate the problem I am attaching a simple test.html file, and 2 screenshots of mozilla displaying that file: with "best shapes" enabled and "subpixel smoothing" enabled. Version-Release number of selected component (if applicable): mozilla-1.2.1 and mozilla-1.4b How reproducible: Always Steps to Reproduce: 1. make sure you have an LCD 2. in fonts properties enable "Best Shapes" 3. open the attached test.html file in mozilla browser. this is how that file is normally supposed to look. (look at best_shapes.png ) 4. in fonts properties enable "Subpixel Smooting" 5. restart mozilla, and open test.html again. the armenian part of the sentence is missing ( subpixel.png )
Created attachment 91947 [details] a simple test file which illustrates the bug
Created attachment 91948 [details] mozilla rendering the test file with "Best Shapes" enabled
Created attachment 91949 [details] mozilla rendering the test file with "Subpixel Smooting" turned on
What happens if you highlight the selection? Are the characters drawn? Try moving back and forth across the text selecting it.
I can see the text when it is selected, but then it disappears again after I diselect it.
It's actually a dup of an upstream bug in xft/fontconfig that's in mozilla's bugtracker: http://bugzilla.mozilla.org/show_bug.cgi?id=190278
this bugs has been fixed in XFree86-4.3.0-22.1
Unfortunately, the fix has caused a number of other problems for us, and may have to be removed if they can't be resolved soon. I've wasted a lot of time trying to resolve weird Xft/Xrender build failures and am lost in a maze of libtool/autoconf problems. If I can't fix it soon, I may remove the new Xft/Xrender, at which point this bug will return. See bug #103319 for details if anyone is interested in helping determine the proper solution for the build failure.