From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux ppc; en-US; rv:1.0.0+) Gecko/20020524 BuildID: 2002052323 The page at the given URL doesn't display correctly: random blank horizontal spaces appear (i.e. some parts of the page are not visible). Reproducible: Always Steps to Reproduce: 1. Open the URL. 2. See what is displayed. 3. You may also scroll. Actual Results: Random parts (sometimes a thin horizontal line), mainly math formulas but also normal text, are not visible. This changes when scrolling. The page layout seems to be OK, though. Expected Results: The document should be correctly displayed.

Same problem with Mozilla 1.0.

-> changing component to MathML (otherwise it doesn't get appropriate visibility...) Got a screenshot?

After some tests, the problem seems to only occur when using a VNC server. Finally, perhaps a bug in the VNC implementation?

I've put some snapshots here: http://www.vinc17.org/download/mathml1.png http://www.vinc17.org/download/mathml2.png http://www.vinc17.org/download/mathml3.png http://www.vinc17.org/download/mathml4.png

Here's a new snapshot: http://www.vinc17.org/download/mathml5.png You can see that this is a redraw problem, but I don't know why it occurs on this page only. If I iconize the window and reopen it, nothing is redrawn in the part of the window corresponding to the document, i.e. I can see what's behind just like if the background were transparent. When launching ksnapshot, I got the message: Xlib: extension "RENDER" missing on display "ay:1.0". but nothing else wrong. I've attached the output of xdpyinfo in case Mozilla requires specific extensions.

@reporter Does it happens with newer builds? You don't need the RENDER extension of your X ... but you can enable it for KDE for faster output :-) See your XFree86.conf

I now use a Xft-enabled version and I no longer see any blank spaces problem. The characters (when they exist) are incorrectly positioned, but I suppose that this comes from bug 128153.

So ok ... discussion go over to bug 128153 and I'll close this bug as Invalide.

It wasn't invalid. There are two display problems: one probably due to bug 128153, and one that occurred in old versions of Mozilla (compiled without Xft support, but I now always compile with Xft support). Contrary to the former problem, the latter one (this bug) depended on how the page was redrawn (e.g. when scrolling).

I've recompiled Mozilla (latest nightly source + make -f client.mk) without Xft support. The bug is still there. So, I reopen the bug.

hmm, maybe duplicate of bug 190278? (that bug is not really limited to hebrew, contrary to what the summary says)

Vincent, do you use trunk or 1.4 branch? Is it a problem with Xft on ppc or also with Xft on x86?

No, it is not a dup of bug 190278 (which is related to Xft): I get the problem when I compile Mozilla *without* Xft support. So, it is *not* a problem with Xft in my case. I always use the trunk.

Now that I have a working configuration for Xft-enabled Mozilla, I can confirm that this bug doesn't occur *with* Xft.

Over to Gfx:Gtk, then.

gfx/src/gtk/ has been removed on trunk. Does this bug occur in a recent (cairo) trunk build?

There's still some blank (I get "a η b" instead of "a ≤ η ≤ b") and incorrect font display, but I don't know if this is the same bug. Tried with Mozilla/5.0 (X11; U; Linux ppc; en-US; rv:1.8.1.4) Gecko/20070508 Iceweasel/2.0.0.4 (Debian-2.0.0.4-0etch1). I'm going to attach a snapshot.

There are some troubleshooting tips and fonts at this URL that might help: http://www.mozilla.org/projects/mathml/fonts/

(In reply to comment #19) > There are some troubleshooting tips and fonts at this URL that might help: > http://www.mozilla.org/projects/mathml/fonts/ I did the font installation a long time ago (and I don't get the warning box saying that some fonts for MathML are missing). Also note that this is a build *with* Xft (about:buildconfig says that --enable-xft is provided), though the problem occurred without Xft. But I no longer use builds without Xft. The problem I mentioned in comment #17 may be another (new) bug.

As far as I can tell, both the --enable-xft and --disable-xft modes have been completely superseded by cairo and freetype in the current source base. The browser I'm typing this in (distro build of Firefox 8) renders the test page correctly. Closing.