Shadow behind transparent chat window isn't updated

[yorn]

I installed Safari 3 beta, got hit by #7164 so I compiled from SVN (since the bug is already fixed there). Now it occured to me that transparency works fine now, but the shadow isn't updated at message sending or retrieval. Forcing the shadow to be repainted (by resizing the window or taking it's focus) fixes the problem, but it occurs again when a new message is sent/received. You can see this effect on the attached screenshot; See the message "whoa this message is ghosted! :o", a few pixels below you see the same message "ghosting". This happend because the text was there before the message "I bet this should not happen" appeared and moved it up. The "ghosting" happens to both text and images.

I believe this can be fixed by forcing the shadow to be repainted everytime something changes in the message area

[yorn]

The ghosting effect in action

[ccsccs7]

Ahh. It's a little hard to see if your LCD screen brightness isn't turned up high enough.

[jchang]

i'm seeing this as well.

[catfish_man]

(In [20840]) Invalidate the window shadow when using a transparent background. Fixes #7769

[jchang]

i'm still seeing this in 1.1.2

[yorn]

Well sorry, the bug isn't entirely fixed, shadow is updated after sending/receiving messages, but not after changing the tab..

[bigDee]

email me at bigDee2007@… if there is ever a fix for this

[yorn]

I just figured that the shadow also isn't updated after scrolling, and when typing very long messages (the input box gets higher, thus moving the text up, but not the shadow)

[cbarrett]

It's kind of ridiculous that we have to invalidate the shadow ourselves. "Perf win" for webkit? Perhaps. Annoying for us? Definitely.

[evands]

The old solution overrode drawRect: and invalidated it; this was effective. It appears WebKit is no longer calling drawRect to draw a WebView :P A perusal of the source should be able to indicate the appropriate method to override.

[evands]

I can't figure out how WebView draws based on the source code.

[cbarrett]

If they're not using drawRect, they are probably just drawing "directly" to the CGContext for that view.

We should file a bug with Apple -- they should at least be calling the drawRect: for their WebView, even if their implementation does nothing.

[evands]

Would somebody please file a bug with the webkit team on this issue? I believe David talked with someone about it; a link to the ticket would be nice if one exists.

[zacw]

Closing all waiting on apple tickets. These just take up space, and are largely ignored anyway. Any functionality we need but cannot add due to apple is a good radar to file, but we are not apple's tracker.

[anonymous]

Milestone Waiting on Apple deleted