129893 - Cocoa changes to GFX/layout/view

Status: RESOLVED FIXED

(SeaMonkey :: General, defect)

Description

[David Hyatt]

This bug tracks changes to GFX/layout/view for Cocoa.

Comment 1

[David Hyatt]

These changes keep Chimera from crashing, turn off double buffering for 
our views, and fix the caret blinking problem in Chimera.

Comment 2

[David Hyatt]

Attachment: Ignore the silly WS stuff.

Comment 3

[David Hyatt]

Attachment: Better patch.

Comment 4

[David Hyatt]

Ok, since this got + (yay), can I get r/sr here?  Pink? Simon? Saari?

Comment 5

[Mike Pinkerton (not reading bugmail)]

+#ifdef MOZ_WIDGET_COCOA
     WindowPtr firstWindow = GetTheWindowList();
     if (firstWindow)
       ::SetGWorld(::GetWindowPort(firstWindow), ::GetMainDevice());
+#else
+    static GrafPtr gGoodPort = nsnull;
+    if (!gGoodPort) {
+      Rect macRect;
+      ::SetRect(&macRect, 0, 0, 2, 2);
+      ::NewGWorld(&gGoodPort, 16, &macRect, nsnull, nsnull, 0);
+    }
+    ::SetGWorld(gGoodPort, nsnull);
+#endif

isn't this backwards? why is the _existing_ code wrapped in MOZ_WIDGET_COCOA?

+#ifndef MOZ_WIDGET_COCOA
        ::SetOrigin(-mGS->mOffx, -mGS->mOffy);          // line order...
+#endif

can you explain this with a comment in the code (and here as well)?

+#ifdef MOZ_WIDGET_COCOA
+   // Emulate the Mac IE behavior of scrolling 2 lines instead of 1
+   // on a button press.  This makes scrolling appear smoother and
+   // keeps us competitive with IE.
+   increment *= 2;
+#endif

shouldn't we, um, try to find out why scrolling is so slow? is this a problem 
with mozilla as well as chimchim?

+++ view/src/nsView.cpp 11 Mar 2002 22:45:53 -0000
@@ -803,13 +803,13 @@
-  }
+    }

did a tab get introduced here by accident?

-//#define NO_DOUBLE_BUFFER
+#ifdef MOZ_WIDGET_COCOA
+#define NO_DOUBLE_BUFFER
+#endif

I'd leave the uncommented version there because a lot of people use that line to 
toggle DB on/off. It's what they're used to looking for.

Comment 6

[David Hyatt]

Hahahah, good grief.  No wonder that good port stuff was behaving so 
oddly.  I had it backwards!  I need to try that out now and see if I still need 
all the code in nsImageMac.cpp.

Comment 7

[David Hyatt]

Ok, so I fixed the ifdef to be #ifndef, and once I did that, I was able to 
simplify the code in nsImageMac to not have to ask IsValidPort (I was 
being an idiot).  All that remains in nsImageMac is the call to set the port 
to a known good port (which now works properly with the #ifndef fixup).

Seeking r/sr now with that modification and with leaving the commented 
out code in nsViewManager.cpp.

Comment 8

[Mike Pinkerton (not reading bugmail)]

Comment on attachment 73581 [details] [diff] [review]
Better patch.

I still want to see a comment on 

+#ifndef MOZ_WIDGET_COCOA
	::SetOrigin(-mGS->mOffx, -mGS->mOffy);		// line order...
+#endif

both in the code and here in the bug. just so i know what's going on.

Comment 9

[Mike Pinkerton (not reading bugmail)]

oops, forgot r=pink

Comment 10

[Ben Goodger (use ben at mozilla dot org for email)]

Comment on attachment 73581 [details] [diff] [review]
Better patch.

sr=[e-mail]

Comment 11

[Randell Jesup [:jesup] (needinfo me)]

Comment on attachment 73581 [details] [diff] [review]
Better patch.

a=[e-mail]

Please send email to drivers next time

Comment 12

[David Hyatt]

Fixed.

Comment 13

[Patrick C. Beard]

I checked the changes to

gfx/src/mac/nsGfxUtils.h
gfx/src/mac/nsImageMac.cpp
gfx/src/mac/nsRenderingContextMac.cpp
gfx/src/mac/nsRenderingContextMac.h

into CHIMERA_M1_0_BRANCH.