[Xquartz-dev] Line drawing in 1.5 and 1.6 servers

[Jeremy Huddleston]

Ok, I think I found the cause and a hackish workaround for this issue...

The problem stems from the fact that the core/common X server code no  
longer has hooks for paintWindowBorder and paintWindowBackground.

We were using those hooks for correcting alpha values as well as  
handling cases like this.  After the core X developers nuked those  
callbacks a while back, I went through and fixed hacked the alpha  
values back into place, but I missed this particular problem.

I have a bit of a hack to get functionality back in place, but it's  
not elegant.  I'm going to try finding a better solution for all of  
this.  I essentially resurrected some code that had been punted and  
found a place to put it into the pipe, but my guess is that there's a  
better location for it.

--Jeremy


----


diff --git a/mi/miexpose.c b/mi/miexpose.c
index 961c35c..d540a9f 100644
--- a/mi/miexpose.c
+++ b/mi/miexpose.c
@@ -531,6 +531,14 @@ miWindowExposures(pWin, prgn, other_exposed)
  	REGION_DESTROY( pWin->drawable.pScreen, exposures);
  }

+#ifdef ROOTLESS
+/* Ugly, ugly, but we lost our hooks into miPaintWindow... =/ */
+void RootlessSetPixmapOfAncestors(WindowPtr pWin);
+void RootlessStartDrawing(WindowPtr pWin);
+void RootlessDamageRegion(WindowPtr pWin, RegionPtr prgn);
+Bool IsFramedWindow(WindowPtr pWin);
+#endif
+
  _X_EXPORT void
  miPaintWindow(WindowPtr pWin, RegionPtr prgn, int what)
  {
@@ -556,6 +564,19 @@ miPaintWindow(WindowPtr pWin, RegionPtr prgn, int  
what)
      Bool	solid = TRUE;
      DrawablePtr	drawable = &pWin->drawable;

+#ifdef ROOTLESS
+    if(IsFramedWindow(pWin)) {
+        RootlessStartDrawing(pWin);
+        RootlessDamageRegion(pWin, prgn);
+
+        if(pWin->backgroundState == ParentRelative) {
+            if((what == PW_BACKGROUND) ||
+               (what == PW_BORDER && !pWin->borderIsPixel))
+                RootlessSetPixmapOfAncestors(pWin);
+        }
+    }
+#endif
+
      if (what == PW_BACKGROUND)
      {
  	while (pWin->backgroundState == ParentRelative)
diff --git a/miext/rootless/rootlessWindow.c b/miext/rootless/ 
rootlessWindow.c
index 3787e7e..2e5e84c 100644
--- a/miext/rootless/rootlessWindow.c
+++ b/miext/rootless/rootlessWindow.c
@@ -1692,3 +1692,29 @@ RootlessShowAllWindows (void)
          RootlessScreenExpose (pScreen);
      }
  }
+
+/*
+ * SetPixmapOfAncestors
+ *  Set the Pixmaps on all ParentRelative windows up the ancestor  
chain.
+ */
+void
+RootlessSetPixmapOfAncestors(WindowPtr pWin)
+{
+    ScreenPtr pScreen = pWin->drawable.pScreen;
+    WindowPtr topWin = TopLevelParent(pWin);
+    RootlessWindowRec *topWinRec = WINREC(topWin);
+
+    while (pWin->backgroundState == ParentRelative) {
+        if (pWin == topWin) {
+            // disallow ParentRelative background state on top level
+            XID pixel = 0;
+            ChangeWindowAttributes(pWin, CWBackPixel, &pixel,  
serverClient);
+            RL_DEBUG_MSG("Cleared ParentRelative on 0x%x.\n", pWin);
+            break;
+        }
+
+        pWin = pWin->parent;
+        pScreen->SetWindowPixmap(pWin, topWinRec->pixmap);
+    }
+}
+


On Sep 24, 2009, at 13:38, Martin Otte wrote:

> Hi,
>
> Attached is a simplified test case showing that in a call to  
> XCreateSimpleWindow, the border around the subwindow is not  
> displayed. With the 1.4 server, the program displays:
>
> <server_1.4.png>
>
>
> but with the 1.5 server, the border of the inner window is not shown:
>
> <server_1.5.png>
>
>
>
> I also added this example code to bug report #290.
>
> Hope this helps,
> Martin
>
>
> <border.c>
>
>
>
>
> On Sep 18, 2009, at 12:49 AM, Jeremy Huddleston wrote:
>
>>
>> On Sep 17, 2009, at 11:48, Martin Otte wrote:
>>
>>> Hi,
>>>
>>> The problem with the lines not drawn correctly occurs when calling  
>>> XCreateSimpleWindow. The program I have uses XCreateSimpleWindow  
>>> to create the buttons and their borders. When I increase the size  
>>> of the border in XCreateSimpleWindow and turn off all other  
>>> decorations, this is what I see:
>>>
>>> <border.png>
>>>
>>> It is interesting to see part of the quartz-wm frame imposed on  
>>> the button borders. My guess is that the border pixmap is not  
>>> being created/shown correctly in the server. Looking through dix/ 
>>> window.c and mi/miexpose.c a lot has changed between the 1.4 and  
>>> 1.5 versions of the server, so I don't think I'll be much help in  
>>> tracking down the problem.
>>
>> My guess is that the changes have just resulted in nothing actually  
>> being drawn.  What you're seeing in the background there is  
>> actually just the uninitialized garbage that happened to be at the  
>> memory location allocated for that backing store which was  
>> previously used to draw a quartz-wm widget.
>>
>> Can you provide me with your simplified test case.  That'll be  
>> easier than dragging the xterm window resize around every time I  
>> want to debug this.
>>
>> Thanks,
>> Jeremy
>>
>>
>> _______________________________________________
>> Xquartz-dev mailing list
>> Xquartz-dev at lists.macosforge.org
>> http://lists.macosforge.org/mailman/listinfo.cgi/xquartz-dev
>
> _______________________________________________
> Xquartz-dev mailing list
> Xquartz-dev at lists.macosforge.org
> http://lists.macosforge.org/mailman/listinfo.cgi/xquartz-dev

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3333 bytes
Desc: not available
URL: <http://lists.macosforge.org/pipermail/xquartz-dev/attachments/20090927/4b9aa9dd/attachment.bin>