[Tcl-bugs] [ tktoolkit-Bugs-220783 ] [BITMAP config -maskdata $data] doesn't work

[SourceForge.net <no...@so...>]

Bugs item #220783, was opened at 2000-11-01 00:53
Message generated for change (Comment added) made by dkf
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=112997&aid=220783&group_id=12997

Category: 40. Bitmap Images
Group: = 8.0.3
Status: Open
Resolution: None
>Priority: 6
Submitted By: Brent B. Welch (welch)
Assigned to: Donal K. Fellows (dkf)
Summary: [BITMAP config -maskdata $data] doesn't work

Initial Comment:
OriginalBugID: 970 Bug
Version: 8.0.3
SubmitDate: '1998-12-15'
LastModified: '1999-08-15'
Severity: MED
Status: Assigned
Submitter: welch
ChangedBy: hobbs
OS: Linux-Red Hat
Machine: Other


Name:

D. Richard Hipp



CustomShell:

Plus/Dash patches



ReproducibleScript:

image create bitmap t1 -data $data -maskdata $data -background red

label .x -image t1

pack .x

update

t1 config -maskdata $NewMaskData







ObservedBehavior:

The new mask specified by the "configure" does not

take effect.







DesiredBehavior:

I expect the mask to change.  Here's a workaround:



    t1 config -maskdata {}

    t1 config -maskdata $NewMaskData



The problem results from a unfortunate interaction between the

XID allocator and the Tk_GetGC allocator.







Patch:

*** tkImgBmap.c.orig    Mon Dec  7 17:45:41 1998

--- tkImgBmap.c Mon Dec  7 18:11:49 1998

***************

*** 353,358 ****

--- 353,359 ----

      XGCValues gcValues;

      GC gc;

      unsigned int mask;

+     Pixmap oldMask;

  

      /*

       * For each of the options in masterPtr, translate the string

***************

*** 395,410 ****

                (unsigned) masterPtr->height);

      }

  

!     if (instancePtr->mask != None) {

!       Tk_FreePixmap(Tk_Display(instancePtr->tkwin), instancePtr->mask);

!       instancePtr->mask = None;

!     }

      if (masterPtr->maskData != NULL) {

        instancePtr->mask = XCreateBitmapFromData(

                Tk_Display(instancePtr->tkwin),

                RootWindowOfScreen(Tk_Screen(instancePtr->tkwin)),

                masterPtr->maskData, (unsigned) masterPtr->width,

                (unsigned) masterPtr->height);

      }

  

      if (masterPtr->data != NULL) {

--- 396,418 ----

                (unsigned) masterPtr->height);

      }

  

!     /*

!      * Careful:  We have to allocate a new mask Pixmap before deleting

!      * the old one.  Otherwise, The XID allocator will always return

!      * the same XID for the new Pixmap as was used for the old Pixmap.

!      * And that will prevent the mask from changing in the GC below.

!      */

!     oldMask = instancePtr->mask;

!     instancePtr->mask = None;

      if (masterPtr->maskData != NULL) {

        instancePtr->mask = XCreateBitmapFromData(

                Tk_Display(instancePtr->tkwin),

                RootWindowOfScreen(Tk_Screen(instancePtr->tkwin)),

                masterPtr->maskData, (unsigned) masterPtr->width,

                (unsigned) masterPtr->height);

+     }

+     if (oldMask != None) {

+       Tk_FreePixmap(Tk_Display(instancePtr->tkwin), oldMask);

      }

  

      if (masterPtr->data != NULL) {









PatchFiles:

tkImgBmap.c









This fix is a quick hack.  We really need to think long and hard about a
better way to fix the interaction between AllocXId() and Tk_GetGC().

It also seems like the same bitmap could be shared across multiple
image instances, at least on the same display.  Is there any particular
reason why this optimization is not done?

02/03/1999 17:17 - stanton - I have applied the suggested patch, but the fundamental problem still exists.  The core issue is that Tk_GetGC doesn't notice when an XID is reused for a different object of the same type, so it will return a stale GC.



----------------------------------------------------------------------

>Comment By: Donal K. Fellows (dkf)
Date: 2003-05-13 21:39

Message:
Logged In: YES 
user_id=79902

Unless something gets done on this as part of TkGS (e.g.
wrapping the pixmap ids in something that lists the GCs to
be invalidated when the pixmap is deleted) ensuring that the
new id is allocated before the old one is released is
probably the only way to take this on.  Rats, but that is
inelegant. :^/

----------------------------------------------------------------------

Comment By: Donal K. Fellows (dkf)
Date: 2003-02-25 09:41

Message:
Logged In: YES 
user_id=79902

I think I've hit this fairly recently (i.e. since I became a core 
maintainer...)

----------------------------------------------------------------------

Comment By: Don Porter (dgp)
Date: 2002-07-05 23:04

Message:
Logged In: YES 
user_id=80530

still valid in 8.4 ?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=112997&aid=220783&group_id=12997