Fix object/function pointer inconsistencies which are a problem for gcc 3.4.2.

[freeglut] / ChangeLog

diff --git a/ChangeLog b/ChangeLog
index 732fb76..ffc15be 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -891,3 +891,30 @@ default.
 
 (232) Improved error message a bit when no suitable visual could be found
 (X11 only).
+
+(233) Fixed part of bug #926883 (Video mode matching code, memory leaks,
+fullscreen), i.e. memory leak caused by not freeing the mode lines returned
+by XF86VidModeGetAllModeLines
+
+(234) Fixed part of bug #926883 (Video mode matching code, memory leaks,
+fullscreen): Now we first try to get an exact mode match, ignoring the
+refresh rate if none could be found. This way the X11 part and the WinDoze
+behave similarly. NOTE: We still don't behave like GLUT, because it has a
+wider notion of "best" match. We have to refactor and extend freeglut quite
+a bit to do that.
+
+(235) Fixed part of bug #926883 (Video mode matching code, memory leaks,
+fullscreen), i.e. issue warnings when XF86VidModeFOO fails.
+
+(236) Reworked the X11 part of glutSetCursor, fixing bug #764187 (Variable
+mouse pointers don't work) on the way:
+   * Use slightly more compatible cursor shapes for GLUT_CURSOR_RIGHT_ARROW
+     and GLUT_CURSOR_LEFT_ARROW.
+   * Refactored and fixed the erroneous code for GLUT_CURSOR_NONE.
+   * Removed the incorrect use of XFreeCursor and use a cache of cursors
+     instead. Cursors are never freed now, we could do this e.g. via
+     reference countig if this is really needed.
+   * Fixed error handling.
+   * Unknown cursor types are an error now.
+   * Now the window state always corresponds to the wanted cursorID, even in
+     the case of GLUT_CURSOR_FULL_CROSSHAIR.