various frame-geometry fixes -------------------- ChangeLog entries follow: -------------------- src/ChangeLog addition: 2010-02-15 Ben Wing <ben@xemacs.org> * EmacsFrame.c: * EmacsFrame.c (EmacsFrameResize): * console-msw-impl.h: * console-msw-impl.h (struct mswindows_frame): * console-msw-impl.h (FRAME_MSWINDOWS_TARGET_RECT): * device-tty.c: * device-tty.c (tty_asynch_device_change): * event-msw.c: * event-msw.c (mswindows_wnd_proc): * faces.c (Fface_list): * faces.h: * frame-gtk.c: * frame-gtk.c (gtk_set_initial_frame_size): * frame-gtk.c (gtk_set_frame_size): * frame-msw.c: * frame-msw.c (mswindows_init_frame_1): * frame-msw.c (mswindows_set_frame_size): * frame-msw.c (mswindows_size_frame_internal): * frame-msw.c (msprinter_init_frame_3): * frame.c: * frame.c (enum): * frame.c (Fmake_frame): * frame.c (adjust_frame_size): * frame.c (store_minibuf_frame_prop): * frame.c (Fframe_property): * frame.c (Fframe_properties): * frame.c (Fframe_displayable_pixel_height): * frame.c (Fframe_displayable_pixel_width): * frame.c (internal_set_frame_size): * frame.c (Fset_frame_height): * frame.c (Fset_frame_pixel_height): * frame.c (Fset_frame_displayable_pixel_height): * frame.c (Fset_frame_width): * frame.c (Fset_frame_pixel_width): * frame.c (Fset_frame_displayable_pixel_width): * frame.c (Fset_frame_size): * frame.c (Fset_frame_pixel_size): * frame.c (Fset_frame_displayable_pixel_size): * frame.c (frame_conversion_internal_1): * frame.c (get_frame_displayable_pixel_size): * frame.c (change_frame_size_1): * frame.c (change_frame_size): * frame.c (generate_title_string): * frame.h: * gtk-xemacs.c: * gtk-xemacs.c (gtk_xemacs_size_request): * gtk-xemacs.c (gtk_xemacs_size_allocate): * gtk-xemacs.c (gtk_xemacs_paint): * gutter.c: * gutter.c (update_gutter_geometry): * redisplay.c (end_hold_frame_size_changes): * redisplay.c (redisplay_frame): * toolbar.c: * toolbar.c (update_frame_toolbars_geometry): * window.c: * window.c (frame_pixsize_valid_p): * window.c (check_frame_size): Various fixes to frame geometry to make it a bit easier to understand and fix some bugs. 1. IMPORTANT: Some renamings. Will need to be applied carefully to the carbon repository, in the following order: -- pixel_to_char_size -> pixel_to_frame_unit_size -- char_to_pixel_size -> frame_unit_to_pixel_size -- pixel_to_real_char_size -> pixel_to_char_size -- char_to_real_pixel_size -> char_to_pixel_size -- Reverse second and third arguments of change_frame_size() and change_frame_size_1() to try to make functions consistent in putting width before height. -- Eliminate old round_size_to_char, because it didn't really do anything differently from round_size_to_real_char() -- round_size_to_real_char -> round_size_to_char; any places that called the old round_size_to_char should just call the new one. 2. IMPORTANT FOR CARBON: The set_frame_size() method is now passed sizes in "frame units", like all other frame-sizing functions, rather than some hacked-up combination of char-cell units and total pixel size. This only affects window systems that use "pixelated geometry", and I'm not sure if Carbon is one of them. MS Windows is pixelated, X and GTK are not. For pixelated-geometry systems, the size in set_frame_size() is in displayable pixels rather than total pixels and needs to be converted appropriately; take a look at the changes made to mswindows_set_frame_size() method if necessary. 3. Add a big long comment in frame.c describing how frame geometry works. 4. Remove MS Windows-specific character height and width fields, duplicative and unused. 5. frame-displayable-pixel-* and set-frame-displayable-pixel-* didn't use to work on MS Windows, but they do now. 6. In general, clean up the handling of "pixelated geometry" so that fewer functions have to worry about this. This is really an abomination that should be removed entirely but that will have to happen later. Fix some buggy code in frame_conversion_internal() that happened to "work" because it was countered by oppositely buggy code in change_frame_size(). 7. Clean up some frame-size code in toolbar.c and use functions already provided in frame.c instead of rolling its own. 8. Fix check_frame_size() in window.c, which formerly didn't take pixelated geometry into account.

/* Hey Emacs, this is -*- C -*- code! */

/* Synched up with: Not synched with FSF. */

/* Think twice before editing this file.  Generated automatically by configure.

 The file startup.el guesses at reasonable values for load-path, exec-path,
 and lock-directory.  This means that if you move emacs and its associated
 sub-tree to a different place in the filesystem, or to a different machine,
 you won't have to do anything for it to work.

 If you define the paths in this file then they will take precedence over
 any value generated by the heuristic in startup.el.  The hardcoded paths
 will be checked to see if they are valid, in which case they will be used.
 Otherwise the editor will attempt to make its normal guess.

 See the NEWS file for a description of the heuristic used to locate the lisp
 and exec directories at startup time.  If you are looking at this file
 because you are having trouble, then you would be much better off arranging
 for those heuristics to succeed than defining the paths in this file.

   **  Let me say that again.  If you're editing this file, you're making
   **  a mistake.  Re-read the section on installation in ../etc/NEWS.

 If it defines anything, this file should define some subset of the following:

   PATH_PREFIX		The default value of `prefix-directory'.  This is the
   			default root for everything.

   PATH_LOADSEARCH	The default value of `load-path'.

   PATH_EXEC		The default value of `exec-directory' and `exec-path'.
			(exec-path also contains the value of whatever is in
			the PATH environment variable.)

   PATH_DATA		The default value of `data-directory'.  This
			is where architecture-independent files are
			searched for.

   PATH_LOCK		The name of the directory that contains lock files
			with which we record what files are being modified in 
			Emacs.  This directory should be writable by everyone.
			If this is specified, the string must end with a slash!

   PATH_SUPERLOCK	The name of the file !!!SuperLock!!! in the lock 
			directory.  You probably should let this default...

   PATH_INFO		The default value of `Info-directory-list'.
			This is where info files are searched for. */