+2014-06-15 Glenn Morris <rgm@gnu.org>
+
+ * Makefile.in: Use `make -C' rather than `cd && make' throughout.
+
+2014-06-15 Eli Zaretskii <eliz@gnu.org>
+
+ * xdisp.c (Fmove_point_visually): Don't use the glyph matrix
+ information if we are in the middle of executing a keyboard macro,
+ since redisplay doesn't update the screen until the macro is
+ finished. (Bug#17777)
+
+ * alloc.c (cleanup_vector): Don't dereference a font driver
+ pointer if it is NULL. (Bug#17771)
+
+2014-06-13 Glenn Morris <rgm@gnu.org>
+
+ * Makefile.in ($(leimdir)/leim-list.el, $(srcdir)/macuvs.h)
+ ($(lispsource)/international/charprop.el)
+ ($(libsrc)/make-docfile$(EXEEXT), $(lwlibdir)/liblw.a)
+ ($(oldXMenudir)/libXMenu11.a, ns-app, .el.elc)
+ ($(lispsource)/loaddefs.el, bootstrap-emacs$(EXEEXT)):
+ GNU make automatically passes command-line arguments to sub-makes.
+
+2014-06-13 Paul Eggert <eggert@cs.ucla.edu>
+
+ Avoid hangs in accept-process-output (Bug#17647).
+ * lisp.h, process.c (wait_reading_process_input):
+ Return int, not bool. All uses changed.
+ * process.c (SELECT_CANT_DO_WRITE_MASK):
+ Remove macro, replacing with ...
+ (SELECT_CAN_DO_WRITE_MASK): ... new constant, with inverted sense.
+ All uses changed.
+ (status_notify): New arg WAIT_PROC. Return int, not void.
+ All uses changed.
+
+2014-06-13 Eli Zaretskii <eliz@gnu.org>
+
+ * menu.c (Fx_popup_menu): Don't call the frame's menu_show_hook if
+ the frame is the initial frame, because the hook is not set up
+ then, and Emacs crashes.
+ Reported by Fabrice Popineau <fabrice.popineau@gmail.com>.
+
+2014-06-12 Stefan Monnier <monnier@iro.umontreal.ca>
+
+ * keymap.c (silly_event_symbol_error): Don't recommend the use
+ of strings.
+
+2014-06-11 Eli Zaretskii <eliz@gnu.org>
+
+ * xdisp.c (set_cursor_from_row): Fix an off-by-one error when
+ matching overlay strings with 'cursor' property against buffer
+ positions traversed in the glyph row. (Bug#17744)
+
+2014-06-11 Jan Djärv <jan.h.d@swipnet.se>
+
+ * nsterm.h (EmacsApp): Always compile in shouldKeepRunning, isFirst
+ on Cocoa.
+
+ * nsterm.m (run): Always compile for Cocoa. Use runtime check to
+ determine 10.9 (Bug#17751).
+
+ * macfont.m (macfont_draw): Positions were not freed.
+
+2014-06-10 Dmitry Antipov <dmantipov@yandex.ru>
+
+ * dispextern.h (PREPARE_FACE_FOR_DISPLAY): Remove as a duplicate of ...
+ * xfaces.c (prepare_face_for_display) [HAVE_WINDOW_SYSTEM]: ... this
+ function. Also adjust comment.
+ * fringe.c, w32term.c, xdisp.c, xterm.c: All users changed.
+
+ * dispextern.h (struct face) [HAVE_XFT]: Ifdef 'extra' member.
+ * font.c (font_done_for_face):
+ * xface.c (realize_non_ascii_face): Adjust user.
+ * font.h (struct font_driver): Convert 'prepare_face' to return
+ void because its return value is never used anyway.
+ * xfont.c (xfont_prepare_face): Return void.
+ * xftfont.c (xftfont_prepare_face): Likewise. Use xmalloc.
+ (xftfont_done_face): Use xfree.
+
+ * dispextern.h (last_tool_bar_item): Remove declaration.
+ * frame.h (struct frame): New member last_tool_bar_item.
+ * frame.c (make_frame): Initialize it.
+ * xdisp.c (toplevel): Remove last_tool_bar_item.
+ (handle_tool_bar_click, note_tool_bar_highlight):
+ * w32term.c (w32_read_socket, w32_initialize):
+ * xterm.c (handle_one_xevent, x_initialize): Adjust users.
+
+ * frame.h (window_system_available) [!HAVE_WINDOW_SYSTEM]: Always false.
+ * frame.c (window_system_available) [HAVE_WINDOW_SYSTEM]: Now here.
+
+2014-06-09 Paul Eggert <eggert@cs.ucla.edu>
+
+ Say (accept-process-output P)'s result pertains to P if P is non-nil.
+ * process.c (Faccept_process_output)
+ (wait_reading_process_output): Mention that if PROCESS is non-nil,
+ the return value is about PROCESS, not about other processes.
+
+2014-06-09 Dmitry Antipov <dmantipov@yandex.ru>
+
+ Further adjustments to mark_object and friends.
+ Now the mark_object's stack is just 32 bytes on a 64-bit
+ system, which means extra 20% off the stack usage.
+ * alloc.c (mark_save_value): As before, refactored out from ...
+ (mark_object): ... adjusted user. Also add comment.
+
+2014-06-09 Paul Eggert <eggert@cs.ucla.edu>
+
+ Fix core dump after a dropped X connection (Bug#17704).
+ * sysdep.c (stuff_char): Don't abort merely because the selected frame
+ is dead, as we may be shutting down.
+
+2014-06-08 Glenn Morris <rgm@gnu.org>
+
+ * fileio.c (write-region-inhibit-fsync): Doc tweak.
+
+ * data.c (Flss, Fgtr, Fleq, Fgeq): Doc tweaks.
+
+2014-06-08 Paul Eggert <eggert@cs.ucla.edu>
+
+ If a C name must be extern on some platforms, make it extern on all.
+ * dispextern.h (set_vertical_scroll_bar, erase_phys_cursor)
+ (load_color):
+ * font.h (ftxfont_driver) [HAVE_XFT]:
+ * keyboard.h (menu_items_inuse, ignore_mouse_drag_p, make_ctrl_char):
+ * lisp.h (get_frame_param):
+ * menu.h (tty_menu_show):
+ * process.h (conv_sockaddr_to_lisp, catch_child_signal):
+ * termhooks.h (encode_terminal_code):
+ * xterm.h (x_menu_wait_for_event):
+ Always declare.
+ * frame.c (get_frame_param):
+ * fringe.c (max_used_fringe_bitmap):
+ * ftxfont.c (ftxfont_driver):
+ * keyboard.c (ignore_mouse_drag_p, make_ctrl_char):
+ * menu.c (menu_items_inuse):
+ * process.c (conv_sockaddr_to_lisp, catch_child_signal):
+ * term.c (encode_terminal_code, tty_menu_show):
+ * xdisp.c (set_vertical_scroll_bar, erase_phys_cursor):
+ * xfaces.c (load_color):
+ * xmenu.c (x_menu_wait_for_event):
+ Now always extern.
+
+2014-06-08 Dmitry Antipov <dmantipov@yandex.ru>
+
+ Change object marking routines to minimize stack usage.
+ This change moves a few cold paths from mark_object to NO_INLINE
+ functions and adjusts symbol marking loop. According to GCC 4.8.2
+ -Wstack-usage, this reduces mark_object's stack usage from 80 to
+ 48 bytes on a 64-bit system. For a long byte-force-recompile runs,
+ stack usage at the mark phase is reduced up to 28%. Surprisingly,
+ it also gains up to 3% in speed (with default '-O2 -g3' flags).
+ * alloc.c (mark_compiled, mark_localized_symbol): New functions,
+ refactored out from ...
+ (mark_object): ... adjusted user. Also mark symbols in a tight
+ inner loop.
+ (mark_face_cache): Add NO_INLINE.
+
2014-06-08 Eli Zaretskii <eliz@gnu.org>
* sysdep.c (reset_sys_modes): Use cursorX, not curX, as the latter
was moved to Fgarbage_collect.
(Fgarbage_collect): Calculate the end address of the stack portion
that needs to be examined by mark_stack, and pass that address to
- garbage_collect_1, which will pass it to mark_stack. See
- http://lists.gnu.org/archive/html/emacs-devel/2014-05/msg00270.html
+ garbage_collect_1, which will pass it to mark_stack.
+ See http://lists.gnu.org/archive/html/emacs-devel/2014-05/msg00270.html
for more details about the underlying problems. In particular,
this avoids dumping Emacs with the large hash-table whose value is
held in purify-flag for most of the time loadup.el runs.