used by any configuration right now, but he's trying to make it
work.
#include "sunos4-1.h" instead of "bsd4-2.h".
(O_NDELAY): Don't define this.
(SYSTEM_MALLOC): Don't define this, either.
(LD_SWITCH_SYSTEM): Remove the definition for this.
according to overlays and text properties; use
compute_char_face and compute_glyph_face to figure out what
face to use, and where a new face starts.
* xterm.c (dumpglyphs): Use the upper bits of the glyphs to decide
which frame face to use. Call GLYPH_FOLLOW_ALIASES to make sure
we're implementing the glyph table properly. If we're not using
the default or mode line face, call intern_face to find a display
face for the frame face selected by the glyph code. Implement
underlining. Remove the `font' argument; we have to derive this
from the frame and face anyway. Change all callers.
* disptab.h (GLYPH_FOLLOW_ALIASES): New macro.
* xterm.c (x_destroy_window): Call free_frame_faces.
init_frame_faces to set up the first two faces.
* xfaces.c (init_frame_faces): Don't just try to copy the default
and mode line faces from some other random frame; instead, consult
the normal_gc and reverse_gc members of the frame, and build the
faces based on their parameters.
* dispextern.h (struct face): Add cached_index member.
* xfaces.c (get_cached_face): Use it to avoid unnecessary
searches of face_vector.
* xfaces.c (intern_face): Renamed from get_display_face.
* xfns.c (x_make_gc): After building the GC's for the frame, call
init_frame_faces to set up the first two faces.
* xfaces.c (init_frame_faces): Don't just try to copy the default
and mode line faces from some other random frame; instead, consult
the normal_gc and reverse_gc members of the frame, and build the
faces based on their parameters.
Adjust the face computation functions to return frame face ID's,
not pointers to display faces; since we call these functions
during display construction, we don't want the display faces yet.
* xfaces.c (intern_frame_face): New function.
(compute_char_face, compute_glyph_face): Apply intern_frame_face
to the computed face, and return the frame face's ID, instead of
calling intern_face and returning a pointer to a display frame.
* xfaces.c: Describe the facial data structures. It took me a
while to figure them out; perhaps this will save someone else the
trouble.
* xfaces.c (get_display_face): Use face_eql instead of writing it out.
* xdisp.c (redisplay_region): New function.
* buffer.c (Fmove_overlay): Call redisplay_region on the areas the
overlay has enclosed or left.
(Fdelete_overlay): Call redisplay_region on the area the overlay
used to occupy.
(Foverlay_put): Call redisplay_region on the area the overlay now
occupies; we may have put a face property on it.
* xdisp.c (redisplay): If we're doing a thorough redisplay (all
windows on all frames involved), go ahead and flush the GC cache -
call clear_face_vector.
* xdisp.c (display_text_line): Apply faces to characters
according to overlays and text properties; use
compute_char_face and compute_glyph_face to figure out what
face to use, and where a new face starts.
* xterm.c (dumpglyphs): Use the upper bits of the glyphs to decide
which frame face to use. Call GLYPH_FOLLOW_ALIASES to make sure
we're implementing the glyph table properly. If we're not using
the default or mode line face, call intern_face to find a display
face for the frame face selected by the glyph code. Implement
underlining. Remove the `font' argument; we have to derive this
from the frame and face anyway. Change all callers.
* disptab.h (GLYPH_FOLLOW_ALIASES): New macro.
defined everywhere this file is #included; merge the two
definitions for defined (SIGIO) and ! defined (SIGIO) into one,
which calls reinvoke_input_signal if interrupt_input_pending is
set.
* keyboard.c (reinvoke_input_signal): New function.
* keyboard.c (syms_of_keyboard): Doc fix for
extra-keyboard-modifiers.
to overlays and text properties; use compute_char_face to figure out
what face to use, and where a new face starts.
* xterm.c (dumpglyphs): Use the upper bits of the glyphs to decide
which frame face to use. Call GLYPH_FOLLOW_ALIASES to make sure
we're implementing the glyph table properly. If we're not using
the default or mode line face, call intern_face to find a display
face for the frame face selected by the glyph code. Implement
underlining. Remove the `font' argument; we have to derive this
from the frame and face anyway. Change all callers.
* disptab.h (GLYPH_FOLLOW_ALIASES): New macro.
instead of AC_OUTPUT.
* config.h.in: Remove mention of GLYPH datatype; that shouldn't be
a user option.
* lisp.h (GLYPH, MAKE_GLYPH, GLYPH_CHAR, GLYPH_FACE): New macros.
* xdisp.c (redisplay_region): New function.
* buffer.c (Fmove_overlay): Call redisplay_region on the areas the
overlay has enclosed or left.
(Fdelete_overlay): Call redisplay_region on the area the overlay
used to occupy.
(Foverlay_put): Call redisplay_region on the area the overlay now
occupies; we may have put a face property on it.
* buffer.c (Fmove_overlay): Doc fix.
defined everywhere this file is #included; merge the two
definitions for defined (SIGIO) and ! defined (SIGIO) into one,
which calls reinvoke_input_signal if interrupt_input_pending is
set.
* keyboard.c (reinvoke_input_signal): New function.
config.h file directly, instead of having lots of -D flags.
* Makefile.in (CFLAGS): Don't make this carry DEFS from the
configure script; the coding standards say that CFLAGS should be
left for the user to tweak.
* Makefile.in (LN_S): New variable, edited by top Makefile.
(SUBMAKEFILE): New variable, containing all flags to pass to
recursive makes.
* keyboard.c (do_mouse_tracking): Now static.
* keyboard.c (read_char_menu_prompt): Expect Fx_popup_menu
to return a list of events. Don't lose any of them.
* keyboard.c (syms_of_keyboard): Doc fix for
extra-keyboard-modifiers; use the same modifier bits as we do for
characters.
(read_char): Apply all the modifiers in extra_keyboard_modifiers
to the input characters, so you can get hyper, super, and the rest
of the gang.
* xterm.c (x_emacs_to_x_modifiers): New function.
(x_convert_modifiers): Renamed to x_x_to_emacs_modifiers, for
consistency. Callers changed.
(XTread_socket): Apply x_emacs_to_x_modifiers to
extra_keyboard_modifiers before setting the state member of the
event; this will get all the modifiers on ASCII characters.
* xterm.c (x_text_icon): Don't call XSetIconName; it should be
unnecessary, and perhaps it's killing the icon pixmap.
* s/mach2.h: copied from the Emacs 18.59 distribution.
Don't define NO_REMAP, define START_FILES as
`pre-crt0.o' instead. Define LIB_MATH as `-lm', to override the
default `-lm -lc' (there is no libc on the NeXT).
* ymakefile (STARTFILES): Allow config.h to set this value even if
ORDINARY_LINK is defined.
* unexnext.c: Fix subdirectories for the machine dependent include
files for NeXTStep 3.0; #include <mach/mach.h> and
<mach-o/loader.h> instead of <mach.h> and <sys/loader.h>.
(getsectbyname): Remove prototype for this; the system #include
files take care of that.
(malloc_cookie): New variable.
(unexec_doit): Set malloc_cookie to the result returned by
malloc_freezedry.
* emacs.c (main): Declare malloc_cookie to be extern, so that we can
get the value set when we dumped and pass it to malloc_jumpstart.
* systime.h: The NeXT has a timezone function.
case we need them.
* termhooks.h (alt_modifier, super_modifier, hyper_modifier,
shift_modifier, ctrl_modifier, meta_modifier): Define these in
terms of the CHAR_mumble macros, to avoid having the same thing
defined in two places.
Make the modifier manipulation functions more robust. The old way
caused a bug once, and probably would again.
* termhooks.h (alt_modifier, super_modifier, hyper_modifier,
shift_modifier, ctrl_modifier, meta_modifier): Shift these all
down one bit in value, to avoid sign extension problems.
* lisp.h (CHAR_META, CHAR_CTL, CHAR_SHIFT): Fix these definitions too.
* keyboard.c (lispy_modifier_list): Ignore modifier bits beyond
what our table of modifier names can handle.
(apply_modifiers): Don't abort if you see extra modifier bits,
just remove them.
(EMACS_SET_USECS): Remember that a `usec' is a microsecond, not a
millisecond. What's three orders of magnitude between friends?
* dispnew.c (Fsit_for, Fsleep_for): Remember to multiply the
`milliseconds' argument by 1000 to get microseconds.
Changes for Emacs 19 from Thorsten Ohl <ohl@chico.harvard.edu>:
* s/mach2.h: copied from the Emacs 18.59 distribution.
Don't define NO_REMAP, define START_FILES as
`pre-crt0.o' instead. Define LIB_MATH as `-lm', to override the
default `-lm -lc' (there is no libc on the NeXT).
* ymakefile (STARTFILES): Allow config.h to set this value even if
ORDINARY_LINK is defined.
* unexnext.c: Fix subdirectories for the machine dependent include
files for NeXTStep 3.0; #include <mach/mach.h> and
<mach-o/loader.h> instead of <mach.h> and <sys/loader.h>.
(getsectbyname): Remove prototype for this; the system #include
files take care of that.
(malloc_cookie): New variable.
(unexec_doit): Set malloc_cookie to the result returned by
malloc_freezedry.
* emacs.c (main): Declare malloc_cookie to be extern, so that we can
get the value set when we dumped and pass it to malloc_jumpstart.
* systime.h: The NeXT has a timezone function.
functions in sysdep.c.
* sysdep.c (emacs_get_tty, emacs_set_tty): Here they are.
* sysdep.c (emacs_set_tty): Call tcsetattr over and over again
until it does all of what we ask it to, or returns an error.
* lread.c (read_filtered_char): New function, which contains the
code which used to be in Fread_char, Fread_event, and
Fread_char_exclusive; there was a lot of common code.
(Fread_char, Fread_event, Fread_char_exclusive): Rewrite in terms
of read_filtered_char.
* lisp.h (read_filtered_char): Declare this extern here.
* fns.c (Fy_or_n_p): Call read_filtered_char, arranging to delay
switch-frame events.
case we need them.
* termhooks.h (alt_modifier, super_modifier, hyper_modifier,
shift_modifier, ctrl_modifier, meta_modifier): Define these in
terms of the CHAR_mumble macros, to avoid having the same thing
defined in two places.
Arrange for Fy_or_n_p to put off switch-frame events.
* lread.c (read_filtered_char): New function, which contains the
code which used to be in Fread_char, Fread_event, and
Fread_char_exclusive; there was a lot of common code.
(Fread_char, Fread_event, Fread_char_exclusive): Rewrite in terms
of read_filtered_char.
* lisp.h (read_filtered_char): Declare this extern here.
* fns.c (Fy_or_n_p): Call read_filtered_char, arranging to delay
switch-frame events.
Make the modifier manipulation functions more robust. The old way
caused a bug once, and probably would again.
* termhooks.h (alt_modifier, super_modifier, hyper_modifier,
shift_modifier, ctrl_modifier, meta_modifier): Shift these all
down one bit in value, to avoid sign extension problems.
* lisp.h (CHAR_META, CHAR_CTL, CHAR_SHIFT): Fix these definitions too.
* keyboard.c (lispy_modifier_list): Ignore modifier bits beyond
what our table of modifier names can handle.
(apply_modifiers): Don't abort if you see extra modifier bits,
just remove them.
* lisp.h (Qrange_error, Qdomain_error, Qsingularity_error,
Qoverflow_error, Qunderflow_error): Add extern to these declarations.
* keyboard.c (syms_of_keyboard): Doc fix for
extra-keyboard-modifiers; use the same modifier bits as we do for
characters.
(read_char): Apply all the modifiers in extra_keyboard_modifiers
to the input characters, so you can get hyper, super, and the rest
of the gang.
* xterm.c (x_emacs_to_x_modifiers): New function.
(x_convert_modifiers): Renamed to x_x_to_emacs_modifiers, for
consistency. Callers changed.
(XTread_socket): Apply x_emacs_to_x_modifiers to
extra_keyboard_modifiers before setting the state member of the
event; this will get all the modifiers on ASCII characters.
* keyboard.c (kbd_buffer_get_event): Don't generate switch-frame
events if they'd only switch to the frame already selected. This
avoids lots of extra switch-frame events when using a separate
minibuffer.
* keyboard.c (Fcurrent_input_mode): New function.
* keyboard.c (read_key_sequence): Let the `modifiers' variable in
the code which deals with KEY being unbound be an int, not a
Lisp_Object.
Make the modifier manipulation functions more robust. The old way
caused a bug once, and probably would again.
* termhooks.h (alt_modifier, super_modifier, hyper_modifier,
shift_modifier, ctrl_modifier, meta_modifier): Shift these all
down one bit in value, to avoid sign extension problems.
* lisp.h (CHAR_META, CHAR_CTL, CHAR_SHIFT): Fix these definitions too.
* keyboard.c (lispy_modifier_list): Ignore modifier bits beyond
what our table of modifier names can handle.
(apply_modifiers): Don't abort if you see extra modifier bits,
just remove them.
':' elsewhere, just have it default to ':' if not #defined, and
#define it to be ',' in s/vms.h; OS/2 will need it to be ';'.
* s/vms.h (SEPCHAR): #define this to be ','.
* s/template.h (SEPCHAR): Mention this.
Changes for Emacs 19 from Thorsten Ohl <ohl@chico.harvard.edu>:
* s/mach2.h: copied from the Emacs 18.59 distribution.
Don't define NO_REMAP, define START_FILES as
`pre-crt0.o' instead. Define LIB_MATH as `-lm', to override the
default `-lm -lc' (there is no libc on the NeXT).
* ymakefile (STARTFILES): Allow config.h to set this value even if
ORDINARY_LINK is defined.
* unexnext.c: Fix subdirectories for the machine dependent include
files for NeXTStep 3.0; #include <mach/mach.h> and
<mach-o/loader.h> instead of <mach.h> and <sys/loader.h>.
(getsectbyname): Remove prototype for this; the system #include
files take care of that.
(malloc_cookie): New variable.
(unexec_doit): Set malloc_cookie to the result returned by
malloc_freezedry.
* emacs.c (main): Declare malloc_cookie to be extern, so that we can
get the value set when we dumped and pass it to malloc_jumpstart.
* systime.h: The NeXT has a timezone function.
(EMACS_SET_USECS): Remember that a `usec' is a microsecond, not a
millisecond. What's three orders of magnitude between friends?
* dispnew.c (Fsit_for, Fsleep_for): Remember to multiply the
`milliseconds' argument by 1000 to get microseconds.
* dispnew.c (Fsleep_for, Fsit_for): Allow SECONDS to be a
floating point value.
* dispnew.c (getenv): Extern declaration deleted; this is done in
config.h.
should exclude the getenv declaration; instead, test NOT_C_CODE.
Per suggestion from Francesco Potorti`.
* ymakefile (NOT_C_CODE): Define this; it's true, and useful.
* config.h.in (volatile): Don't define this to be the empty string
if some file has #defined HAVE_VOLATILE.
* buffer.c (Fmake_overlay, Fmove_overlay): New optional BUFFER
arguments.
(recenter_overlay_lists): New argument BUF, to use instead of the
current buffer.
(Foverlay_recenter): Pass the appropriate arguments to
recenter_overlay_lists.
* buffer.c (Fdelete_overlay): Don't assume that overlay is in the
current buffer. Don't forget to declare the argument a Lisp_Object.
':' elsewhere, just have it default to ':' if not #defined, and
#define it to be ',' in s/vms.h; OS/2 will need it to be ';'.
* s/vms.h (SEPCHAR): #define this to be ','.
* s/template.h (SEPCHAR): Mention this.
* s/vms.h (xfree): #define this to emacs_xfree, to avoid case
conflict with XFree; on VMS, external symbols are case-insensitive.
':' elsewhere, just have it default to ':' if not #defined, and
#define it to be ',' in s/vms.h; OS/2 will need it to be ';'.
* s/vms.h (SEPCHAR): #define this to be ','.
* s/template.h (SEPCHAR): Mention this.
