* vip.el (vip-escape-to-emacs, vip-prefix-arg-value,
vip-prefix-arg-com): Use unread-command-event instead of
unread-command-char; respect its new semantics.
* comint.el (comint-dynamic-list-completions): Same.
(free_bloc): This can now be simplified.
* ralloc.c (r_alloc_sbrk): When we allocate new space for the
malloc heap, zero it out even if we don't have any blocs in the
free list.
Added VMS changes from Roland Roberts.
* process.c (read_process_output): Save, widen, insert the process
output, and then restore the restriction if inserting text outside
the visible region.
* process.c (Fstart_process): Establish an unwind-protect to
remove PROC from the process list if an error occurs while
starting it.
(start_process_unwind): New function to help with that.
(create_process): There's no need to explicitly call
remove_process if the fork fails; the record_unwind_protect in
Fstart_process will take care of it.
* process.c (wait_reading_process_input): Test the C preprocessor
symbol "ultrix", not "__ultrix__" to see if we should ignore
ENOMEM errors from select.
* process.c (process_send_signal): On systems which have both
the TIOCGETC and TCGETA ioctls, just use the former.
* s/bsd4-2.h, s/bsd4-3.h: #define SIGNALS_VIA_CHARACTERS.
* process.c (process_send_signal): Put all the code for sending
signals via characters in a #ifdef SIGNALS_VIA_CHARACTERS. Decide
whether to use the Berkeley-style or SYSV-style ioctls by seeing
which ioctl commands are #defined.
* process.c (process_send_signal): Doc fix.
unread_command_char to unread_command_event.
* minibuf.c (read_minibuf): If get_minibuffer gives the
new minibuffer a nil default directory, find another buffer with a
better default directory and use that one's instead.
(Fread_char, Fread_char_exclusive): Don't signal an
error for or throw away switch-frame events; instead, put them off
until after we've found a character we can respond to.
Rename unread_command_char to unread_command_event; it has
subtly different semantics now, and we should use
`make-obsolete-variable' to warn people.
* lread.c (Fread_char): Change reference.
* keyboard.c (Vlast_event_frame): Doc fix.
* keyboard.c (read_char): Write composite events to the dribble
file properly.
* keyboard.c (init_keyboard): Initialize Vlast_event_frame to
Qnil, rather than the selected frame.
* keyboard.c (lispy_modifier_list): Added sanity check before
indexing into modifier_symbols.
* keyboard.c (add_command_key): When copying the contents of the
old this_command_keys to new_keys, remember to multiply size by
sizeof (Lisp_Object) to get the amount we really need to copy.
Rename unread_command_char to unread_command_event; it has
subtly different semantics now, and we should use
`make-obsolete-variable' to warn people.
* keyboard.c (unread_command_char): Rename.
(command_loop_1, read_char, Finput_pending, Fdiscard_input,
quit_throw_to_read_char, init_keyboard, syms_of_keyboard): Change
references.
* keyboard.c (read_key_sequence): Don't use
save_excursion_{save,restore} to protect the caller against buffer
switches; use Fset_buffer and Fcurrent_buffer; redisplay might
change point, and we don't want to undo that.
* keyboard.c (kbd_buffer_get_event): When checking a mouse
movement for a frame switch, don't assume Vlast_event_frame
contains a Lisp_Frame object.
number of arguments to mark_window_display_accurate.
* dispnew.c (safe_bcopy): Use the right terminating condition in
the loop which uses multiple bcopy calls to transfer a block to an
overlapping higher block.
* buffer.c (Fbury_buffer): Make this behave as in 18.59, although
that behavior is very odd - only remove the buffer from the
selected window if BUFFER was nil or omitted.
* process.c (process_send_signal): Put all the code for sending
signals via characters in a #ifdef SIGNALS_VIA_CHARACTERS. Decide
whether to use the Berkeley-style or SYSV-style ioctls by seeing
which ioctl commands are #defined.
* make-dist: Don't forget that the way to avoid filenames starting
with `=' is to use the pattern `[a-zA-Z0-9]*.h', not
`[a-zA-Z0-9].h'. Add a new section for dealing with files that we
couldn't make hard links to, since we have two already, and
perhaps more to come.
with `=' is to use the pattern `[a-zA-Z0-9]*.h', not
`[a-zA-Z0-9].h'. Add a new section for dealing with files that we
couldn't make hard links to, since we have two already, and
perhaps more to come.
can indicate that it depends on ../src/config.h.
* Makefile.in (CONFIG_CFLAGS): Let the configure script edit this
instead of CFLAGS.
(CFLAGS): Add -Demacs and -I../src to CONFIG_CFLAGS to produce this.
replay_sequence_new_buffer label; replay_sequence should be here
instead.
* keyboard.c (Vlast_event_frame): Arrange for this to exist iff
MULTI_FRAME is defined.
[not MULTI_FRAME] (syms_of_keyboard): Don't DEFVAR Vlast_event_frame.
[not MULTI_FRAME] (read_char): Don't try to set Vlast_event_frame.
[not MULTI_FRAME] (kbd_buffer_store_event): Don't try to set
Vlast_event_frame for quit characters.
[not MULTI_FRAME] (kbd_buffer_get_event): Don't try to generate
switch-frame events.
fact that (eq TYPE 'keymap) means FUNCTION will become a keymap
when loaded.
(Fmacroexpand): Instead of assuming that every autoload
form with a fifth element is a macro, actually check the fifth
element against t and `macro', which are the only values which
denote macroness.
* configure: When editing parameters into lib-src/Makefile, change
the definition of CONFIG_CFLAGS instead of CFLAGS itself; CFLAGS
needs some other flags too.
replaced by...
(isearch-gnu-emacs-events): non-nil if should expect events in the
style generated by GNU Emacs 19. Set if set-frame-height is
fboundp; this is true on any GNU Emacs 19, whether or not it was
compiled with multiple frame support.
(isearch-mode-map): Test isearch-gnu-emacs-events instead of
isearch-frames-exist to see if we should bind switch-frame events.
(isearch-update): Test isearch-gnu-emacs-events instead of
isearch-frames-exist to see if unread-command-char's quiescent
value is nil or -1.
