1. 18 Jul, 1993 1 commit
  2. 19 Jun, 1993 1 commit
  3. 18 Jun, 1993 1 commit
    • Jim Blandy's avatar
      Changes for Irix 4.0, tested this time: · 5a570e37
      Jim Blandy authored
      	* s/irix4-0.h: Get rid of our fake definitions for setpgrp and
      	getpgrp.
      	* callproc.c (Fcall_process): Go ahead and use the USG calling
      	convention for setpgrp.
      	* ymakefile (pre-crt0.o): Add rule for this.  Perhaps it will help
      	separate-source-directory compilation.
      	* emacs.c (shut_down_emacs): Some USG systems #define
      	EMACS_HAVE_TTY_PGRP; call getpgrp as appropriate for such systems.
      	* sysdep.c (sys_suspend): Call getpgrp as appropriate for USG and
      	non-USG.
      	* process.c [IRIX] (create_process): Don't put child in process
      	group zero before opening the tty; Irix is like USG in this
      	regard.
      5a570e37
  4. 17 Jun, 1993 1 commit
    • Jim Blandy's avatar
      Changes for correct pgrp behavior; approach suggested by Bob · 65aa44ac
      Jim Blandy authored
      	Glickstein <bobg@zindigo.z-code.com>:
      	* m/iris4d.h (LIB_STANDARD): Do list -lbsd here.
      	* s/irix4-0.h (getpgrp, setpgrp): #define these to call BSDgetpgrp
      	and BSDsetpgrp.
      	(GETPGRP_NO_ARG): Don't #define this.
      	* emacs.c (main): Don't test GETPGRP_NO_ARG.
      	* sysdep.c (sys_suspend): Don't test GETPGRP_NO_ARG.
      
      	* config.h.in (HAVE_CLOSEDIR): Add an #undef for autoconf to
      	change.
      	* sysdep.c (closedir): Only define this if autoconf can't find
      	HAVE_CLOSEDIR.  If we're running under solaris (defined (sun) &&
      	defined (USG5_4)), then don't free the buffer separately.
      65aa44ac
  5. 16 Jun, 1993 2 commits
  6. 12 Jun, 1993 1 commit
  7. 11 Jun, 1993 1 commit
    • Jim Blandy's avatar
      Interact properly with shells lacking job control (sh, rc, es...) · 9ae8f997
      Jim Blandy authored
      	* sysdep.c [BSD] (inherited_pgroup): New variable.
      	(narrow_foreground_group, widen_foreground_group): New functions.
      	(init_sys_modes): Call narrow_foreground_group.
      	(reset_sys_modes): Call widen_foreground_group.
      	* emacs.c [BSD] (inherited_pgroup): Add extern declaration.
      	[BSD] (main): Set inherited_pgroup, and put ourselves in our own
      	pgroup.
      9ae8f997
  8. 10 Jun, 1993 1 commit
  9. 09 Jun, 1993 1 commit
  10. 07 Jun, 1993 1 commit
  11. 04 Jun, 1993 1 commit
  12. 31 May, 1993 1 commit
  13. 30 May, 1993 3 commits
  14. 29 May, 1993 2 commits
  15. 27 May, 1993 1 commit
    • Jim Blandy's avatar
      * systty.h (HAVE_LTCHARS, HAVE_TCHARS): New macros; define them if · 50b8cf60
      Jim Blandy authored
      	we have those structures, but *don't* define them if we have
      	TERMIOS, whose functions take care of those parameters; that
      	screws up AIX.
      	(struct emacs_tty): Test those symbols, instead of the ioctl
      	commands.
      	* sysdep.c (emacs_get_tty, emacs_set_tty, new_ltchars, new_tchars,
      	init_sys_modes): Same.
      50b8cf60
  16. 26 May, 1993 1 commit
  17. 25 May, 1993 2 commits
  18. 22 May, 1993 1 commit
  19. 21 May, 1993 2 commits
  20. 19 May, 1993 1 commit
    • Jim Blandy's avatar
      Some changes from Michael K. Johnson for Linux. · 00eaaa32
      Jim Blandy authored
      	* sysdep.c (sys_siglist): Don't define this if HAVE_SYS_SIGLIST is
      	#defined.  That lets the system provide it, if it has it.
      	* syssignal.h (sigmask): Only define this if <signal.h> hasn't
      	given us a definition already.
      	* syssignal.h (sys_sigpause): Fix argument in prototype.
      	* sysdep.c (init_signals): The masks are called empty_mask and
      	full_mask, not signal_empty_mask and signal_full_mask.
      	(signal_handler_t): Moved ....
      	* syssignal.h: ... to here.
      	* systty.h (EMACS_SET_TTY_PGRP): Call tcsetpgrp with the correct
      	arguments.
      	* emacs.c (main): Don't try to establish signal handlers for
      	SIGBUS and SIGSYS unless they're actually #defined.
      00eaaa32
  21. 18 May, 1993 1 commit
  22. 04 May, 1993 1 commit
  23. 26 Apr, 1993 1 commit
  24. 31 Mar, 1993 1 commit
    • Jim Blandy's avatar
      Put interrupt input blocking in a separate file from xterm.h. · 9ac0d9e0
      Jim Blandy authored
      	This isn't specific to X, and it allows us to avoid #including
      	xterm.h in files that don't really have anything to do with X.
      	* blockinput.h: New file.
      	* xterm.h (BLOCK_INPUT, UNBLOCK_INPUT, TOTALLY_UNBLOCK_INPUT,
      	UNBLOCK_INPUT_RESIGNAL): These are now in blockinput.h.
      	(x_input_blocked, x_pending_input): Deleted; there are analogs
      	in blockinput.h called interrupt_input_blocked and
      	interrupt_input_pending.
      	* keyboard.c (interrupt_input_blocked, interrupt_input_pending):
      	New variables, used by the macros in blockinput.h.
      	* xterm.c: #include blockinput.h.
      	(x_input_blocked, x_pending_input): Deleted.
      	(XTread_socket): Test and set interrupt_input_blocked and
      	interrupt_input_pending instead of the old variables.
      	* alloc.c, xfaces.c, xfns.c, xmenu.c, xselect.c, keymap.c:
      	#include blockinput.h.
      	* eval.c: #include blockinput.h instead of xterm.h.
      	* keyboard.c: #include blockinput.h.
      	(input_poll_signal): Just test
      	interrupt_input_blocked, instead of testing HAVE_X_WINDOWS and
      	x_input_blocked.
      
      	Block the processing of interrupt input while we're manipulating the
      	malloc heap.
      	* alloc.c: (xfree): New function, to make it easy to free things
      	safely.
      	(xmalloc, xrealloc): Block X input while doing the deed.
      	(VALIDATE_LISP_STORAGE, gc_sweep, compact_strings): Use xfree
      	instead of free.
      	(uninterrupt_malloc): New function, to install input-blocking
      	hooks into the GNU malloc routines.
      	* emacs.c [not SYSTEM_MALLOC] (main): Call uninterrupt_malloc
      	on startup.
      	* alloc.c: (make_interval, make_float, Fcons, Fmake_vector,
      	Fmake_symbol, Fmake_marker, make_uninit_string, Fgarbage_collect):
      	Use xmalloc instead of malloc; don't bother to check if out of
      	memory here.
      	(Fgarbage_collect): Call xrealloc instead of realloc.
      	* buffer.c: Use xmalloc and xfree instead of malloc and free;
      	don't bother to check if out of memory here.
      	(Fget_buffer_create): Put BLOCK_INPUT/UNBLOCK_INPUT pair around
      	calls to ralloc routines.
      	* insdel.c: Same.
      	* lisp.h (xfree): New extern declaration.
      	* xfaces.c (xfree): Don't #define this to be free; use the
      	definition in alloc.c.
      	* dispnew.c, doc.c, doprnt.c, fileio.c, lread.c, term.c, xfns.c,
      	xmenu.c, xterm.c: Use xfree instead of free.
      	* hftctl.c: Use xfree and xmalloc instead of free and malloc.
      	* keymap.c (current_minor_maps): BLOCK_INPUT while calling realloc
      	and malloc.
      	* search.c: Since the regexp routines can malloc, BLOCK_INPUT
      	while runing them.  #include blockinput.h.
      	* sysdep.c: #include blockinput.h.  Call xfree and xmalloc instead
      	of free and malloc.  BLOCK_INPUT around routines which we know
      	will call malloc.
      
      	ymakefile (keyboard.o, keymap.o, search.o, sysdep.o, xfaces.o,
      	xfns.o, xmenu.o, xterm.o, xselect.o, alloc.o, eval.o): Note that
      	these depend on blockinput.h.
      9ac0d9e0
  25. 23 Mar, 1993 1 commit
  26. 20 Mar, 1993 1 commit
  27. 19 Mar, 1993 1 commit
    • Jim Blandy's avatar
      Some VMS changes from Richard Levitte <levitte@e.kth.se>: · 210b2b4f
      Jim Blandy authored
      	* [VMS] systime.h: Include vmstime.h.  VMS has the timezone
      	variable and the tzname array.
      	* s/vms.h: VMS does have select.
      	mth$dmod is the same as Unix's drem.
      	Use the time functions in vmstime.c.
      	No need to rename the malloc routines if we're using GNU malloc.
      	PURESIZE needs to be 330000.
      	* vmstime.c, vmstime.h: New files.
      	* systty.h: Don't try to initialize extern declarations under VAX C.
      	* vmspaths.h (PATH_LOADSEARCH): Include EMACS_LIBRARY:[LOCAL-LISP]
      	in PATH_LOADSEARCH.
      	(PATH_EXEC): Use EMACS_LIBRARY:[LIB-SRC] instead of [ETC].
      	* sysdep.c [VMS] (init_sys_modes): Don't allocate process_ef.
      	[VMS] (queue_kbd_input): Build events structure correctly.
      	[VMS] (gethostname): New function.
      	[VMS] (getwd): Don't get the PATH environment variable; that's
      	dumb.  Call getcwd.
      210b2b4f
  28. 11 Mar, 1993 1 commit
    • Jim Blandy's avatar
      * sysdep.c (read_input_waiting): Set e.frame_or_window, not · 6c65530f
      Jim Blandy authored
      	e.frame; the latter doesn't exist anymore.
      
      	* sysdep.c (start_of_data): If ORDINARY_LINK is #defined, we don't
      	have the data_start symbol defined, so we'll just use the address
      	of environ.
      
      	* sysdep.c (init_baud_rate): Use input_fd, instead of fd; the
      	latter is undefined.
      6c65530f
  29. 22 Feb, 1993 1 commit
  30. 16 Nov, 1992 1 commit
  31. 07 Nov, 1992 1 commit
  32. 29 Oct, 1992 1 commit
  33. 23 Sep, 1992 1 commit
    • Jim Blandy's avatar
      * sysdep.c (sys_suspend): Don't try to use "nice (- nice (0))" to · 1593c2fe
      Jim Blandy authored
      	set the subshell's priority to normal; nice doesn't return a
      	defined value on all systems.  Instead, since emacs_priority gives
      	the priority that Emacs was nastied to, we can use it to reset the
      	priority in a straightforward way.
      	[BSD4_1], [USG], [VMS] (setpriority): Remove dummy and
      	compatibility definitions of setpriority.
      1593c2fe
  34. 19 Sep, 1992 1 commit