Mercurial > hg > xemacs-beta
view src/lisp-disunion.h @ 1333:1b0339b048ce
[xemacs-hg @ 2003-03-02 09:38:37 by ben]
To: xemacs-patches@xemacs.org
PROBLEMS: Include nt/PROBLEMS and update. Add note about incremental
linking badness.
cmdloop.el, custom.el, dumped-lisp.el, files.el, keydefs.el, keymap.el, lisp-mode.el, make-docfile.el, replace.el, simple.el, subr.el, view-less.el, wid-edit.el: Lots of syncing with FSF 21.2.
Use if-fboundp in wid-edit.el.
New file newcomment.el from FSF.
internals/internals.texi: Fix typo.
(Build-Time Dependencies): New node.
PROBLEMS: Delete.
config.inc.samp, xemacs.mak: Eliminate HAVE_VC6, use SUPPORT_EDIT_AND_CONTINUE in its place.
No incremental linking unless SUPPORT_EDIT_AND_CONTINUE, since it
can cause nasty crashes in pdump. Put warnings about this in
config.inc.samp. Report the full compile flags used for src
and lib-src in the Installation output.
alloc.c, lisp.h, ralloc.c, regex.c: Use ALLOCA() in regex.c to avoid excessive stack allocation.
Also fix subtle problem with REL_ALLOC() -- any call to malloc()
(direct or indirect) may relocate rel-alloced data, causing
buffer text to shift. After any such call, regex must update
all its pointers to such data. Add a system, when
ERROR_CHECK_MALLOC, whereby regex.c indicates all the places
it is prepared to handle malloc()/realloc()/free(), and any
calls anywhere in XEmacs outside of this will trigger an abort.
alloc.c, dialog-msw.c, eval.c, event-stream.c, general-slots.h, insdel.c, lisp.h, menubar-msw.c, menubar-x.c: Change *run_hook*_trapping_problems to take a warning class, not
a string. Factor out code to issue warnings, add flag to
call_trapping_problems() to postpone warning issue, and make
*run_hook*_trapping_problems issue their own warnings tailored
to the hook, postponed in the case of safe_run_hook_trapping_problems()
so that the appropriate message can be issued about resetting to
nil only when not `quit'. Make record_unwind_protect_restoring_int()
non-static.
dumper.c: Issue notes about incremental linking problems under Windows.
fileio.c: Mule-ize encrypt/decrypt-string code.
text.h: Spacing changes.
author | ben |
---|---|
date | Sun, 02 Mar 2003 09:38:54 +0000 |
parents | 184461bc8de4 |
children | ae48681c47fa |
line wrap: on
line source
/* Fundamental definitions for XEmacs Lisp interpreter -- non-union objects. Copyright (C) 1985, 1986, 1987, 1992, 1993 Free Software Foundation, Inc. Copyright (C) 2001, 2002 Ben Wing. This file is part of XEmacs. XEmacs is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. XEmacs is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with XEmacs; see the file COPYING. If not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ /* Synched up with: FSF 19.30. Split out from lisp.h. */ /* This file has diverged greatly from FSF Emacs. Syncing is no longer desirable or possible */ /* Format of a non-union-type Lisp Object 3 2 1 0 bit 10987654321098765432109876543210 -------------------------------- VVVVVVVVVVVVVVVVVVVVVVVVVVVVVVTT Integers are treated specially, and look like this: 3 2 1 0 bit 10987654321098765432109876543210 -------------------------------- VVVVVVVVVVVVVVVVVVVVVVVVVVVVVVVT For integral Lisp types, i.e. integers and characters, the value bits are the Lisp object. Some people call such Lisp_Objects "immediate". The object is obtained by masking off the type bits. Bit 1 is used as a value bit by splitting the Lisp integer type into two subtypes, Lisp_Type_Int_Even and Lisp_Type_Int_Odd. By this trickery we get 31 bits for integers instead of 30. For non-integral types, the value bits of a Lisp_Object contain a pointer to a structure containing the object. The pointer is obtained by masking off the type and mark bits. All pointer-based types are coalesced under a single type called Lisp_Type_Record. The type bits for this type are required by the implementation to be 00, just like the least significant bits of word-aligned struct pointers on 32-bit hardware. This requires that all structs implementing Lisp_Objects have an alignment of at least 4 bytes. Because of this, Lisp_Object pointers don't have to be masked and are full-sized. There are no mark bits in the Lisp_Object itself (there used to be). Integers and characters don't need to be marked. All other types are lrecord-based, which means they get marked by setting the mark bit in the struct lrecord_header. Here is a brief description of the following macros: XTYPE The type bits of a Lisp_Object XPNTRVAL The value bits of a Lisp_Object storing a pointer XCHARVAL The value bits of a Lisp_Object storing a Ichar XREALINT The value bits of a Lisp_Object storing an integer, signed XUINT The value bits of a Lisp_Object storing an integer, unsigned INTP Non-zero if this Lisp_Object is an integer Qzero Lisp Integer 0 EQ Non-zero if two Lisp_Objects are identical, not merely equal. */ typedef EMACS_INT Lisp_Object; #define Lisp_Type_Int_Bit (Lisp_Type_Int_Even & Lisp_Type_Int_Odd) #define VALMASK (((1UL << VALBITS) - 1UL) << GCTYPEBITS) #define XTYPE(x) ((enum Lisp_Type) (((EMACS_UINT)(x)) & ~VALMASK)) #define XPNTRVAL(x) (x) /* This depends on Lisp_Type_Record == 0 */ #define XCHARVAL(x) ((x) >> GCBITS) #define XREALINT(x) ((x) >> INT_GCBITS) #define XUINT(x) ((EMACS_UINT)(x) >> INT_GCBITS) #define wrap_pointer_1(ptr) ((Lisp_Object) (ptr)) DECLARE_INLINE_HEADER ( Lisp_Object make_int_verify (EMACS_INT val) ) { Lisp_Object obj = (Lisp_Object) ((val << INT_GCBITS) | Lisp_Type_Int_Bit); type_checking_assert (XREALINT (obj) == val); return obj; } #define make_int(x) ((Lisp_Object) (((x) << INT_GCBITS) | Lisp_Type_Int_Bit)) #define make_char_1(x) ((Lisp_Object) (((x) << GCBITS) | Lisp_Type_Char)) #define INTP(x) ((EMACS_UINT)(x) & Lisp_Type_Int_Bit) #define INT_PLUS(x,y) ((x)+(y)-Lisp_Type_Int_Bit) #define INT_MINUS(x,y) ((x)-(y)+Lisp_Type_Int_Bit) #define INT_PLUS1(x) INT_PLUS (x, make_int (1)) #define INT_MINUS1(x) INT_MINUS (x, make_int (1)) #define Qzero make_int (0) #define Qnull_pointer ((Lisp_Object) 0) #define EQ(x,y) ((x) == (y)) /* WARNING!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! You can only VOID_TO_LISP something that had previously been LISP_TO_VOID'd. You cannot go the other way, i.e. create a bogus Lisp_Object. If you want to stuff a void * into a Lisp_Object, use make_opaque_ptr(). */ /* Convert between a (void *) and a Lisp_Object, as when the Lisp_Object is passed to a toolkit callback function */ #define VOID_TO_LISP(varg) ((Lisp_Object) (varg)) #define LISP_TO_VOID(larg) ((void *) (larg)) /* Convert a Lisp_Object into something that can't be used as an lvalue. Useful for type-checking. */ #define NON_LVALUE(larg) ((larg) + 0)