Mercurial > hg > xemacs-beta
view src/lisp-disunion.h @ 4678:b5e1d4f6b66f
Make #'floor, #'ceiling, #'round, #'truncate conform to Common Lisp.
lisp/ChangeLog addition:
2009-08-11 Aidan Kehoe <kehoea@parhasard.net>
* cl-extra.el (ceiling*, floor*, round*, truncate*):
Implement these in terms of the C functions; mark them as
obsolete.
(mod*, rem*): Use #'nth-value with the C functions, not #'nth with
the CL emulation functions.
man/ChangeLog addition:
2009-08-11 Aidan Kehoe <kehoea@parhasard.net>
* lispref/numbers.texi (Bigfloat Basics):
Correct this documentation (ignoring for the moment that it breaks
off in mid-sentence).
tests/ChangeLog addition:
2009-08-11 Aidan Kehoe <kehoea@parhasard.net>
* automated/lisp-tests.el:
Test the new Common Lisp-compatible rounding functions available in
C.
(generate-rounding-output): Provide a function useful for
generating the data for the rounding functions tests.
src/ChangeLog addition:
2009-08-11 Aidan Kehoe <kehoea@parhasard.net>
* floatfns.c (ROUNDING_CONVERT, CONVERT_WITH_NUMBER_TYPES)
(CONVERT_WITHOUT_NUMBER_TYPES, MAYBE_TWO_ARGS_BIGNUM)
(MAYBE_ONE_ARG_BIGNUM, MAYBE_TWO_ARGS_RATIO)
(MAYBE_ONE_ARG_RATIO, MAYBE_TWO_ARGS_BIGFLOAT)
(MAYBE_ONE_ARG_BIGFLOAT, MAYBE_EFF, MAYBE_CHAR_OR_MARKER):
New macros, used in the implementation of the rounding functions.
(ceiling_two_fixnum, ceiling_two_bignum, ceiling_two_ratio)
(ceiling_two_bigfloat, ceiling_one_ratio, ceiling_one_bigfloat)
(ceiling_two_float, ceiling_one_float, ceiling_one_mundane_arg)
(floor_two_fixnum, floor_two_bignum, floor_two_ratio)
(floor_two_bigfloat, floor_one_ratio, floor_one_bigfloat)
(floor_two_float, floor_one_mundane_arg, round_two_fixnum)
(round_two_bignum_1, round_two_bignum, round_two_ratio)
(round_one_bigfloat_1, round_two_bigfloat, round_one_ratio)
(round_one_bigfloat, round_two_float, round_one_float)
(round_one_mundane_arg, truncate_two_fixnum)
(truncate_two_bignum, truncate_two_ratio, truncate_two_bigfloat)
(truncate_one_ratio, truncate_one_bigfloat, truncate_two_float)
(truncate_one_float, truncate_one_mundane_arg):
New functions, used in the implementation of the rounding
functions.
(Fceiling, Ffloor, Fround, Ftruncate, Ffceiling, Fffloor)
(Ffround, Fftruncate):
Revise to fully support Common Lisp conventions. This means:
-- All functions have optional DIVISOR arguments
-- All functions return multiple values; see #'values
-- All functions do their arithmetic with the correct number types
according to the contamination rules.
-- #'round and #'fround always round towards the even number
in ambiguous cases.
* doprnt.c (emacs_doprnt_1):
* number.c (internal_coerce_number):
Call Ftruncate with two arguments, not one.
* floatfns.c (Ffloat):
Correct this, if NUMBER is a bignum.
* lisp.h:
Declare Ftruncate as taking two arguments.
* number.c:
Provide scratch_ratio2, init it appropriately.
* number.h:
Make scratch_ratio2 available.
* number.h (BIGFLOAT_ARITH_RETURN):
* number.h (BIGFLOAT_ARITH_RETURN1):
Correct these functions.
author | Aidan Kehoe <kehoea@parhasard.net> |
---|---|
date | Tue, 11 Aug 2009 17:59:23 +0100 |
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)