* src/floatfns.c (double_integer_scale): Distinguish Inf from NaN.
(rounding_driver): Handle (ceil 0 0.0) and (ceil 0 1.0e+INF).
* test/src/floatfns-tests.el (special-round): Add tests for
weird cases like this.
Avoid crash with (floor 0 0.0)
* src/floatfns.c (rounding_driver): Signal an arithmetic
error if divisor is 0.0 or -0.0, instead of crashing.
This is doable now that we have bignums.
* src/floatfns.c (integer_value): Remove; no longer used.
(rescale_for_division): New function.
(rounding_driver): Use it to divide properly (by using bignums)
even when arguments are float, fixing a double-rounding FIXME.
* src/lisp.h (LOG2_FLT_RADIX): Move here ...
* src/timefns.c (frac_to_double): ... from here.
* test/src/floatfns-tests.el (big-round):
Add a test to catch the double-rounding bug.
This doesn’t alter behavior, and simplifies a future commit.
* src/floatfns.c (double_integer_scale): New function,
with body adapted from the old timefns.c.
* src/timefns.c (decode_float_time): Use it.
This uses ‘const’ to be better at catching bugs that
mistakenly attempt to modify a bignum value.
Lisp bignums are supposed to be immutable.
* src/alloc.c (make_pure_bignum):
* src/fns.c (sxhash_bignum):
Accept Lisp_Object instead of struct Lisp_Bignum *, as that’s
simpler now. Caller changed.
* src/bignum.h (bignum_val, xbignum_val): New inline functions.
Prefer them to &i->value and XBIGNUM (i)->value, since they
apply ‘const’ to the result.
* src/timefns.c (lisp_to_timespec): Use mpz_t const *
to point to a bignum value.
Change logb to return -infinity, +infinity, and NaN respectively.
Formerly logb returned an extreme fixnum to represent
infinity, but this is no longer the right thing to do now that
we have bignums and there is no extreme integer.
* doc/lispref/numbers.texi (Float Basics), etc/NEWS: Document.
* src/floatfns.c (Flogb): Implement this.
* src/floatfns.c (rounding_driver): Fix rounding error
that can occur when both args have values exactly
representable as integers but at least one arg is a float.
* test/src/floatfns-tests.el (big-round): New test.
* src/floatfns.c (rounding_driver):
New arg fixnum_divide. All callers changed.
(ceiling2, floor2, truncate2, round2): New functions.
Not that new, actually; these are essentially taken from Emacs 26.
(Fceiling, Ffloor, Fround, Ftruncate): Use them.
* src/floatfns.c (rounding_driver): Fix bug when one
argument is a float and the other is a bignum.
* test/src/floatfns-tests.el (bignum-round): Test for the bug.
Improve arithmetic performance by avoiding bignums until needed.
Also, simplify bignum memory management, fixing some unlikely leaks.
This patch improved the performance of (+ 2 2) by a factor of ten
on a simple microbenchmark computing (+ x 2), byte-compiled,
with x a local variable initialized to 2 via means the byte
compiler could not predict: performance improved from 135 to 13 ns.
The platform was Fedora 28 x86-64, AMD Phenom II X4 910e.
Performance also improved 0.6% on ‘make compile-always’.
* src/bignum.c (init_bignum_once): New function.
* src/emacs.c (main): Use it.
* src/bignum.c (mpz): New global var.
(make_integer_mpz): Rename from make_integer. All uses changed.
* src/bignum.c (double_to_bignum, make_bignum_bits)
(make_bignum, make_bigint, make_biguint, make_integer_mpz):
* src/data.c (bignum_arith_driver, Frem, Flogcount, Fash)
(expt_integer, Fadd1, Fsub1, Flognot):
* src/floatfns.c (Fabs, rounding_driver, rounddiv_q):
* src/fns.c (Fnthcdr):
Use mpz rather than mpz_initting and mpz_clearing private
temporaries.
* src/bignum.h (bignum_integer): New function.
* src/data.c (Frem, Fmod, Fash, expt_integer):
* src/floatfns.c (rounding_driver):
Use it to simplify code.
* src/data.c (FIXNUMS_FIT_IN_LONG, free_mpz_value):
Remove. All uses removed.
(floating_point_op): New function.
(floatop_arith_driver): New function, with much of the guts
of the old float_arith_driver.
(float_arith_driver): Use it.
(floatop_arith_driver, arith_driver):
Simplify by assuming NARGS is at least 2.
All callers changed.
(float_arith_driver):
New arg, containing the partly converted value of the next arg.
Reorder args for consistency. All uses changed.
(bignum_arith_driver): New function.
(arith_driver): Use it. Do fixnum-only integer calculations
in intmax_t instead of mpz_t, when they fit.
Break out mpz_t calculations into bignum_arith_driver.
(Fquo): Use floatop_arith_driver instead of float_arith_driver,
since the op is known to be valid.
(Flogcount, Fash): Simplify by coalescing bignum and fixnum code.
(Fadd1, Fsub1): Simplify by using make_int.
* src/bignum.c, src/bignum.h: New files. Only modules that
need to know how bignums are implemented should include
bignum.h. Currently these are alloc.c, bignum.c (of course),
data.c, emacs.c, emacs-module.c, floatfns.c, fns.c, print.c.
* src/Makefile.in (base_obj): Add bignum.o.
* src/alloc.c (make_bignum_str): Move to bignum.c.
(make_number): Remove; replaced by bignum.c’s make_integer.
All callers changed.
* src/conf_post.h (ARG_NONNULL): New macro.
* src/json.c (json_to_lisp): Use it.
* src/data.c (Fnatnump):
Move NATNUMP’s implementation here from lisp.h.
* src/data.c (Fnumber_to_string):
* src/editfns.c (styled_format):
Move conversion of string to bignum to bignum_to_string, and
call it here.
* src/emacs-module.c (module_make_integer):
* src/floatfns.c (Fabs):
Simplify by using make_int.
* src/emacs.c: Include bignum.h, to expand its inline fns.
* src/floatfns.c (Ffloat): Simplify by using XFLOATINT.
(rounding_driver): Simplify by using double_to_bignum.
(rounddiv_q): Clarify use of temporaries.
* src/lisp.h: Move decls that need to know bignum internals to
bignum.h. Do not include gmp.h or mini-gmp.h; that is now
bignum.h’s job.
(GMP_NUM_BITS, struct Lisp_Bignum, XBIGNUM, mpz_set_intmax):
Move to bignum.h.
(make_int): New function.
(NATNUMP): Remove; all callers changed to use Fnatnump.
(XFLOATINT): If arg is a bignum, use bignum_to_double, so that
bignum internals are not exposed here.
* src/print.c (print_vectorlike): Use SAFE_ALLOCA to avoid the
need for a record_unwind_protect_ptr.
Also, since Emacs historically reported a range error when
rounding operations overflowed, do that consistently for all
bignum overflows.
* doc/lispref/errors.texi (Standard Errors):
* doc/lispref/numbers.texi (Integer Basics): Document range errors.
* src/alloc.c (range_error): Rename from integer_overflow.
All uses changed.
* src/floatfns.c (rounding_driver): When the result of a floating
point rounding operation does not fit into a fixnum, put it
into a bignum instead of always signaling an range error.
* test/src/floatfns-tests.el (divide-extreme-sign):
These tests now return the mathematically-correct answer
instead of signaling an error.
(bignum-round): Check that integers round to themselves.
Problem reported by Andy Moreton (Bug#32463#35 (d)).
* src/floatfns.c (rounding_driver): Change the signature
of the integer rounder to use mpz_t rather than EMACS_INT.
All uses changed. Support bignums.
(ceiling2, floor2, truncate2, round2): Remove.
All uses changed to rounddiv_q or to a GMP library function.
(rounddiv_q): New function.
* test/src/floatfns-tests.el (bignum-round): New test.
libgmp calls ‘abort’ when given numbers too big for its
internal data structures. The numeric limit is large and
platform-dependent; with 64-bit GMP 6.1.2 it is around
2**2**37. Work around the problem by refusing to call libgmp
functions with arguments that would cause an abort. With luck
libgmp will have a better way to do this in the future.
Also, introduce a variable integer-width that lets the user
control how large bignums can be. This currently defaults
to 2**16, i.e., it allows bignums up to 2**2**16. This
should be enough for ordinary computation, and should
help Emacs to avoid thrashing or hanging.
Problem noted by Pip Cet (Bug#32463#71).
* doc/lispref/numbers.texi, etc/NEWS:
Document recent bignum changes, including this one.
Improve documentation for bitwise operations, in the light
of bignums.
* src/alloc.c (make_number): Enforce integer-width.
(integer_overflow): New function.
(xrealloc_for_gmp, xfree_for_gmp):
Move here from emacs.c, as it's memory allocation.
(init_alloc): Initialize GMP here, rather than in emacs.c.
(integer_width): New var.
* src/data.c (GMP_NLIMBS_MAX, NLIMBS_LIMIT): New constants.
(emacs_mpz_size, emacs_mpz_mul)
(emacs_mpz_mul_2exp, emacs_mpz_pow_ui): New functions.
(arith_driver, Fash, expt_integer): Use them.
(expt_integer): New function, containing integer code
that was out of place in floatfns.c.
(check_bignum_size, xmalloc_for_gmp): Remove.
* src/emacs.c (main): Do not initialize GMP here.
* src/floatfns.c (Fexpt): Use expt_integer, which
now contains integer code moved from here.
* src/lisp.h (GMP_NUMB_BITS): Define if gmp.h doesn’t.
Problem reported by Federico in:
https://lists.gnu.org/r/emacs-devel/2018-08/msg00619.html
* src/floatfns.c (Fexpt): Use TYPE_RANGED_FIXNUMP, not
RANGED_FIXNUMP, to fix bug with unsigned comparison on
platforms built --without-wide-int.
Problem and initial solution reported by Andy Moreton in:
https://lists.gnu.org/r/emacs-devel/2018-08/msg00503.html
* doc/lispref/numbers.texi (Math Functions): expt integer
overflow no longer causes truncation; it now signals an error
since bignum overflow is a big deal.
* src/floatfns.c (Fexpt): Support bignum arguments.
* test/src/floatfns-tests.el (bignum-expt): New test.
* src/floatfns.c (Fabs): Improve performance by not copying
the argument if it would eql the result. As a minor detail,
don't assume fixnums are two’s complement.
These should be portable nowadays.
* src/data.c (isnan): Remove.
* src/floatfns.c (isfinite, isnan): Remove.
* src/print.c: Include math.h, for isinf and isnan.
(float_to_string): Simplify by using them.
Problem reported by Paul Pogonyshev (Bug#31938).
* src/editfns.c: Include math.h, for trunc.
(styled_format): For %d, truncate floating-point numbers and
convert -0 to 0, going back to how Emacs 26 did things.
* doc/lispref/strings.texi (Formatting Strings):
Document behavior of %o, %d, %x, %X on floating-point numbers.
* src/floatfns.c (trunc) [!HAVE_TRUNC]: Rename from emacs_trunc
and make it an extern function, so that editfns.c can use it.
All callers changed.
* test/src/editfns-tests.el (format-%d-float): New test.
Most of this change is to boilerplate commentary such as license URLs.
This change was prompted by ftp://ftp.gnu.org's going-away party,
planned for November. Change these FTP URLs to https://ftp.gnu.org
instead. Make similar changes for URLs to other organizations moving
away from FTP. Also, change HTTP to HTTPS for URLs to gnu.org and
fsf.org when this works, as this will further help defend against
man-in-the-middle attacks (for this part I omitted the MS-DOS and
MS-Windows sources and the test tarballs to keep the workload down).
HTTPS is not fully working to lists.gnu.org so I left those URLs alone
for now.
* configure.ac (copysign): Remove test.
* src/floatfns.c (signbit): New macro, if not already defined.
(Fcopysign): Use it instead of copysign.
(Fcopysign, syms_of_floatfns): Define the function on all platforms.
* admin/merge-gnulib (GNULIB_MODULES): Add count-leading-zeros.
* etc/NEWS: Document the change.
* lib/count-leading-zeros.c, lib/count-leading-zeros.h:
* m4/count-leading-zeros.m4: New files, copied from Gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* src/floatfns.c: Include count-leading-zeros.h.
(Flogb): Do not convert fixnum to float before taking the log,
as the rounding error can cause the answer to be off by 1.
* src/lisp.h (EMACS_UINT_WIDTH): New constant.
* test/src/floatfns-tests.el (logb-extreme-fixnum): New test.
Turn instances of extract_float into XFLOAT_DATA when possible,
and to a resurrected XFLOATINT when the arg is a number.
The resurrected XFLOATINT is more like XFLOAT and XINT in
that is valid only if its arg is a number. This clarifies
the ways in which floats can be extracted at the C level.
* src/editfns.c (styled_format):
* src/floatfns.c (extract_float, Fexpt):
Use XFLOATINT rather than open-coding it.
* src/fns.c (internal_equal):
* src/image.c (imagemagick_load_image):
* src/xdisp.c (resize_mini_window):
Prefer XFLOAT_DATA to extract_float on values known to be floats.
* src/frame.c (x_set_screen_gamma):
* src/frame.h (NUMVAL):
* src/image.c (x_edge_detection, compute_image_size):
* src/lread.c (read_filtered_event):
* src/window.c (Fset_window_vscroll):
* src/xdisp.c (handle_single_display_spec, try_scrolling)
(redisplay_window, calc_pixel_width_or_height, x_produce_glyphs)
(on_hot_spot_p):
Prefer XFLOATINT to extract_float on values known to be numbers.
* src/lisp.h (XFLOATINT): Bring back this function, except
it now assumes its argument is a number.
* src/lisp.h (XFLOATINT): Remove this alias for extract_float.
All callers changed to use extract_float.
* src/frame.h (NUMVAL): Now an inline function, not a macro.
Without this fix, (ceiling most-negative-fixnum -1.0) returns
most-negative-fixnum instead of correctly signaling range-error,
and similarly for floor, round, and truncate.
* configure.ac (trunc): Add a check, since Gnulib’s doc says
‘trunc’ is missing from MSVC 9. The Gnulib doc says ‘trunc’ is
also missing from some other older operating systems like Solaris
9 which I know we don’t care about any more, so MSVC is the only
reason to worry about ‘trunc’ here.
* src/editfns.c (styled_format): Formatting a float with %c is now an
error. The old code did not work in general, because FIXNUM_OVERFLOW_P
had rounding errors. Besides, the "if (FLOATP (...))" was in there
only as a result of my misunderstanding old code that I introduced
2011. Although %d etc. is sometimes used on floats that represent
huge UIDs or PIDs etc. that do not fit in fixnums, this cannot
happen with characters.
* src/floatfns.c (rounding_driver): Rework to do the right thing
when the intermediate result equals 2.305843009213694e+18, i.e.,
is exactly 1 greater than MOST_POSITIVE_FIXNUM on a 64-bit host.
Simplify so that only one section of code checks for overflow,
rather than two.
(double_identity): Remove. All uses changed to ...
(emacs_trunc): ... this new function. Add replacement for
platforms that lack ‘trunc’.
* src/lisp.h (FIXNUM_OVERFLOW_P, make_fixnum_or_float):
Make it clear that the arg cannot be floating point.
* test/src/editfns-tests.el (format-c-float): New test.
* test/src/floatfns-tests.el: New file, to test for this bug.
Work around Bug#22884 by rewording comments and strings to avoid ‘(’
at the start of a line unless it starts a function. This change
is a short-term hack; in the longer run we plan to fix cc-mode’s
performance for C files that have ‘(’ at the start of a line in a
comment or string.
