* libguile/numbers.c (mem2decimal_from_point, mem2ureal, mem2complex):
Fix a number of cases where, for invalid numbers, we could read past
the end of the buffer. This happened in e.g. "1.0+", "1/" and "1.0f".
But I couldn't figure out how to test for these, given that the
behavior depended on the contents of uninitialized memory in the
reader buffer. We'll just have to be happy with this.
Thanks to Kjetil S. Matheussen for the report.
This adds the 32-bit standalone characters. Strings are still
8-bit. Characters larger than 8-bit can only be entered or
displayed in octal format at this point. At this point, the
terminal's display encoding is expected to be Latin-1.
* module/language/assembly/compile-bytecode.scm (write-bytecode):
add 32-bit char
* module/language/assembly.scm (object->assembly): add 32-bit char
(assembly->object): add 32-bit char
* libguile/vm-i-system.c (make-char32): new op
* libguile/print.c (iprin1): print 32-bit char
* libguile/numbers.h: add type scm_t_wchar
* libguile/numbers.c: add type scm_t_wchar
* libguile/chars.h: new type scm_t_wchar
(SCM_CODEPOINT_MAX): new
(SCM_IS_UNICODE_CHAR): new
(SCM_MAKE_CHAR): operate on 32-bit char
* libguile/chars.c: comparison operators now use Unicode
codepoints
(scm_c_upcase): now receives and returns scm_t_wchar
(scm_c_downcase): now receives and returns scm_t_wchar
Thanks to Bill Schottstaedt for reporting this problem!
* libguile/numbers.c (mem2ureal): Don't be misled by *p_exactness
being INEXACT on entry (as is possible when reading a complex
number): use local exactness variable x which starts as EXACT.
Call mem2decimal_from_point () with &x instead of p_exactness.
* test-suite/tests/numbers.test ("string->number"): Add complex number
tests suggested by Bill.
(reported by Bill Schottstaedt)
* libguile/numbers.c (scm_gcd): When only one arg given, use scm_abs
to ensure that result is non-negative.
* test-suite/tests/numbers.test ("gcd"): New test, (gcd -2).
directly, and only if GUILE_I was defined by the configure step.
(scm_log, scm_log10, scm_exp, scm_sqrt): Use SCM_COMPLEX_VALUE
code only if SCM_COMPLEX_VALUE is defined.
* configure.in (--without-64-calls): Use AC_MSG_CHECKING and
AC_MSG_RESULT instead of just echo.
(GUILE_I): New programs to try using _Complex_I or 1.0fi for the
imaginary unit.
before using clog(). It's possible for gcc to provide the "complex
double" type, but for the system not to have the complex funcs.
(scm_exp): Ditto HAVE_CEXP for cexp().
(clog, cexp, carg): Remove fallback definitions. These only
duplicated the code within scm_log and scm_exp, and the latter have to
exist for the case when there's no "complex double". So better just
fix up the conditionals selecting between the complex funcs and plain
doubles than worry about fallbacks.
scm_i_make_ratio(), so fractions are only read.
scm_i_fraction_reduce() modifies a fraction when reading it. A
race condition might lead to fractions being corrupted by reading
them concurrently.
* numbers.h: remove SCM_FRACTION_SET_NUMERATOR,
SCM_FRACTION_SET_DENOMINATOR, SCM_FRACTION_REDUCED_BIT,
SCM_FRACTION_REDUCED_SET, SCM_FRACTION_REDUCED_CLEAR,
SCM_FRACTION_REDUCED.
integer-expt and multiply/divide. Inexacts and fractions no longer
supported (they happened to work before for left shifts, but not
right). Don't really need inexacts and fractions, since ash is
documented as a "bitwise operation", and all the rest of those only
take exact integers.
scm_t_intmax values.
(scm_uint2str): New, for scm_t_uintmax.
(scm_iint2str): Argument type changed to scm_t_intmax,
reimplemented in terms of scm_uint2str.
