- Aug 01, 2014
-
-
Raymond Toy authored
This appears to be a very old bug where clear-output didn't clear the output for fd-streams.
-
Raymond Toy authored
See ticket #101. This fixes item 1 of that bug report.
-
Raymond Toy authored
-
- Jul 31, 2014
-
-
Raymond Toy authored
-
Raymond Toy authored
This is the patch from the ticket except that the boot file name has been changed to conform to the usual naming rules. To test it, try (c::info :source-location :class 'error). This should return something like #S(C::FILE-SOURCE-LOCATION :FORM-NUMBERS 52 :PATHNAME "target:code/error.lisp") Notes: * compiler/globaldb.lisp: New info-type source-location/class. As class names can theoretically also be variable names it seemed reasonable to introduce this instead of using the existing source-location/defvar. * code/error.lisp (%compiler-define-condition): Take source-location as argument and store it in the infodb. (define-condition): Pass source-location along. * bootfiles/20e/boot-2014-06-1.lisp: New bootfile needed because error.lisp is compiled before globaldb.lisp.
-
Raymond Toy authored
-
Raymond Toy authored
duplicated in irrat.lisp.
-
Raymond Toy authored
-
Raymond Toy authored
now in C.
-
Raymond Toy authored
-
- Jul 26, 2014
-
-
Raymond Toy authored
-
Raymond Toy authored
* code/exports.lisp: * Export %ieee754-rem-pi/2 and %sincos. * code/irrat.lisp: * Remove some conditionalization that is always true now. * compiler/float-tran.lisp: * %sincos is exported so we don't need the package qualifier.
-
Raymond Toy authored
-
Raymond Toy authored
Update def-math-rtn to allow specifying the C function name and the lisp function name so we can use def-math-rtn with the new trig functions.
-
Raymond Toy authored
-
Raymond Toy authored
converted cis to (complex (cos x) (sin x)). Besides, that was blocking the other deftransform for cis that replaced cis with a call to %sincos.
-
Raymond Toy authored
-
- Jul 24, 2014
-
-
Raymond Toy authored
This is untested, but I'm assuming NetBSD needs fdlibm as well.
-
Raymond Toy authored
Sparc's libm almost gets everything right but it fails two tests; | Failed Form: (COS (* 13/4 PI)) | Expected -0.7071067811865471d0 but saw -0.7071067811865472d0 | | Failed Form: (SIN (SCALE-FLOAT 1.0d0 120)) | Expected 0.377820109360752d0 but saw 0.37782010936075205d0 In both of these cases, the expected value is the correct one, as confirmed by computing these values with maxima to 64 digits of precision. (Care must be used to look at the actual bits instead of rounding the printed decimal form.)
-
Raymond Toy authored
to C.
-
- Jul 23, 2014
-
-
Raymond Toy authored
and cos. Just like on darwin/x86.
-
- Jul 22, 2014
-
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
functions on Darwin/x86 even though Darwin's libm is accurate.
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
to C. Also add interface to sincos function in C. This needs more testing and currently some tests fail.
-
Raymond Toy authored
Darwin/x86 doesn't have sincos in libm so include sincos when building on Darwin.
-
Raymond Toy authored
-
- Jul 21, 2014
-
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
Raymond Toy authored
-
- Jul 02, 2014
-
-
Raymond Toy authored
It's an int, not a double.
-
- Jun 24, 2014
-
-
Raymond Toy authored
* lisp/gencgc.c * Set number of generations to GC to 3 instead of NUM_GENERATIONS - 1. * Add simple interface to allow user to set the number of generations and return the old value. * code/gc.lisp: * Add Lisp interface to set the number of generations to GC.
-
- Jun 17, 2014
-
-
Raymond Toy authored
- Jun 01, 2014
-
-
Raymond Toy authored
Change the warning to a style-warning and update the message to be more precious.
-
- May 29, 2014
-
-
Raymond Toy authored
-
- May 13, 2014
-
-
Raymond Toy authored
-