- cmucl-cvs - mailman3.common-lisp.net

[git] CMU Common Lisp branch master updated. snapshot-2013-12-a-29-gd0aaa33
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via d0aaa33ba7afbd83575e1cb9e23832c3a563ca6d (commit) from bf1127321b08ef7e02eb7159f194c054078b95f2 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit d0aaa33ba7afbd83575e1cb9e23832c3a563ca6d Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat Dec 21 11:22:58 2013 -0800 Move src/tests/trig.lisp to tests/trig.lisp. diff --git a/src/tests/trig.lisp b/tests/trig.lisp similarity index 100% rename from src/tests/trig.lisp rename to tests/trig.lisp ----------------------------------------------------------------------- Summary of changes: {src/tests => tests}/trig.lisp | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename {src/tests => tests}/trig.lisp (100%) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2014-06-80-gae3e203
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via ae3e2030908db293f7deeebb2ef079dcd1024bc5 (commit) from 55767f82d8cef9e280e142d1c4c5db978e9d1310 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit ae3e2030908db293f7deeebb2ef079dcd1024bc5 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat Aug 2 20:58:37 2014 -0700 -march=pentium4 imples -msse2, so drop the latter. diff --git a/src/lisp/Config.x86_linux b/src/lisp/Config.x86_linux index 213e42c..cba5ec6 100644 --- a/src/lisp/Config.x86_linux +++ b/src/lisp/Config.x86_linux @@ -1,7 +1,7 @@ # -*- Mode: makefile -*- include Config.x86_common -CPPFLAGS += -m32 -rdynamic -D__NO_CTYPE -D_GNU_SOURCE -march=pentium4 -mfpmath=sse -msse2 +CPPFLAGS += -m32 -rdynamic -D__NO_CTYPE -D_GNU_SOURCE -march=pentium4 -mfpmath=sse UNDEFSYMPATTERN = -Xlinker -u -Xlinker & ASSEM_SRC += linux-stubs.S ----------------------------------------------------------------------- Summary of changes: src/lisp/Config.x86_linux | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2014-06-79-g55767f8
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 55767f82d8cef9e280e142d1c4c5db978e9d1310 (commit) from 07a07a799922a073fde0093867fbb4bef756b219 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 55767f82d8cef9e280e142d1c4c5db978e9d1310 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat Aug 2 19:46:05 2014 -0700 Forgot to remove e_rem_pio2 and friends. diff --git a/src/lisp/Config.ppc_darwin b/src/lisp/Config.ppc_darwin index b0768e0..eab5efa 100644 --- a/src/lisp/Config.ppc_darwin +++ b/src/lisp/Config.ppc_darwin @@ -40,7 +40,7 @@ ASFLAGS = $(OSX_VERSION) -traditional -g -O3 -DDARWIN -Dppc $(LINKAGE) $(GENCGC) UNDEFSYMPATTERN = -Xlinker -u -Xlinker & ASSEM_SRC = ppc-assem.S linux-stubs.S ARCH_SRC = ppc-arch.c -OS_SRC = ppc-darwin-dlshim.c os-common.c Darwin-os.c e_rem_pio2.c k_rem_pio2.c +OS_SRC = ppc-darwin-dlshim.c os-common.c Darwin-os.c ifdef FEATURE_EXECUTABLE ----------------------------------------------------------------------- Summary of changes: src/lisp/Config.ppc_darwin | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2014-03-1-g508b659
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 508b6594cf003d6a9d400d59d4ec32255728d0e9 (commit) from 901d7539202d485429c76b4262d0f6e500826a07 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 508b6594cf003d6a9d400d59d4ec32255728d0e9 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat Mar 8 11:27:06 2014 -0800 Revert change for ticket 92. It's incorrect. We really want log(-0) to be -inf + i*pi. It's needed to get the correct values on the branch cuts of the special functions from the definitions. Update trac test to test for this. diff --git a/src/compiler/float-tran.lisp b/src/compiler/float-tran.lisp index b856a6c..08ce90b 100644 --- a/src/compiler/float-tran.lisp +++ b/src/compiler/float-tran.lisp @@ -1494,15 +1494,11 @@ (two-arg-derive-type x y #'expt-derive-type-aux #'expt)) +;;; Note must assume that a type including 0.0 may also include -0.0 +;;; and thus the result may be complex -infinity + i*pi. +;;; (defun log-derive-type-aux-1 (x) - (elfun-derive-type-simple x - #'(lambda (z) - ;; log(0) and log(-0) is -infinity. - ;; Return NIL to indicate that. - (if (zerop z) - nil - (log z))) - -0d0 nil nil nil)) + (elfun-derive-type-simple x #'log 0d0 nil nil nil)) (defun log-derive-type-aux-2 (x y same-arg) (let ((log-x (log-derive-type-aux-1 x)) diff --git a/tests/trac.lisp b/tests/trac.lisp index 7fd9408..f8b4258 100644 --- a/tests/trac.lisp +++ b/tests/trac.lisp @@ -329,8 +329,20 @@ (declare (type (double-float 0d0) x)) (log x))))) (assert-equal - 'double-float - (third (kernel:%function-type f))))) + '(or double-float (complex double-float)) + (third (kernel:%function-type f))) + ;; It is important that log(-0) = -inf + i*pi. This is needed to + ;; ensure that all of the special functions have the right values + ;; on the branch cuts. (Although the implementation of the special + ;; function may not use log internally, the definition might and + ;; this property is needed to derive the correct value from the + ;; definition.) + (assert-equal + (complex ext:double-float-negative-infinity pi) + (ext:with-float-traps-masked (:divide-by-zero) (log -0d0))) + (assert-equal + (complex ext:single-float-negative-infinity (float pi 1f0)) + (ext:with-float-traps-masked (:divide-by-zero) (log -0f0))))) (define-test trac.93 (:tag :trac) ----------------------------------------------------------------------- Summary of changes: src/compiler/float-tran.lisp | 12 ++++-------- tests/trac.lisp | 16 ++++++++++++++-- 2 files changed, 18 insertions(+), 10 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. begin-x87-removal-6-g260d94f
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 260d94f1a9057fb9ae592a0b7b898f881316f289 (commit) from 1dbe0c4b5e5e05cc6a9377bc3c6505a9a3e60efc (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 260d94f1a9057fb9ae592a0b7b898f881316f289 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Thu Apr 10 20:03:20 2014 -0700 Simplify call_into_c more, putting more of the functionality into call-out. * lisp/x86-assem.S * Rip out all of cruft dealing with float return values and the need for maintaining the FPU state. Just call the foreign function and return, letting call-out deal with the result. * compiler/x86/sse2-c-call.lisp * Handle the different return values depending on whether we declared the function to return a float or not. diff --git a/src/compiler/x86/sse2-c-call.lisp b/src/compiler/x86/sse2-c-call.lisp index edd226b..83f21b5 100644 --- a/src/compiler/x86/sse2-c-call.lisp +++ b/src/compiler/x86/sse2-c-call.lisp @@ -44,35 +44,26 @@ (:generator 0 (cond ((policy node (> space speed)) (move eax function) - (inst call (make-fixup (extern-alien-name "call_into_c") :foreign)) - (when (and results (location= (tn-ref-tn results) xmm0-tn)) - ;; If there is a float result from the foreign call, - ;; call_into_c has arranged for the result to be in XMM0, - ;; as a double. If we wanted a single float, do the - ;; conversion here. - (sc-case (tn-ref-tn results) - (single-reg - (inst cvtsd2ss xmm0-tn xmm0-tn)) - (double-reg - ;; Nothing needed for double because call_into_c saved - ;; the result as a double. - )))) + (inst call (make-fixup (extern-alien-name "call_into_c") :foreign))) (t (inst call function) ;; To give the debugger a clue. XX not really internal-error? - (note-this-location vop :internal-error) - - (when (and results - (location= (tn-ref-tn results) xmm0-tn)) - ;; If there's a float result, it would have been returned - ;; in fr0 according to the ABI. We want it in xmm0. - (sc-case (tn-ref-tn results) - (single-reg - (inst fstp (ea-for-sf-stack temp-single)) - (inst movss xmm0-tn (ea-for-sf-stack temp-single))) - (double-reg - (inst fstpd (ea-for-df-stack temp-double)) - (inst movsd xmm0-tn (ea-for-df-stack temp-double))))))))) + (note-this-location vop :internal-error))) + ;; FIXME: check that a float result is returned when expected. If + ;; we don't, we'll either get a NaN when doing the fstp or we'll + ;; leave an entry on the FPU and we'll eventually overflow the FPU + ;; stack. + (when (and results + (location= (tn-ref-tn results) xmm0-tn)) + ;; If there's a float result, it would have been returned + ;; in ST(0) according to the ABI. We want it in xmm0. + (sc-case (tn-ref-tn results) + (single-reg + (inst fstp (ea-for-sf-stack temp-single)) + (inst movss xmm0-tn (ea-for-sf-stack temp-single))) + (double-reg + (inst fstpd (ea-for-df-stack temp-double)) + (inst movsd xmm0-tn (ea-for-df-stack temp-double))))))) (define-vop (alloc-number-stack-space) (:info amount) diff --git a/src/lisp/x86-assem.S b/src/lisp/x86-assem.S index b01fa5f..f9dee43 100644 --- a/src/lisp/x86-assem.S +++ b/src/lisp/x86-assem.S @@ -100,42 +100,10 @@ FUNCDEF(call_into_c) call *%eax # normal callout using Lisp stack - movl %eax,%ecx # remember integer return value - -/* Check for a return FP value */ - fxam - fnstsw %ax - andl $0x4500,%eax - cmpl $0x4100,%eax - jne Lfp_rtn_value - -/* The return value is in eax, or eax,edx? */ -no_fldz: -/* Restore the return value */ - movl %ecx,%eax # maybe return value - movl $0,GNAME(foreign_function_call_active) /* Return */ jmp *%ebx -Lfp_rtn_value: - -/* - * The float result is in st(0). We want it in xmm0. For - * consistency, save st(0) out as a double and load it into xmm0 as a - * double. The call-out vop will arrange to convert it to a single-float - * if necessary. - */ - subl $8, %esp # Space for a double float - fstpl 0(%esp) - movsd 0(%esp), %xmm0 - addl $8, %esp - -/* Don't need to restore eax as the result is in st(0) */ - - movl $0,GNAME(foreign_function_call_active) -/* Return */ - jmp *%ebx ENDFUNC(call_into_c) ----------------------------------------------------------------------- Summary of changes: src/compiler/x86/sse2-c-call.lisp | 43 +++++++++++++++---------------------- src/lisp/x86-assem.S | 32 --------------------------- 2 files changed, 17 insertions(+), 58 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2013-04-18-g431765c
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 431765cfb80004b7bbde8cc6dcd43bbfc1e17e4c (commit) from e6aeed9a036e71f217f5de8db8e497ec4a606253 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 431765cfb80004b7bbde8cc6dcd43bbfc1e17e4c Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat May 4 12:45:30 2013 -0700 Remove the -M option that was recently added. This isn't needed since we no longer set CC in the Config files. bin/build.sh:: * Revert change that added -M bin/rebuild-lisp.sh:: * Remove the -M option, but leave in the getops and usage changes so we get a little command help if needed. diff --git a/bin/build.sh b/bin/build.sh index fa0b3df..496d641 100755 --- a/bin/build.sh +++ b/bin/build.sh @@ -96,7 +96,6 @@ usage () echo " The flags always include -noinit -nositeinit" echo " -R Force recompiling the C runtime. Normally, just runs make to " echo " recompile anything that has changed." - echo " -M opt Additional flags to be passed to make." exit 1 } @@ -128,17 +127,13 @@ buildit () $TOOLDIR/clean-target.sh $CLEAN_FLAGS $TARGET || { echo "Failed: $TOOLDIR/clean-target.sh"; exit 1; } time $BUILDWORLD $TARGET $OLDLISP $BOOT || { echo "Failed: $BUILDWORLD"; exit 1; } if [ "$REBUILD_LISP" = "yes" ]; then - if [ -z $MFLAGS ]; then - $TOOLDIR/rebuild-lisp.sh $TARGET - else - $TOOLDIR/rebuild-lisp.sh -M "$MFLAGS" $TARGET - fi + $TOOLDIR/rebuild-lisp.sh $TARGET else # Set the LANG to C. For whatever reason, if I (rtoy) don't # do this on my openSuSE system, any messages from gcc are # basically garbled. This should be harmless on other # systems. - LANG=C $MAKE -C $TARGET/lisp $MFLAGS $MAKE_TARGET || { echo "Failed: $MAKE -C $TARGET/lisp"; exit 1; } + LANG=C $MAKE -C $TARGET/lisp $MAKE_TARGET || { echo "Failed: $MAKE -C $TARGET/lisp"; exit 1; } fi if [ "$BUILD_WORLD2" = "yes" ]; @@ -157,7 +152,7 @@ BUILDWORLD="$TOOLDIR/build-world.sh" BUILD_POT="yes" UPDATE_TRANS= -while getopts "123PRo:b:v:uB:C:Ui:f:w:O:M:?" arg +while getopts "123PRo:b:v:uB:C:Ui:f:w:O:?" arg do case $arg in 1) ENABLE2="no" ;; @@ -177,7 +172,6 @@ do U) UPDATE_TRANS="yes";; O) OLDLISPFLAGS="$OLDLISPFLAGS $OPTARG" ;; R) REBUILD_LISP="yes";; - M) MFLAGS="$OPTARG" ;; \?) usage ;; esac diff --git a/bin/rebuild-lisp.sh b/bin/rebuild-lisp.sh index 1e49d01..7d81f56 100755 --- a/bin/rebuild-lisp.sh +++ b/bin/rebuild-lisp.sh @@ -1,9 +1,8 @@ #!/bin/sh -while getopts "M:h?" arg +while getopts "h?" arg do case $arg in - M) MFLAGS="$OPTARG" ;; h) usage ;; \?) usage ;; esac @@ -11,11 +10,11 @@ done usage() { - echo "rebuild.-lisp.sh [-h?] [-M opts] target-directory" + echo "rebuild.-lisp.sh [-h?] target-directory" echo " -h this help" echo " -? this help" - echo " -M opts Options to pass to make" - + echo "" + echo "Force recompiling all the C code in the specified directory". } shift `expr $OPTIND - 1` @@ -43,4 +42,4 @@ fi export MAKE -${MAKE} -C $TARGET/lisp $MFLAGS clean && ${MAKE} -C $TARGET/lisp $MFLAGS +${MAKE} -C $TARGET/lisp clean && ${MAKE} -C $TARGET/lisp ----------------------------------------------------------------------- Summary of changes: bin/build.sh | 12 +++--------- bin/rebuild-lisp.sh | 11 +++++------ 2 files changed, 8 insertions(+), 15 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2013-11-9-g881903d
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 881903dddf1e3f8feed193981b0ad7af5c4cd030 (commit) from 5d7af879d56be571ab7818d3a547c08035287967 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 881903dddf1e3f8feed193981b0ad7af5c4cd030 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat Dec 7 16:22:33 2013 -0800 Use C sincos for CIS on linux/x86. diff --git a/src/code/irrat.lisp b/src/code/irrat.lisp index 5944182..d1e9a26 100644 --- a/src/code/irrat.lisp +++ b/src/code/irrat.lisp @@ -309,7 +309,49 @@ #+(or ppc sse2) (frob %%sin %%cos %%tan)) - +;; Linux has a sincos function in the C library. Use it. But we need +;; to do pi reduction ourselves because the C library doesn't do +;; accurate reduction. +#+(or (and linux x86)) +(progn +(declaim (inline %%sincos)) +(export '%%sincos) +(alien:def-alien-routine ("sincos" %%sincos) c-call:void + (x double-float) + (sin double-float :out) + (cos double-float :out)) + +(defun %sincos (theta) + (declare (double-float theta)) + ;; Accurately reduce theta. + (multiple-value-bind (n y0 y1) + (%ieee754-rem-pi/2 theta) + (multiple-value-bind (ignore s c) + (%%sincos y0) + (declare (ignore ignore)) + ;; Figure out which quadrant to use, and finish out the + ;; computation using y1. This is done by using a 1st-order + ;; Taylor expansion about y0. + (flet ((sin2 (s c y) + ;; sin(x+y) = sin(x) + cos(x)*y + (+ s (* c y))) + (cos2 (s c y) + ;; cos(x+y) = cos(x) - sin(x)*y + (- c (* s y)))) + (case (logand n 3) + (0 + (values (cos2 s c y1) + (sin2 s c y1))) + (1 + (values (- (sin2 s c y1)) + (cos2 s c y1))) + (2 + (values (- (cos2 s c y1)) + (- (sin2 s c y1)))) + (3 + (values (sin2 s c y1) + (- (cos2 s c y1))))))))) +) ;;;; Power functions. @@ -964,7 +1006,24 @@ "Return cos(Theta) + i sin(Theta), AKA exp(i Theta)." (if (complexp theta) (error (intl:gettext "Argument to CIS is complex: ~S") theta) - (complex (cos theta) (sin theta)))) + #-(or (and linux x86)) + (complex (cos theta) (sin theta)) + #+(or (and linux x86)) + (number-dispatch ((theta real)) + ((rational) + (let ((arg (coerce theta 'double-float))) + (multiple-value-bind (s c) + (%sincos arg) + (complex (coerce s 'single-float) + (coerce c 'single-float))))) + (((foreach single-float double-float)) + (multiple-value-bind (s c) + (%sincos (coerce theta 'double-float)) + (complex (coerce s '(dispatch-type theta)) + (coerce c '(dispatch-type theta))))) + #+double-double + ((double-double-float) + (complex (cos theta) (sin theta)))))) (defun asin (number) "Return the arc sine of NUMBER." diff --git a/src/compiler/float-tran.lisp b/src/compiler/float-tran.lisp index 5fbc2e0..ecca5f1 100644 --- a/src/compiler/float-tran.lisp +++ b/src/compiler/float-tran.lisp @@ -731,6 +731,30 @@ (deftransform name ((x) '(double-float) rtype :eval-name t :when :both) `(,prim x)))) +#+(or (and linux x86)) +(progn +(defknown (kernel::%sincos) + (double-float) (values double-float double-float) + (movable foldable flushable)) + +(deftransform cis ((x) (single-float) * :when :both) + `(multiple-value-bind (s c) + (kernel::%sincos (coerce x 'double-float)) + (complex (coerce s 'single-float) + (coerce c 'single-float)))) + +(deftransform cis ((x) (double-float) * :when :both) + `(multiple-value-bind (ignore s c) + (kernel::%%sincos x) + (declare (ignore ignore)) + (complex s c))) + +#+double-double +(deftransform cis ((z) (double-double-float) *) + ;; Cis. + '(complex (cos z) (sin z))) +) + ;;; The argument range is limited on the x86 FP trig. functions. A ;;; post-test can detect a failure (and load a suitable result), but ;;; this test is avoided if possible. @@ -1777,6 +1801,7 @@ (deftransform * ((z w) (,real-type (complex ,type)) *) ;; Real * complex '(complex (* z (realpart w)) (* z (imagpart w)))) + #-(or (and linux x86)) (deftransform cis ((z) ((,type)) *) ;; Cis. '(complex (cos z) (sin z))) ----------------------------------------------------------------------- Summary of changes: src/code/irrat.lisp | 63 ++++++++++++++++++++++++++++++++++++++++-- src/compiler/float-tran.lisp | 25 +++++++++++++++++ 2 files changed, 86 insertions(+), 2 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2014-05-5-g7534898
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 7534898c2e414172cda3d7d8486a868038420398 (commit) from c1e7b4aad739a7e71d0b0095893df6c257b7724a (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 7534898c2e414172cda3d7d8486a868038420398 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Sat May 10 09:29:05 2014 -0700 Use %make-double-double-float directly for the double-double constants because we don't need to renormalize the values. Also gets rid of some compiler warnings about constant folding. diff --git a/src/code/float.lisp b/src/code/float.lisp index de80bc4..9a7080d 100644 --- a/src/code/float.lisp +++ b/src/code/float.lisp @@ -140,34 +140,34 @@ (progn (defconstant least-positive-normalized-double-double-float ;; What is the right value? - (kernel:make-double-double-float least-positive-normalized-double-float - 0d0)) + (kernel:%make-double-double-float least-positive-normalized-double-float + 0d0)) (defconstant least-negative-normalized-double-double-float ;; What is the right value? - (kernel:make-double-double-float least-negative-normalized-double-float - 0d0)) + (kernel:%make-double-double-float least-negative-normalized-double-float + 0d0)) (defconstant least-positive-double-double-float - (kernel:make-double-double-float least-positive-double-float - 0d0)) + (kernel:%make-double-double-float least-positive-double-float + 0d0)) (defconstant least-negative-double-double-float - (kernel:make-double-double-float least-negative-double-float - 0d0)) + (kernel:%make-double-double-float least-negative-double-float + 0d0)) (defconstant most-positive-double-double-float ;; What is the right value? - (kernel:make-double-double-float most-positive-double-float - 0d0)) + (kernel:%make-double-double-float most-positive-double-float + 0d0)) (defconstant most-negative-double-double-float ;; What is the right value? - (kernel:make-double-double-float most-negative-double-float - 0d0)) + (kernel:%make-double-double-float most-negative-double-float + 0d0)) (defconstant double-double-float-positive-infinity ;; What is the right value? - (kernel:make-double-double-float double-float-positive-infinity - 0d0)) + (kernel:%make-double-double-float double-float-positive-infinity + 0d0)) (defconstant double-double-float-negative-infinity ;; What is the right value? - (kernel:make-double-double-float double-float-negative-infinity - 0d0)) + (kernel:%make-double-double-float double-float-negative-infinity + 0d0)) ); double-double ----------------------------------------------------------------------- Summary of changes: src/code/float.lisp | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2014-08-24-g6ae7424
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via 6ae74240bfa83dafcc6251b8a014a080d757df33 (commit) via 1626be46a98e3b18e7609227ffd7823155dcffc4 (commit) from 2c4a13afde093c0e1eb415c7252efaad6ca362f5 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 6ae74240bfa83dafcc6251b8a014a080d757df33 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Fri Aug 22 20:19:00 2014 -0700 Give up trying to get the operands if we don't know what the FP operation was. This shows up when fdlibm_setexception signals an exception. It seems to set the exception bits directly instead of trying to do the operation. This was confusing get-fp-operands. diff --git a/src/compiler/ppc/float.lisp b/src/compiler/ppc/float.lisp index 200772f..c5c944b 100644 --- a/src/compiler/ppc/float.lisp +++ b/src/compiler/ppc/float.lisp @@ -1445,6 +1445,9 @@ ;; operands, if we can. (multiple-value-bind (fop format rd rs1 rs2) (get-fp-operation scp) + (unless fop + ;; Give up if we don't know the operation + (return-from get-fp-operands (values nil (list nil nil)))) (let ((traps (logand (ldb float-exceptions-byte modes) (ldb float-traps-byte modes))) (fs1 (and fop rs1 (sigcontext-float-register scp rs1 format))) commit 1626be46a98e3b18e7609227ffd7823155dcffc4 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Thu Aug 21 23:16:02 2014 -0700 On x86/darwin, we want and only need the sse2 mode bits. Darwin doesn't use x87. diff --git a/src/code/float-trap.lisp b/src/code/float-trap.lisp index d97d04e..db0a9c4 100644 --- a/src/code/float-trap.lisp +++ b/src/code/float-trap.lisp @@ -87,7 +87,7 @@ (setf (x87-floating-point-modes) x87-modes))) ) -#+sse2 +#+(and sse2 (not darwin)) (progn (defun floating-point-modes () ;; Combine the modes from the FPU and SSE2 units. Since the sse @@ -119,7 +119,20 @@ (setf (vm::sse2-floating-point-modes) new-mode) (setf (vm::x87-floating-point-modes) x87-modes)) new-mode) -) + ) + +#(and sse2 darwin) +(progn + (defun floating-point-modes () + ;; Get just the SSE2 mode bits. + (vm::sse2-floating-point-modes)) + + (defun (setf floating-point-modes) (new-mode) + (declare (type (unsigned-byte 24) new-mode)) + ;; Set the floating point modes for SSE2. + (setf (vm::sse2-floating-point-modes) new-mode) + new-mode) + ) ;;; SET-FLOATING-POINT-MODES -- Public ;;; ----------------------------------------------------------------------- Summary of changes: src/code/float-trap.lisp | 17 +++++++++++++++-- src/compiler/ppc/float.lisp | 3 +++ 2 files changed, 18 insertions(+), 2 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0

[git] CMU Common Lisp branch master updated. snapshot-2013-05-7-gd879838
by rtoy＠common-lisp.net 08 Apr '15

08 Apr '15

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "CMU Common Lisp". The branch, master has been updated via d879838d33c578bcecf282b8b95628b6d9bbc476 (commit) from 59c50055f4538c8ba49f6f8cae403563ea76cfbb (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit d879838d33c578bcecf282b8b95628b6d9bbc476 Author: Raymond Toy <toy.raymond(a)gmail.com> Date: Mon May 20 19:08:31 2013 -0700 Don't reverse surrogate pairs in strings; it's not allowed by the CLHS. diff --git a/src/code/string.lisp b/src/code/string.lisp index f23db5b..5176edf 100644 --- a/src/code/string.lisp +++ b/src/code/string.lisp @@ -1177,18 +1177,7 @@ (src-index start (1+ src-index))) ((minusp dst-index)) (declare (fixnum src-index dst-index)) - (let ((current-char (schar sequence src-index))) - (cond ((and (lisp::surrogatep current-char :leading) - (plusp dst-index)) - ;; Reverse surrogate pairs correctly, which means the - ;; pair isn't reversed at all. - (incf src-index) - (setf (schar r dst-index) (schar sequence src-index)) - (decf dst-index) - (setf (schar r dst-index) current-char)) - (t - ;; Easy case - (setf (schar r dst-index) current-char))))) + (setf (schar r dst-index) (schar sequence src-index))) r))) #+unicode @@ -1196,16 +1185,15 @@ (declare (optimize (speed 3) (space 0) (safety 0)) (type string sequence)) (with-string sequence - (flet ((rev (start end) - (do ((i start (1+ i)) - (j (1- end) (1- j))) - ((>= i j)) - (declare (type kernel:index i j)) - (rotatef (schar sequence i) (schar sequence j))))) - (let ((len end)) - (loop for i = start then n as n = (%glyph-f sequence i) do - (rev i n) while (< n len)) - (rev start end)))) + (let ((length (- end start))) + (declare (fixnum length)) + (do ((left-index 0 (1+ left-index)) + (right-index (1- length) (1- right-index)) + (half-length (truncate length 2))) + ((= left-index half-length) sequence) + (declare (fixnum left-index right-index half-length)) + (rotatef (aref sequence left-index) + (aref sequence right-index))))) sequence) diff --git a/src/general-info/release-20e.txt b/src/general-info/release-20e.txt index d9aeb5e..858f585 100644 --- a/src/general-info/release-20e.txt +++ b/src/general-info/release-20e.txt @@ -60,6 +60,8 @@ New in this release: #79. * Fix error in (format t "~ve" 21 5d-324). (See ticket #80). * String reverse is much faster (upto 20 times) + * REVERSE and NREVERSE on strings will reverse surrogate pairs. + (Previously, surrogate pairs weren't reversed.) * Trac Tickets: ----------------------------------------------------------------------- Summary of changes: src/code/string.lisp | 32 ++++++++++---------------------- src/general-info/release-20e.txt | 2 ++ 2 files changed, 12 insertions(+), 22 deletions(-) hooks/post-receive -- CMU Common Lisp

1 0