0.

on x86.

double-float-bits for x86/sse2. This gives x86 the same
micro-optimizations that were available for sparc and ppc.


 o code/kernel.lisp:
   o Enable fast double-float-bits using the vop instead of calling
     double-float-high-bits/double-float-low-bits. 

 o compiler/float-tran.lisp:
   o Make fast-unary-ftruncate known to compiler and enable optimizer
     for it. 
   o Make double-float-bits known to compiler

 o compiler/x86/float-sse2.lisp:
   o Implement fast-unary-ftruncate for singles and doubles.
   o Implement double-float-bits.

these were needed for sparc v7 and earlier which didn't have a
multiply or divide instruction (multiply step, and divide step).  We
don't support v7 anymore so the existing vops for multiplication and
division work just fine.

Could optimize SUB-OUTPUT-INTEGER and DIGIT-TO-CHAR if needed, but I'm
assuming printing of fixnums and bignums is not limited by the
conversion of each digit to a character.

 * src/code/print.lisp:
   * Print integers in lowercase if *print-case* is :downcase
   * Update a docstring.

 * src/i18n/locale/cmucl.pot:
   * Update

 * src/general-info/release-20f.txt:
   * New file with updated info.

The comment says the field should be proceeded with a + or -, but
depended on negative integers producing the needed -.  When
disassembling with a radix, this doesn't produce the expected output
(#x7 vs #x-7 vs -#x7). So really print + or - followed by the absolute
value.

 * src/compiler/byte-comp.lisp
   * Update fasl file version to 20e
 * bootfiles/20d/boot-20e.lisp
   * Use this during a normal compile to update to 20e.

 * src/compiler/main.lisp:
   * Add *FILE-COMMENT-FROM-GIT* to control whether to use git to
     derive the file-comment.  Default is T.  Otherwise, the actual
     file-comment is used.
   * Update PROCESS-FILE-COMMENT to use *FILE-COMMENT-FROM-GIT*.

 * bin/build-all.sh
 * bin/build.sh
 * bin/build-world.sh
   * Add -G option to control whether file-comment's are derived from
     git.

I almost always build with sparc_sunc so make it the default.
I only do sparc_gcc for testing that the build still works with gcc.

implementation details of DISASSEM:DISASSEMBLE.

 code/exports.lisp::
 * Update packages so CL:DISASSEMBLE is not DISASSEM:DISASSEMBLE.

 code/misc.lisp::
 * Define CL:DISASSEMBLE.

 compiler/fndb.lisp::
 * Update defknow for disassemble.

 compiler/disassem.lisp::
 * Print of source codes uses standard I/O syntax instead of
   inheriting from the environment.
 * Add new keyword arguments to DISASSEM:DISASSEMBLE for the base,
   case, and radix.  These default to 16, :downcase, and
   *print-radix*, respectively.  This means disassembly now prints out
   all numbers in base 16 and is in lowercase.
 * When printing a note for an assembler routine, we don't need to
   print the (hex) address if *print-base* is 16.

 compiler/x86/insts.lisp::
 * Fix some issues when in print-mem-access.
   * Sometimes the absolute value of the value was printed instead of
     the value (displaying the wrong value).
   * Print out the value as an unsigned in some cases instead of
     signed value.
 * Fix print-label to print addresses as unsigned integers.  This
   fixes the issue where things like call #x-4xxxxxxx were printed.

 i18n/locale/cmucl.pot::
 * Update because of new or changed docstrings.

With these changes, you can bind *print-base* to 16 and *print-case*
to :downcase to get correctly printed disassembly in lower case with
all numerical values in hex.  If you want the radix marker as well,
bind *print-radix* to T; the disassembly is still correct.

src/compiler/disassem.lisp:
o When printing out the source code, bind *print-base* to 10 so that
  the code is printed "naturally".

src/compiler/x86/insts.lisp:
o Use Carl's much cleaned-up print-mem-access
o Update print-imm-data to use princ for all values, removing the
  special casing for NIL
o When printing out xmm registers, honor *print-case*.

src/general-info/release-20e.txt:
o Update

* Remove the name dependence on EAX in move-to-word/integer.

* Change move-from-signed to do a fixnum check using one branch
  instead of two.

* Remove move-from-signed and move-from-unsigned assembly routines and
  the commented out code that refers to them.

This change allow cmucl to fold identity operations as in

(defun foo (x)
  (declare (float x))
  (* x 1))

Previously, cmucl wouldn't change (* x 1) to just x. because the
declaration of x is represented internally as a union type.

are set but never used.