* libguile/ports.c (scm_ungetc_unlocked): Inline the conversion from
codepoint to bytes for UTF-8 and latin-1 ports. Speeds up a
numbers-reading test case by 100% (!).
* module/language/tree-il/compile-cps.scm: New module implementing CPS
conversion of Tree-IL.
* module/Makefile.am:
* module/language/tree-il/spec.scm:
* module/language/cps/spec.scm: Integrate CPS in the build and language
system.
* lib/jit_arm-cpu.c, lib/jit_arm.c: Correct wrong test and update
of the thumb offset information, when checking if needing to
patch a jump from arm to thumb mode. The problem would happen when
remapping the code buffer, and the new address being lower than
the previous one.
* module/scripts/disassemble.scm (disassemble): Update to work with
RTl (and only RTL, as that's the future).
* module/system/vm/debug.scm (for-each-elf-symbol): New public
interface.
(debug-context-from-image): New helper.
(find-debug-context): Use the helper.
* module/system/vm/disassembler.scm (disassemble-image): New public
interface.
* configure.ac: Extend FreeBSD test to also handle NetBSD.
* lib/jit_x86-cpu.c: Correct wrongly defined offset type of
ldxi_ui. Problem detected when building on NetBSD.
* lib/lightning.c: Adjust code to handle NetBSD mremap,
where arguments do not match Linux mremap.
lib/jit_ppc.c: Correct C sequence point problem miscalculating
the actual function address in a function descriptor. Problem
happens with gcc 4.8.1 at least.
* module/system/vm/assembler.scm (link-data): Give stringbufs the
"shared" flag already, so we don't attempt to set it at runtime. Give
.data sections the SHF_WRITE flag.
* module/system/base/compile.scm (compile-file): Pass #:to-disk? as an
option to indicate that the result will be being loaded from disk.
Perhaps a linker might want to page-align in that case.
* module/language/elisp/compile-tree-il.scm (process-options!): Accept
and ignore the #:to-file compiler option.
* libguile/vm-engine.c (return-values): Remove NVALUES operand.
Don't reset the frame.
* test-suite/tests/rtl.test ("cached-toplevel-set!"): Adapt to the fact
that 'return-values' has no operand now, and that 'reset-frame' must
be done first.
* module/language/tree-il/primitives.scm (maybe-simplify-to-eq): Wrap
within another lambda, binding the primitive name. If there are less
than two arguments, expand to constant #t. If there are more than two
arguments, convert into a conjunction of binary applications.
(expand-chained-comparisons): New procedure.
(*primitive-expand-table*): Add entries for (< <= = >= > eq?).
* module/ice-9/curried-definitions.scm (cdefine, cdefine*): Simplify,
and improve error reporting by making the patterns more strict.
(define-public): Fix bug in generated 'export' form.
* module/ice-9/boot-9.scm (module-add!): Check that the symbol argument
is actually a symbol.
(define-public): Expand into 'define' in such a way that curried
definitions will immediately fail.
Fixes <http://bugs.gnu.org/15100>.
Reported by Göran Weinholt <goran@weinholt.se>.
* libguile/bytevectors.c (INTEGERS_TO_LIST): Enforce the R6RS
requirement that SIZE divides the bytevector length.
* test-suite/tests/bytevectors.test: Add test. Remove tests that
assumes that this "divides" check is not enforced.
Partially fixes <http://bugs.gnu.org/15100>.
Reported by Göran Weinholt <goran@weinholt.se>.
* libguile/bytevectors.c (INTEGERS_TO_LIST): Make sure SIZE isn't 0.
Allow SIZE to be greater than the bytevector length, for consistency
with allowing extra bytes at the end when the bytevector length is
non-zero. Use scm_from_size_t instead of scm_from_uint.
* test-suite/tests/bytevectors.test: Add tests. Remove a test that
checks for an exception when SIZE is greater than the bytevector
length.
* module/Makefile.am: match.go depends on match.upstream.scm.
* module/ice-9/match.scm (error): Accept any number of arguments.
* module/ice-9/match.upstream.scm (match-next): Call 'error' in non-tail
context, and include the value that failed to match in the call.
* test-suite/tests/peval.test ("partial evaluation"): Adapt tests to the
fact that arithmetic operators are now expanded into binary ops with
left-to-right associativity.
* module/language/tree-il/primitives.scm (define-primitive-expander):
Use 'match-lambda*' instead of 'case-lambda' for pattern matching.
(*primitive-expand-table*): In primitive expanders for '+', '*', '-',
'/', 'logior', and 'logand', assume conventional left-to-right
associativity.
