+2002-08-09 Han-Wen Nienhuys <hanwen@cs.uu.nl>
+
+ * ports.c ("scm_new_port_table_entry"): return a boxed SCM in
+ stead of scm_t_port*. The function now takes a tag argument.
+
+2002-08-08 Han-Wen Nienhuys <hanwen@cs.uu.nl>
+
+ * gc.h: add scm_debug_cells_gc_interval to public interface
+
+ * gc-card.c ("sweep_card"): set scm_gc_running while sweeping.
+
+ * gc.c (scm_i_expensive_validation_check): separate expensive
+ validation checks from cheap ones.
+
+2002-08-06 Han-Wen Nienhuys <hanwen@cs.uu.nl>
+
+ * read.c (scm_input_error): new function: give meaningful error
+ messages, and throw read-error
+
+ * gc-malloc.c (scm_calloc): add scm_calloc.
+
+2002-08-05 Han-Wen Nienhuys <hanwen@cs.uu.nl>
+
+ * tags.h: remove GC bits documentation from the tags table.
+
+ * read.c (INPUT_ERROR): Prepare for file:line:column error
+ messages for errors in scm_lreadr() and friends.
+
+2002-08-04 Han-Wen Nienhuys <hanwen@cs.uu.nl>
+
+ * gc-malloc.c (scm_malloc): use scm_realloc() (simplifies
+ implementation).
+ (scm_gc_calloc): new function
+
+2002-08-04 Han-Wen <hanwen@cs.uu.nl>
+
+ * ports.c (scm_new_port_table_entry): init port entry to 0
+ completely.
+
+ * ports.c (scm_new_port_table_entry): change function from
+ scm_add_to_port_table. This prevents cells with null-pointers from
+ being exposed to GC.
+
+ * vports.c (s_scm_make_soft_port) strports.c (scm_mkstrport),
+ fports.c (scm_fdes_to_port): Use scm_new_port_table_entry().
+
+ * gc.c (s_scm_gc_stats): add cell-yield and malloc-yield statistic
+ to gc-stats.
+
+ * numbers.c (big2str): return "0" for 0 iso. ""
+
+ * gc-segment.c, gc-malloc.c gc-mark.c, gc-freelist.c, gc-card.c,
+ private-gc.h: new file
+
+ * gc.c: completely revised and cleaned up the GC. It now uses lazy
+ sweeping. More documentation in workbook/newgc.text
+
+2002-07-25 Marius Vollmer <marius.vollmer@uni-dortmund.de>
+
+ * random.c (rstate_free): Return zero.
+
+2002-07-24 Dirk Herrmann <D.Herrmann@tu-bs.de>
+
+ * environments.c (remove_key_from_alist): Removed.
+
+ (obarray_remove): Simplified.
+
+2002-07-24 Stefan Jahn <stefan@lkcc.org>
+
+ * continuations.h: ia64: Include <signal.h> before
+ <sys/ucontext.h>.
+
+2002-07-21 Dirk Herrmann <D.Herrmann@tu-bs.de>
+
+ * modules.c (scm_sym2var): Don't compare SCM values with ==.
+
+2002-07-21 Han-Wen <hanwen@cs.uu.nl>
+
+ * goops.c (scm_compute_applicable_methods): use
+ scm_remember_upto_here_1 iso scm_remember_upto_here
+
+ * macros.c: include deprecation.h
+
+ * vectors.c (s_scm_vector_move_right_x): remove side effect in
+ macro arg.
+ (s_scm_vector_move_left_x): idem.
+
+ * net_db.c, posix.c, socket.c: variable naming: change ans to
+ result.
+
+ * sort.c (scm_merge_vector_x): accept vector as argument
+ iso. SCM*. This is needed for full GC correctness.
+
+ * gc.h: undo previous undocumented changes related to #ifdef
+ GENGC.
+
+2002-07-20 Han-Wen <hanwen@cs.uu.nl>
+
+ * *.c: add space after commas everywhere.
+
+ * *.c: use SCM_VECTOR_SET everywhere, where a vector is written.
+ Document cases where SCM_WRITABLE_VELTS() is used.
+
+ * vectors.h (SCM_VELTS): prepare for write barrier, and let
+ SCM_VELTS() return a const pointer
+ (SCM_VECTOR_SET): add macro.
+
+2002-07-15 Dirk Herrmann <D.Herrmann@tu-bs.de>
+
+ * eval.c (SCM_CEVAL), macros.c (macro_print, scm_makmacro,
+ scm_sym_macro, scm_macro_type), macros.h (scm_makmacro):
+ Deprecated the special kind of built-in dynamic syntax transformer
+ that was inaccurately named "macro". Note: The built-in syntax
+ transformers that are named "mmacro" or "memoizing-macro" still
+ exist, and it is these which come much closer to what one would
+ call a macro.
+
+2002-07-13 Neil Jerram <neil@ossau.uklinux.net>
+
+ * eval.c (unmemocopy): Fix for
+ 1001-local-eval-error-backtrace-segfaults (unmemoization crash
+ with internal definitions and local-eval).
+
+2002-07-12 Gary Houston <ghouston@arglist.com>
+
+ * dynl.c: Don't define stub procedures if DYNAMIC_LINKING is not
+ defined. They don't do anything useful, especially since the
+ only case where DYNAMIC_LINKING is undefined seems to be
+ when --with-modules=no is given to configure, which is basically
+ requesting that the "dynamic linking module" be omitted.
+
+ * Makefile.am (libguile_la_SOURCES): move dynl.c from
+ libguile_la_SOURCES to EXTRA_libguile_la_SOURCES.
+
+ * extensions.c (load_extension): check DYNAMIC_LINKING for
+ scm_dynamic_call.
+ * init.c (scm_init_guile_1): check DYNAMIC_LINKING for
+ scm_init_dynamic_linking.
+
+2002-07-10 Marius Vollmer <mvo@zagadka.ping.de>
+
+ * guile.c, iselect.h, net_db.c, posix.c, socket.c: No need to
+ check for Cygwin when including <winsock2.h>, this is already
+ check for by configure. Thus, revert change from 2002-07-07.
+
+2002-07-10 Gary Houston <ghouston@arglist.com>
+
+ * eq.c: include <string.h>
+ * dynl.c: docstring editing.
+
+2002-07-09 Gary Houston <ghouston@arglist.com>
+
+ * dynl.c (scm_dynamic_call): docstring editing.
+
+2002-07-08 Rob Browning <rlb@defaultvalue.org>
+
+ * gc_os_dep.c: HURD fixes.
+
+2002-07-07 Marius Vollmer <mvo@zagadka.ping.de>
+
+ Crosscompiling and Cygwin fixes by Jan Nieuwenhuizen. Thanks!
+
+ * Makefile.am: Override default rule for c-tokenize.$(OBJECT);
+ this should be compiled for BUILD host.
+ Override default rule for
+ guile_filter_doc_snarfage$(EEXECT); this should run on BUILD host.
+ Add missing $(EXEEXT) to guile_filter_doc_snarfage invocation.
+ (snarf2checkedtexi): Use GUILE_FOR_BUILD instead of preinstguile.
+
+ * guile.c, iselect.h, net_db.c, posix.c, socket.c: Do not include
+ <winsock2.h> on Cygwin even when we have it.
+
+2002-07-01 Dirk Herrmann <D.Herrmann@tu-bs.de>
+
+ * __scm.h (SCM_CAUTIOUS), eval.c (scm_eval_args, deval_args,
+ SCM_CEVAL): Removed compile time option SCM_CAUTIOUS to clean up
+ the code. Full number of arguments checking of closures is
+ mandatory now. However, the option to disable the checking has
+ most probably not been used anyway.
+
+2002-06-30 Dirk Herrmann <D.Herrmann@tu-bs.de>
+
+ * __scm.h (SCM_RECKLESS), backtrace.c (SCM_ASSERT), debug.c
+ (scm_debug_options), eval.c (scm_lookupcar, scm_lookupcar1,
+ scm_badargsp, SCM_CEVAL, SCM_APPLY, scm_map, scm_for_each),
+ feature.c (scm_init_feature), gsubr.c (scm_gsubr_apply), numbers.c
+ (scm_logand, scm_logior, scm_logxor, scm_i_dbl2big), srcprop.c
+ (scm_source_properties, scm_set_source_properties_x,
+ scm_source_property): Removed compile time option SCM_RECKLESS to
+ clean up the code. Full number of arguments checking of closures
+ is mandatory now. However, the option to disable the checking has
+ most probably not been used anyway.
+
+ * srcprop.c (scm_source_properties, scm_set_source_properties_x,
+ scm_source_property): Use !SCM_CONSP instead of SCM_NCONSP.
+
+2002-06-30 Gary Houston <ghouston@arglist.com>
+
+ * dynl.c: Removed all SCM_DEFER_INTS/SCM_ALLOW_INTS, which won't
+ do anything useful. Added a comment about need for a mutex if
+ pre-emptive threading is supported.
+
+ * posix.c (scm_convert_exec_args), dynl.c
+ (scm_make_argv_from_stringlist): static procs: 1) renamed both to
+ allocate_string_pointers. 2) simplified: don't reallocate the
+ strings, just make an array of pointers 3) avoid memory leaks on
+ error 4) let the procedure report errors in its own name.
+ Consequences: 1) the procedures now assume that SCM strings are
+ nul-terminated, which should always be the case. 2) Since strings
+ are not reallocated, it's now possible for strings passed to
+ dynamic-args-call to be mutated.
+
+2002-06-28 Dirk Herrmann <D.Herrmann@tu-bs.de>
+
+ * __scm.h, eval.c, eval.h: Removed compile time option
+ MEMOIZE_LOCALS to clean up the code. Now, caching of local
+ variable positions during memoization is mandatory. However, the
+ option to disable the caching has most probably not been used
+ anyway.
+
+2002-06-18 Marius Vollmer <mvo@zagadka.ping.de>
+
+ * print.c (scm_simple_format): Print missing part of format before
+ ~% control. Thanks to Daniel Skarda!
+
+2002-06-01 Marius Vollmer <mvo@zagadka.ping.de>
+
+ * mkstemp.c: Added exception notice to license statement.
+
2002-05-22 Marius Vollmer <mvo@zagadka.ping.de>
* numbers.c (mem2ureal): When returning an inexact zero, make sure