(SCM_GC_TRIGGER_1, SCM_GC_TRIGGER_2): Default values set to -50.

[bpt/guile.git] / libguile / ChangeLog

2000-03-17  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* gc.c, gc.h: Cleanup of the change of 2000-03-15.
	Cluster sizes are now independent of GC trigger values.
	GUILE_GC_TRIGGER_n can now specify a relative trigger value:
	A negative integer gives fraction of total heap size in percent.
	(SCM_GC_TRIGGER_1, SCM_GC_TRIGGER_2): Default values set to -40.

	* init.c (scm_boot_guile_1): Introduced new variable
	GUILE_MAX_SEGMENT_SIZE; New environment variable names:
	GUILE_INIT_SEGMENT_SIZE_1, GUILE_GC_TRIGGER_1,
	GUILE_INIT_SEGMENT_SIZE_2, GUILE_GC_TRIGGER_2

2000-03-16  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>
	
	* __scm.h (GC_FREE_SEGMENTS): Disable this until we have made
	freeing of segment work with the new GC scheme.  (Thanks to
	Michael Livshin.)  Oops, also happened to make GUILE_NEW_GC_SCHEME
	the default, but I'll let this change stay in CVS Guile since this
	code is not expected to contain serious bugs.
	
2000-03-16  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* gc.c, gc.h (scm_map_free_list): Define also if GUILE_DEBUG is
	defined.
	(scm_free_list_length): New procedure (GUILE_DEBUG).
	Fixed a small but serious bug introduced by the previous change.

	* gc.c (scm_gc_sweep): Moved variable n_objects to inner sweep
	loop and declare as register.

	* gc.c (scm_gc_sweep): Sigh... forgot to clear private freelists
	after GC.

Wed Mar 15 08:27:04 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* fluids.c: Docstring patch from Keisuke Nishida.  Some
	reindentation, too, and a couple formals renamed.  Should
	fluid-set! return UNSPECIFIED instead of a value?

Wed Mar 15 08:24:58 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* Makefile.am: Separate out DOT_X_FILES and DOT_DOC_FILES, and
	generate the latter from the concrete listing of the former.  Then 
	make guile-procedures.txt depend on DOT_DOC_FILES instead of
	*.doc, so that rebuilding it works.

Wed Mar 15 08:12:14 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* libguile.h: Include libguile/validate.h.  Thanks Keisuke Nishida!

	* guile-snarf.awk.in: Replace docstring line-ending \n" and \n\n"
	with nothing and \n, respectively. Thanks Keisuke Nishida for
	noticing this problem.

2000-03-15  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* __scm.h (GUILE_NEW_GC_SCHEME): Define this if you want to test a
	new way of allocating heap.  It makes Guile fast, but still
	contains bugs.

	* gc.c, gc.h, pairs.h, init.c: Implementation of a new way of
	allocating heap.  The basic idea is to trigger GC every Nth
	allocated cell and grow heap when free list runs out.  The scheme
	has been extended so that GC isn't triggered until all remaining
	cells are used.  The implementation is also prepared for
	development in the direction of POSIX threads.

	* gc.c (SCM_EXPHEAP): In order to grow by a factor of 1.5,
	SCM_EXPHEAP should return half of the heap size.

2000-03-14  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	The following change to init.c is only enabled if Guile was
	configured with --enable-guile-debug.
	
	* init.c (scm_i_getenv_int): New function.
	(scm_boot_guile_1): Use the environment variables
	GUILE_INIT_HEAP_SIZE, GUILE_INIT_HEAP_SIZE2 to select heap size if
	they exist.  (This may be replaced by a Scheme level mechanism in
	the future.)

	* objprop.c (s_scm_set_object_property_x): Use scm_assq instead of
	scm_assoc.  (Thanks to Keisuke Nishida.)

2000-03-14  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* eval.c, lang.c, lang.h (scm_lisp_nil, scm_lisp_t): Renamed from
	scm_nil, scm_t.  (Thanks to Keisuke Nishida.)

2000-03-14  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* init.c (scm_boot_guile_1): Use same initial segment size for
	1-word and 2-word segments.  Having the smaller size causes Guile
	to GC too often.  Obviously something needs to be done to allow
	for a smaller 2-word segment without this to happen.  (The amount
	of heap for each type should be automatically adapted to the
	application somehow.)

	[Almost all of these changes should be documented in the NEWS
	file.]

	* gc.h (scm_freelist_t): New type.

	* pairs.h (SCM_NEWCELL, SCM_NEWCELL2): Use new style freelists.

	* gc.c (SCM_INIT_HEAP_SIZE): Changed from 32768 --> 40000 so that
	all of Guile basics fits into one segment and there suitable room
	for work.
	(SCM_EXPHEAP): Now takes an argument.  Grow by a factor of 1.5
	instead of 2.
	(scm_freelist, scm_freelist2): Now of type scm_freelist_t.
	Freelists now contains information about object span, cells
	collected and amount of cells in heap segments belonging to the
	list.
	(scm_heap_size, scm_gc_cells_collected): Removed.

	* init.c (scm_boot_guile_1): Make 2-word segment 8K (512 cells).

	* Makefile.am (libguile_la_LDFLAGS): Bumped library version
	number.

	* __scm.h eq.c, eval.c, gc.c, gc.h, gh_data, hash.c, numbers.c,
	numbers.h, objects.c, ramap.c, random.c, unif.c, unif.h: Extensive
	rewrite of handling of real and complex numbers.
	(SCM_FLOATS, SCM_SINGLES): These #ifdef conditionals have been
	removed along with the support for floats.  (Float vectors are
	still supported.)

	* tags.h (scm_tcs_bignums): Removed.
	(scm_tc16_bigpos, scm_tc16_bigneg): Replaced by scm_tc16_big.
	Use SCM_BIGSIGN(x) to test for sign!
	(scm_tc16_big): The new bignum type.
	(SCM_REAL_PART, SCM_IMAG_PART): Removed.

	* numbers.h (SCM_BIGSIGN): Sign moved to bit 16.
	(scm_makdbl): Deprecated.
	(SCM_NEWREAL, SCM_NEWCOMPLEX): New macros.
	(SCM_SINGP): Deprecated.
	(SCM_FLO): Removed.
	(SCM_INEXP, SCM_CPLXP): Deprecated.
	(SCM_INEXACTP, SCM_COMPLEXP): New macros.
	(SCM_COMPLEX_REAL, SCM_COMPLEX_IMAG): Renamed from
	SCM_REAL, SCM_IMAG (and now only valid for complex numbers).
	(SCM_REAL, SCM_IMAG): New, *deprecated*, selectors which work both
	for doubles and complex numbers.
	(SCM_REAL_VALUE): New selector for doubles.
	(scm_double_t, scm_complex_t): New types.
	(scm_dbl): Removed.

	* numbers.c (scm_floprint, scm_floequal): Removed.
	(scm_print_real, scm_print_complex, scm_real_equalp,
	scm_complex_equalp): New functions.

	* unif.c (scm_makflo): Removed.

	* smob.h (SCM_SMOB_PREDICATE): New macro.
	(SCM_NEWSMOB2, SCM_RETURN_NEWSMOB2, SCM_NEWSMOB3,
	SCM_RETURN_NEWSMOB3): New macros.

1999-11-21  Michael Livshin  <mlivshin@bigfoot.com>

	The following changes implement primitive support for double cells
	(i.e. four-word cells) and change the representation of some
	things to multi-cells instead of cons+malloc.  (Applied and
	modified by mdj.)

	* pairs.h (SCM_NEWCELL2): double-cell variants of SCM_NEWCELL.
	(SCM_CELL_WORD, SCM_CELL_WORDLOC, SCM_SET_CELL_WORD): primitive
	multi-cell access macros (used by the ones below).
	(SCM_CELL_WORD[0-3], SCM_SET_CELL_WORD[0-3]): multi-cell access
 	macros.

	* gc.c (scm_freelist2): multi-cell freelists.
	(inner_map_free_list): map_free_list, parameterized on ncells.
  	"nn cells in segment mm" was misleading for ncells > 1; changed to
 	"objects".  still print cells too, though.
	(scm_map_free_list): rewritten using inner_map_free_list.
	(scm_check_freelist): get freelist as parameter, since now we have
 	more than one.
	(scm_debug_newcell2): multi-cell variants of
 	scm_debug_newcell.
	(scm_gc_for_newcell): take ncells and freelist pointer as
 	parameters.
	(scm_gc_mark): add case for tc7_pws (procedures with setters are
 	now double cells).
	(scm_gc_sweep): don't free the float data, since it's not malloced
 	anymore.
	(init_heap_seg): didn't understand what n_new_objects stood for,
 	so changed to n_new_cells.
	(make_initial_segment): new function, makes an initial segment
 	according to given ncells.
	(scm_init_storage): call make_initial_segment, for ncells={1,2,3}.

	* numbers.c (scm_makdbl): no malloc'ing needed, so the
 	{DEFER,ALLOW}_INTS thing removed.

	* numbers.h (struct scm_dbl): changed to represent a double cell,
 	with the number in the second half.

	* dynwind.c: changed the wind-guards representation to double
 	cell.

	* procs.c, procs.h: changed the procedure-with-setter representation
	to double cell.

	* async.c, async.h: made async representation a double cell.

	* dynl.c: made dynamic_obj representation a double cell.
	
2000-03-13  Gary Houston  <ghouston@arglist.com>

	* ports.c (flush_void_port): renamed to flush_port_default.
	(end_input_void_port): renamed to end_input_default.

	* init.c (scm_standard_stream_to_port): create a void port instead
	of opening /dev/null if the standard file descriptors are bad.
	advantages: no portability problems, doesn't waste a file descriptor,
	simplifies the code (thanks to Marius for the idea).

	* vports.c (s_scm_make_soft_port): call scm_port_non_buffer.

	* void ports: make reading from a void port give EOF instead of
	segv:
	* ports.c (s_scm_sys_make_void_port): modified docstring.
	(fill_input_void_port): new proc.
	(scm_init_ports): set up fill_input_void_port.
	* ports.c (scm_port_non_buffer): new proc.
	(scm_void_port): call scm_port_non_buffer.

	* fports.c (scm_setvbuf): docstring: remove the fcntl documentation
	which was incorrectly appended.

2000-03-13  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* guile-doc-snarf.in: Don't use absolute path for `sed'.  (Note
	that we can't use autoconf for this.  Autoconf itself relies on
	the existence of `sed' somewhere on your path.)  (Thanks to Dirk
	Herrman.)
	
2000-03-13  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* Makefile.am (libguile_la_SOURCES): Moved iselect.c here from
	EXTRA_libguile_la_SOURCES.

	* iselect.h: Always declare scm_internal_select.

	* iselect.c (scm_internal_select): Added SCM_ASYNC_TICK at the
	end.  Also let scm_internal_select be a real function also when
	not using threads.

	* __scm.h (SCM_TICK): Oops! Forgot to put SCM_ASYNC_TICK here...

2000-03-13  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* __scm.h (SCM_ALLOW_INTS, SCM_REALLOW_INTS): Removed call to
	SCM_ASYNC_TICK.  (This is a preparation for POSIX threads support,
	and kind of an experiment: Will this cause problems?)

Sun Mar 12 13:26:30 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* Makefile.am: Added *.doc to DISTCLEANFILES.

2000-03-12  Gary Houston  <ghouston@arglist.com>

	* fports.c (scm_fdes_to_port): call fcntl F_GETFL to test that
	the fdes is valid before doing anything else.  check that
	the file descriptor supports the modes required.
	(scm_fport_buffer_add): don't throw an error if fstat doesn't
	work: just use the default buffer size.

	* throw.c: change an outdated comment about scm_internal_catch
	BODY: it doesn't take a jumpbuf arg.

	* init.c (scm_standard_stream_to_port): install a handler in case
	scm_fdes_to_port throws an error.  don't check here whether the
	file descriptor is valid, since scm_fdes_to_port will do that.
	set the revealed count depending on whether the port got the
	standard file descriptor.
	(stream_body_data): new type.
	(stream_body, stream_handler): new procs.

2000-03-12  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* stacks.c, stacks.h, struct.c, tags.h, unif.c (scm_bits_t):
	Renamed from SCMWORD.

	* tags.h (SCM_NCELLP): Removed (SCMWORD).

	* arbiters.c (SCM_ARB_LOCKED): Use SCM_UNPACK_CAR.

	* async.c, boolean.h, debug.c, dynl.c, dynwind.c, eval.c, eval.h,
	  feature.h, filesys.h, fluids.h, fports.c, fports.h, gc.c, gc.h,
	  hash.c, keywords.h, macros.c, numbers.c, numbers.h, objects.c,
	  objects.h, options.c, pairs.h, ports.c, ports.h, print.c,
	  procs.h, ramap.c, read.c, smob.c, smob.h, srcprop.h, stacks.c,
	  stacks.h, strports.c, struct.c, struct.h, tag.c, tags.h,
	  throw.c, unif.c, unif.h, variable.h, vectors.h, weaks.c,
	  weaks.h (SCM_PACK, SCM_UNPACK, SCM_UNPACK_CAR): Renamed from
	  SCM_ASSCM, SCM_ASWORD, SCM_CARW).

	* numbers.h (SCM_SRS, SCM_INUM): Corrected SCM_ASSCM/ASWORD fixes.

	* alist.c, eval.c, net_db.c, posix.c, print.c, snarf.h, struct.c,
	tags.h: Fixed copyright notices.

	* struct.c, coop-threads.c: SCM_ASSCM/ASWORD fixes.

2000-03-12  Marius Vollmer  <mvo@zagadka.ping.de>
	
	* init.c (scm_standard_stream_to_port): Check whether the file
	descriptor is valid and substitute "/dev/null" when not.
	
2000-03-12  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* coop-defs.h (struct timespec): Conditionally defined.

	* coop.c (coop_condition_variable_timed_wait_mutex): Use ETIMEDOUT
	instead of ETIME.

	* readline.c (match_paren): Bugfix: First arg to select is not
	number of descriptors but the number of the highest descriptor +
	1.

	This is a preliminary attempt at a cleanup of the threads support
	code.  It moves things to better places, makes arguments more
	consistent with the POSIX API (which is used in GNOME's glib), and
	adds new functionality.
	
	* readline.c (scm_init_readline): Added new arg to scm_init_mutex.

	* coop-defs.h (scm_mutex_trylock): New macro: alias for
	coop_mutex_trylock.
	(scm_cond_init): Changed definition to
	coop_new_condition_variable_init.
	
	* coop.c: #include <errno.h>
	(coop_timeout_qinsert): Moved here from iselect.c
	(coop_new_mutex_init, coop_new_condition_variable_init): New
	functions.  The strange names are temporary.  Use scm_mutex_init
	and scm_cond_init instead.
	(coop_mutex_trylock): New function.  Uses errno.h:EBUSY.  errno.h
	is ANSI C, but should we check for individual error codes in
	configure.in?
	(coop_condition_variable_timed_wait_mutex): New function.
	(coop_key_create, coop_setspecific, coop_getspecific,
	coop_key_delete): New functions.

	* iselect.c (coop_timout_qinsert): Moved to coop.c

2000-03-11  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* pairs.h (SCM_SETAND_CAR, SCM_SETAND_CDR, SCM_SETOR_CAR,
	SCM_SETOR_CDR): Corrected SCM_ASSCM/WORD fixes.

	* tags.h (SCM_VOIDP_TEST): Renamed from VOIDP_TEST.
	Layout cleanups.

	* objects.h (SCM_CLASS_FLAGS, SCM_OBJ_CLASS_FLAGS,
	SCM_SET_CLASS_INSTANCE_SIZE), struct.h (SCM_STRUCT_VTABLE_DATA),
	proc.h (SCM_CLOSCAR): SCM_ASSCM/WORD fixes.

	* eval.c (scm_lookupcar1): Inserted SCM_ASWORD in expressions
	dealing with ilocs.

2000-03-11  Dale P. Smith  <dpsm@en.com>, applied by Greg J. Badros, <gjb@cs.washington.edu>

        * numbers.c (scm_copy_big_dec, scm_copy_smaller, scm_big_ior,
        scm_big_xor, scm_big_and, scm_big_test): Added new lowlevel bignum
        logical functions from SCM.

        (logand, logior, logxor, logtest, logbit?): Extended scheme
        logical functions to use bignums from SCM.

        (lognot): Removed call to `SCM_VALIDATE_INUM' that prevented
        lognot from using bignums.

Thu Mar  9 11:33:25 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* vectors.h (SCM_VELTS_AS_STACKITEMS): Added this macro to help in 
	eliminating some warnings.

	* unif.c, strports.c, print.c, options.c: Fix some warnings on
 	mis-use of SCM/long

	* gc.c, gc.h: Added scm_return_first_int(), and added comment re:
 	what the scm_return_first* functions do.

2000-03-09  Han-Wen Nienhuys  <hanwen@cs.uu.nl>, applied by Greg J. Badros, <gjb@cs.washington.edu>

       * libguile/*.[ch]: make a distinction between SCM as a generic
       name for a Scheme object (now a void*), and SCM as 32 bit word for
       storing tags and immediates (now a long int).  Introduced
       SCM_ASWORD and SCM_ASSCM for conversion. Fixed various dubious
       code in the process: arbiter.c (use macros), unif.c (scm_array_p),
       
Wed Mar  8 10:15:59 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* numbers.c:  Use SCM_VALIDATE_LONG_COPY, and longs, not ints, in
	various logXXX primitives.  Thanks Eric Moore!

Tue Mar  7 08:05:22 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* run-test, remaining-docs-needed:  Added these scripts.  The
	second one is only temporary until the docstring additions are
	complete.   run-test may best live on, but is here mostly for
	convenience and awareness for now.

	* hash.c: Docs, minor cleanup patch from Dirk Herrman.

Thu Mar  2 16:06:58 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* error.h, error.c: Added `scm_wrong_type_arg_msg' to support
 	displaying the expected type.  Use SCM_LISTn in a couple places
	instead of scm_cons-ing by hand.

	* __scm.h: Added SCM_ASSERT_TYPE macro.

	* validate.h, scm_validate.h: Added the former, as a renamed
	version of the latter with SCM_ASSERT_TYPE used in
	SCM_MAKE_VALIDATE (instead of just SCM_ASSERT)

	* Makefile.am: Rename scm_validate.h to validate.h.

	* *.c, *.h: Include validate.h, not scm_validate.h (old name's
	prefix was superfluous).

Thu Mar  2 15:33:12 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* hashtab.c: Improved documentation for lots of functions.  Added
	handwritten docs for `hash-fold'.

Thu Mar  2 15:13:25 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* list.c: Added hand-written docs for `del{q,v,ete}1!'.

Thu Mar  2 12:38:30 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* list.c: Moved append docs to append! Thanks Dirk Hermann.  Also, 
	added append docs from R4RS.

	* strings.c: Docstring typo fix, + eliminate unneeded IMP tests.
	Thanks Dirk Hermann!

	* chars.h: Provide SCM_CHARP, SCM_CHAR, SCM_MAKE_CHAR and
	deprecate SCM_ICHRP, SCM_ICHR, SCM_MAKICHR.  Thanks Dirk Hermann!
 
	* *.h, *.c: Use SCM_CHARP, SCM_CHAR, SCM_MAKE_CHAR throughout.
	Drop use of SCM_P for function prototypes... assume an ANSI C
	compiler.  Thanks Dirk Hermann!

Sat Feb 19 12:20:12 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* ports.c: Made `set-port-column!' and `set-port-line!' each
	return SCM_UNSPECIFIED instead of a (not-scheme-object) integer
	that caused a seg fault.  Also fixed `set-port-column!'s
	docstring.  Thanks Han-Wen Nienhuys for finding the bug!

Sun Feb 13 19:11:42 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* arbiters.c, eq.c, gc.c, guardians.c, list.c, ports.c, print.c,
	regex-posix.c, scmsigs.c, stime.c, strings.c, variable.c, stime.c, 
	strings.c, variable.c:  Added lots of documentation, cleaned up
	some existing documentation.  Occasionally changed formal params
	to match docs.  Also folded an #ifdef into the inners of a
	primitive instead of having two copies of the primitive
	(`get-internal-real-time', from stime.c)

Sun Feb 13 18:12:19 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* ports.c: Added docs for primitives missing them.  Written by
	hand.

Sun Feb 13 09:40:36 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* guile-doc-snarf.in: Use ${AWK} -f guile-func-name-check, not
	just execing guile-func-name-check.  Thanks Michael Livshin!

Thu Feb 10 11:43:23 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* guile-snarf.awk.in: Tweak to work with Sun/HP awk, removed some
	dead code.  Patch from Michael Livshin.

	* guile-doc-snarf.in: Tweak to work with Sun/HP sh.  Patch from
	Michael Livshin.

2000-02-09  Gary Houston  <ghouston@arglist.com>

	* init.c (scm_init_standard_ports): when stdout is a tty, make the
	current-output-port unbuffered by default.  this is less confusing
	for interactive use.  it was line-buffered because of a
	performance problem with unbuffered ports, but I think it will be
	OK now.

2000-02-08  Gary Houston  <ghouston@arglist.com>

	* __scm.h: don't define long_long or ulong_long if HAVE_LONG_LONGS
	is not defined.

	* stime.c (scm_localtime, scm_mktime): if neither HAVE_TM_ZONE nor
	HAVE_TZNAME are defined, use an empty string instead of giving two
	spurious compile-time errors.

Tue Feb  8 13:57:46 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* ports.c: Doc patches from Richard Kim.  Pasted from MIT Scheme.
	Thanks Richard!

Mon Feb  7 09:07:31 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* macros.c: Doc patches from Richard Kim.  Pasted from scm.texi.

Sun Feb  6 20:26:21 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* pairs.c: Doc patches from Richard Kim.  Pasted from MIT Scheme
	(which is GNU GPL'd).

2000-01-31  Gary Houston  <ghouston@arglist.com>

	* strings.h: don't use SCM_P.  don't include <string.h>.
	* error.c, gh_data.c, ports.c, script.c, strop.c: include <string.h>.
	
	* strings.c (scm_string_ref): make the 2nd argument compulsory.
	previously it defaulted to zero for no good reason that I can see.
	use a local variable for SCM_INUM (k).  replace
	SCM_VALIDATE_INUM_DEF with SCM_VALIDATE_INUM_COPY.

	(scm_makfromstr): cosmetic changes.

	(scm_string): Accept only chars in the list, not strings, for
	conformance to R5RS (particularly for list->string, which is
	supposed to be the inverse of string->list.)  remove
	SCM_DEFER_INTS/SCM_ALLOW_INTS, which is unnecessary since
	scm_makstr handles the cell allocation.  when reporting wrong-type
	arg, don't report the position as 1.

	* posix.c (scm_init_posix): intern PIPE_BUF if it's defined.

2000-01-29  Gary Houston  <ghouston@arglist.com>

	* posix.c (scm_pipe): rewrote the docstring.

	* filesys.c (scm_select, retrieve_select_type, get_element,
	fill_select_type, set_element): modified so that Scheme
	"select" tests port buffers for the ability to provide input
	or accept output.  Previously only the underlying file descriptors
	were checked.  Rewrote the docstring.
 
Thu Jan 27 10:14:25 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* vectors.c, symbols.c, strorder.c: Documentation cut and pasted
 	from Gregg Reynolds.  Thanks Gregg!

Thu Jan 27 09:59:38 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* strop.c (scm_i_index): Obfuscated commented-out SCM_DEFINE by
	adding "x" prefix to the line so that guile-func-name-check
	doesn't complain unnecessarily.

Wed Jan 26 17:33:52 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* throw.c: Factor out an #ifdef/#else/#endif choice more finely
	for maintainability.

	* strop.c: Documentation added by Gregg A. Reynolds.   Pasted in
	from qdocs, RnRs.

Wed Jan 26 10:02:11 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* tag.c:  Added doc for `tag', but mark as deprecated since Mikael 
	suggests removing tag.c altogether (and using a new `class-of'
	instead).

	* strings.c:  Added documentation from Gregg A. Reynolds.  Edited
	a bit by me to use FOO instead of @var{foo} and to have the
	summary come before preconditions on input.  Also dropped trailing 
	(rnrs) note.

	* gsubr.c: Do not use SCM_DEFINE for `gsubr-apply'. Register the
 	function with scm_make_subr_opt w/ last arg of 0 so it is not
 	visible at the Scheme level. Mikael says that this is the right
 	thing because the first arg to the proc is the guts of a compiled
 	closure and shouldn't be exposed to the Scheme level.

Tue Jan 25 17:15:47 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* sort.c: typo in comment fixed.

	* keywords.c: Added documentation.

	* guardians.c: Added documentation (could be better).

	* gc.c: Added docs for gc-set-debug-check-freelist.

	* eq.c: Added docs for eq?, eqv? equal? abridged from R4RS.

	* boolean.c: Added docs for `not', `boolean?' (by hand).
	
Tue Jan 25 13:28:56 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* random.c: Added documentation, from SLIB page:
	http://angela.ctrl-c.liu.se/~calle/scheme/slib_toc.html

Mon Jan 24 17:50:20 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* variable.c, version.c:  Added documentation, written by hand
	since I could not find anything already written that was
	relevant.

2000-01-23  Gary Houston  <ghouston@arglist.com>

	* filesys.c (scm_chown): omit port/fdes support if HAVE_FCHOWN is 
	not defined (thanks to Richard Y. Kim).

Thu Jan 20 13:00:38 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* Makefile.in: Removed, this is auto-generated.

Thu Jan 20 11:33:47 2000  Dirk Hermann <dirk@ida.ing.tu-bs.de> --applied 01/20/00 gjb

	* list.c: Put some variable initialization code at the point of
	declaration; Added a comment for list*; Formatting changes.

	* load.c: use SCM_NNULLP to make sure the end of a list is not
	reached yet.

2000-01-18  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* backtrace.c (scm_display_error_message): Bugfix: Don't use
	result of scm_list_p as C boolean.
	(scm_display_error_message, scm_set_print_params_x): Use new
	validation macros. (Thanks to Dirk Herrmann.)

	* net_db.c (scm_resolv_error): Cast result from hstrerror.

	* strports.c (st_end_input): Inserted parenthesis to get operator
	grouping correct.

	* list.h (scm_init_list): Removed SCM_P around prototypes.

	* fports.c, list.c, numbers.c, ports.c, stime.c, symbols.c,
	filesys.c, posix.c: Converted docstrings to ANSI C format and
	escaped " occurring inside string literals.

Tue Jan 18 13:21:08 2000  Mikael Djurfeldt  <mdj@r11n07-s.pdc.kth.se>

	* posix.c (scm_mknod): Escape " occuring inside docstring.

2000-01-18  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* alist.c, chars.c, debug.c, dynl.c, dynwind.c, error.c, eval.c,
	   evalext.c, filesys.c, gc.c, hash.c, hashtab.c, ioext.c,
	   keywords.c, list.c, load.c, macros.c, net_db.c, numbers.c,
	   objprop.c, ports.c, posix.c, print.c, procprop.c, procs.c,
	   ramap.c, regex-posix.c, root.c, scmsigs.c, simpos.c, socket.c,
	   stacks.c, stime.c, strings.c, strop.c, strports.c, struct.c,
	   symbols.c, throw.c, unif.c, vectors.c, version.c, vports.c,
	   weaks.c: Converted docstrings to ANSI C format.
	
	* filesys.c (scm_chmod), simpos.c (scm_system), version
	(scm_version), vports (scm_make_soft_port): Escape " occuring
	inside docstring.

Mon Jan 17 11:41:22 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* scm_validate.h: Added SCM_VALIDATE_ULONG_COPY,
	SCM_VALIDATE_LONG_COPY

	* numbers.c: Use SCM_VALIDATE_ULONG_COPY, instead of
	SCM_VALIDATE_INUM_COPY to let bigger numbers be used.  Rename a
	couple of formal arguments (and fix their uses) to make arguments
	match the documentation.

2000-01-14    <mstachow@alum.mit.edu>

	* Makefile.am: Augment path when running guile-doc-snarf so
	guile-func-name-check is found.

Fri Jan 14 09:34:55 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* scm_validate.h (SCM_NUM2LONG_DEF): Fix this macro to just use
	def, not SCM_MAKINUM(def);  thanks Janis Bzerins!

Wed Jan 12 00:06:53 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* net_db.c (s_scm_inet_makeaddr): Use SCM_NUM2ULONG since that's
	the way guile-1.3.4 worked, but #if 0 out the version using
	SCM_VALIDATE_INUM_COPY for stricter testing.

Tue Jan 11 18:24:18 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* guile-doc-snarf.in: Use new $fullfilename for running
	guile-func-name-check, and put "$fullfilename" and "$filename" in
	quotes at uses to make sure re-splitting on whitespace does not
	occur (so filenames w/ embedded whitespace would work okay, though 
	I sure hope we never have to deal with that! :-) ).  Thanks to
	Mikael for pointing out the source_dir != build_dir was broken.

Tue Jan 11 17:42:40 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* scm_validate.h: Added SCM_NUM2LONG_DEF macro.  Make
	SCM_OUT_OF_RANGE use SCM_MAKINUM, not scm_long2num.  Added
	SCM_COERCE_ROSTRING macro.  Added SCM_VALIDATE_NONEMPTYLIST
	macro.  Fix SCM_VALIDATE_STRINGORSUBSTR macro to not use SLOPPY
	variants.

	* ports.c (scm_port_closed_p): Validate that the arg is a PORT,
	then return whether it's an open port (was validating that it was
	an open port -- this was a bug I introduced back in December, but
	my careful reading of diffs caught it).

	* numbers.c: Recombine the two conditional-compilation paths for
	all the log* primitives -- they were split based on #ifndef
	scm_long2num;  factored out a SCM_LOGOP_RETURN macro, and fixed
	some bugs and inconsistencies in the two sets of implementations.
	(scm_lognot) Fixed *atrocious* re-use of a SCM as an integer!

	* ioext.c: Use SCM_ASSERT_RANGE in a couple places, and
	SCM_VALIDATE_INUM_COPY once where it should've been used.

	* fluids.c (scm_internal_with_fluids): Use
	SCM_VALIDATE_LIST_COPYLEN.

	* filesys.c: Use SCM_NUM2LONG instead of SCM_VALIDATE_INUM_COPY;
	this is questionable as it relaxes type safety, but other changes
	were useful and all SCM_NUM2LONG's should probably be
	revisited. Use SCM_OUT_OF_RANGE, SCM_WRONG_TYPE_ARG.

	* evalext.c: line-break change on 1 line.

	* eval.c (nconc2last): Takes a non-empty list as its first
	argument, not just a list.

	* dynl.c: Use new SCM_COERCE_ROSTRING macro.

Tue Jan 11 15:44:23 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* dynl.c, feature.c, filesys.c, fports.c, list.c, load.c,
 	net_db.c, sort.c, stacks.c, unif.c: Use SCM_WTA, SCM_MISC_ERROR
 	where possible.

	* symbols.c (scm_sysintern0): Fixed the function name in a
	scm_misc_error invocation.

	* print.c (scm_simple_format): Do not need SCM_COERCE_SUBSTR, and
	use scm_return_first to ward off latent GC bug that Mikael caught.

	* async.c: Use SCM_VALIDATE_ASYNC_COPY one place where it wasn't
	used before but should've been.

2000-01-12  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* snarf.h (SCM_PROC1): Replaced SCM (*) (...) with
	SCM_FUNC_CAST_ARBITRARY_ARGS.

Tue Jan 11 13:44:07 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* guile-func-name-check.in: Added this script to statically check
	#define FUNC_NAME, #undef FUNC_NAME in the source.

	* sort.c, posix.c: Fix #undef FUNC_NAME lines to not have trailing
	redundant comment, semicolon; caught by new guile-func-name-check
	script.

	* debug.c: Fix mistaken #define FUNC_NAME for scm_make_iloc.
	Caught by new guile-func-name-check-script.

	* Makefile.am: Added guile-func-name-check to bin_SCRIPTS

	* ramap.c: Fix #if 0'd out code to be syntactically acceptable to
	guile-func-name-check.

	* guile-doc-snarf.in: Run guile-func-name-check on the file before 
	doing the snarf.

Tue Jan 11 11:31:10 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* fports.c, ports.c, ports.h, strports.c, vports.c: Make write
 	port function take const void*, not void*.

Tue Jan 11 11:18:07 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* scm_validate.h, chars.c, ports.c, print.c, read.c, strings.c,
 	strop.c: Use SCM_VALIDATE_ICHR, SCM_VALIDATE_ICHR_COPY instead of
 	SCM_VALIDATE_CHAR, SCM_VALIDATE_CHAR_COPY.  Change made for
 	consistency with the other macros dealing with immediate
 	characters.  (Similar to INT -> INUM change a week or so ago).

Tue Jan 11 10:41:46 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* dynl.c, error.c, eval.c, feature.c, filesys.c, fports.c, list.c, load.c,
 	net_db.c, read.c, socket.c: Update error messages to use ~A for
 	%s, ~S for %S to work with new `simple-format' format and be
 	standardized better.

	* print.h, print.c (scm_simple_format): Added `simple-format'
	primitive.  It's the old scm_display_error, with ARGS now a rest
	parameter, and the destination first instead of last (and a couple 
	new capabilities inspired by `format' -- #t as destination means
	current-output-port, #f means return the formatted text as a
	string.

	* gh.h, gh_data.c, ports.h, ports.c: Added some missing const specifications.

	* backtrace.c (scm_display_error_message): Rewrote to use
	scm_simple_format() procedure.

	* __scm.h: Added commented-out #define of GUILE_DEBUG_FREELIST

2000-01-09  Marius Vollmer  <mvo@zagadka.ping.de>

 	Finally applied the libltdl patch from Thomas Tanner, with slight
	modifications.

	* DYNAMIC-LINKING: Removed because it is obsolete.
	* dynl.c: Use ANSI prototypes.
	(sysdep_dynl_link): Use lt_dlopenext instead of lt_dlopen.
	* scmconfig.h.in: Do not change, as it is automatically generated.
	
    1999-07-25  Thomas Tanner  <tanner@ffii.org>

	    * dynl-dl.c, dynl-dld.c, dynl-shl.c, dynl-vms.c: deleted
	    (obsolete)
	    * Makefile.am: likewise, add INCLTDL (libltdl headers) to
	    INCLUDES, set dlpreopened files in LDFLAGS, link libguile
	    against libltdl
	    * dynl.c: use libltdl if DYNAMIC_LINKING is enabled,
	    * guile.c: register preloaded modules
	    * scmconfig.h.in: remove obsolete symbols

2000-01-09  Gary Houston  <ghouston@arglist.com>

	* These changes should make it unnecessary to call tzset from
	Scheme after modifying the TZ environment variable, even if the
	system date facilities cache the value.
	* stime.c (setzone, scm_localtime): added comments.
	(tzset): don't define a noop tzset macro if HAVE_TZSET not defined.
	(setzone): don't call tzset.
	(restorezone): call tzset only if HAVE_TZSET is defined.
	(scm_tzset): don't define if HAVE_TZSET not defined.  Change the
	doc string to indicate that this procedure isn't likely to do
	anything useful.
	(scm_localtime, scm_strftime, scm_mktime): call tzset if
	LOCALTIME_CACHE is defined.

2000-01-09  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* posix.c (scm_sync): Return SCM_UNSPECIFIED.

2000-01-09  Gary Houston  <ghouston@arglist.com>

	* eval.c: define scm_unbound_variable_key ('unbound-variable).
	scm_lookupcar1: throw an error with key 'unbound-variable instead
	of 'misc-error when an unbound variable is encountered.

	* filesys.c (scm_mkdir, scm_rmdir, scm_getcwd, scm_select, 
	scm_symlink, scm_readlink, scm_lstat),
	posix.c (scm_setpgid, scm_setsid, scm_ctermid, scm_tcgetpgrp,
	scm_tcsetpgrp, scm_uname, scm_setlocale, scm_mknod, scm_nice,
	scm_sync),
	simpos.c (scm_system), 
	stime.c (scm_times, scm_strptime):
	move the HAVE_XXX feature tests out of the procedure bodies.
	don't use SCM_SYSMISSING.
	scm_validate.h (SCM_SYSMISSING): removed.
	error.h, error.c (scm_sysmissing): comment that this is deprecated.
	see ChangeLog entry for 1999-12-28.

Sat Jan  8 19:52:04 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* scm_validate.h (SCM_VALIDATE_BOOL_COPY): Fix typo.

Sat Jan  8 17:06:46 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* backtrace.c: Fix spelling typo in a comment.

	* snarf.h: Use new SCM_DOCS macro to encapsulate the non SCM_INIT
 	text.  Reformatted some of the expansions.

Fri Jan  7 15:50:46 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* scm_validate.h (SCM_OUT_OF_RANGE): Use scm_out_of_range_pos to
	report the position of the argument.

	* error.h, error.c (scm_out_of_range_pos):  Added this function to 
	take extra "pos" argument, the position number of the errant
	argument. 

	* debug.c: Use SCM_OUT_OF_RANGE instead of scm_out_of_range.

	* snarf.h: Use SCM_HERE and SCM_INIT as names, not SCM_NOTSNARF
	and SCM_SNARFING.  Also put the %%% in the SCM_INIT since Mikael
	prefers that and I'm reasonably indifferent.

Fri Jan  7 15:03:32 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* snarf.h: Factor out differences between C++ and non-C++ into
	SCM_FUNC_CAST_ARBITRARY_ARGS macro.  Modify all the snarf macro
	definitions to use SCM_NOTSNARF and SCM_SNARFING macros (like
	Mikael's macros, below, but changed names and SCM_SNARFING no
	longer expands to include %%% -- that must appear in the argument
	so that the token appears at the call-site as a reminder).

2000-01-07  Mikael Djurfeldt  <mdj@thalamus.nada.kth.se>

	* snarf.h (SCM_INSITU, SCM_INIT): New snarf macros for use in user
	snarf macro definitions.

2000-01-06  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* chars.c (scm_integer_to_char): Use Greg's nice
	SCM_VALIDATE_INUM_RANGE macro for argument checking for closer
	adherence to R5RS.

Thu Jan  6 11:48:49 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* *.c, snarf.h: Replace GUILE_PROC1 with SCM_DEFINE1 throughout.

Thu Jan  6 11:22:53 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* Makefile.am (ETAGS_ARGS): Switch to SCM_DEFINE, SCM_DEFINE1
	instead of GUILE_PROC.

Thu Jan  6 11:21:49 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* alist.c: Do not report mismatch errors on some uses of `tmp' (do 
	this by using SCM_ARG2 instead of `2' in the SCM_VALIDATE_CONS
	macro call.

Thu Jan  6 09:54:33 2000  Dirk Herrmann <dirk@ida.ing.tu-bs.de> --gjb applied

	* scm_validate.h: Remove some redundant NIMP tests.

	* alist.c: minimize scope of the tmp variables, and initialize
 	them when declared. The strange SCM_NIMP tests are replaced by
 	SCM_CONSP tests that more closely reflect the intended semantics.
  	However, we don't get a performance penalty here, because the
 	SCM_CONSP test was performed by the ALISTCELL test anyway.  * The
 	extremely ugly use of ASRTGO macros was removed: The calls to
 	ASRTGO were not encapsulated by "#ifndef SCM_RECKLESS", but got a
 	label parameter that only exists when SCM_RECKLESS is not defined.
  	This works, because ASRTGO itself is defined in a way that it only
 	makes use of the label parameter if SCM_RECKLESS is not defined
 	(shudder!).  Does guile make at all use of the possibility to
 	define SCM_RECKLESS?  * Codesize is likely to be reduced, since
 	instead of two calls to SCM_ASSERT performed by the ALISTCELL test
 	we now only get one test.

	* list.c: Use SCM_NNULLP, not SCM_NIMP as appropriate.  Also use
 	SCM_NULLP instead of SCM_IMP.  Drop use of "register" keyword on
 	some variables in `list?'.  Fix `reverse' and `reverse!'
 	primitives to handle improper lists better.

Wed Jan  5 11:24:53 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* *.[ch]: Whitespace changes -- added space after SCM_VALIDATE_*
 	macros and SCM_DEFINE macros to match GNU coding standards.

Wed Jan  5 11:04:24 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* *.[ch]: Replace GUILE_PROC w/ SCM_DEFINE.

Wed Jan  5 10:59:06 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* *.[ch]:  Replace SCM_VALIDATE_INT w/ SCM_VALIDATE_INUM for
	better consistency with the names of other SCM_VALIDATE_ macros
	and better conformance to guile naming policy.

Wed Jan  5 10:50:39 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* ports.c (s_scm_close_all_ports_except): Use SCM_ARG1 in a
	SCM_VALIDATE instead of 1 to avoid a check on the argument (since
	it's not the actual name of the formal).

	* guile-snarf.awk.in: Do argument/number mismatch checking and
	print warnings in an Emacs compile-mode parseable format.

	* struct.c: Use SCM_ASSERT_RANGE instead of SCM_ASSERT w/
	SCM_OUTOFRANGE as 3rd argument.

	* random.c: Fix argument/number mismatch (that I introduced :-( ).

	* __scm.h: Do not #define SCM_ARG* when snarfing;
	lets us distinguish between 1 and SCM_ARG1 when snarfing as only
	the former (using the number) requires the argument to match the
	formal in the current argument snarfing check.

	* snarf.h: Give new definition of SCM_ASSERT when in
	snarfing mode to output a lexically-identifiable sequence that the 
	guile-snarf.awk script uses to verify argument/position matching.

	* ramap.c: Remove extraneous #undef FUNC_NAME.

Wed Jan  5 08:36:38 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* guile-doc-snarf.awk.in: Removed -- guile-snarf.awk.in is the
	current version of the same functionality; it writes the .x output 
	to stdout instead of directly into the file.

Wed Jan  5 08:15:04 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* unif.c, symbols.c, strings.c, stacks.c, random.c, print.c,
	posix.c:  Eliminated a bunch of SCM_NIMP(..)s that are now
	redundant with the safer macros.  Patch from Dirk Hermann applied
	by hand. Thanks Dirk!

	* scm_validate.h: Added SCM_VALIDATE_VECTOR_OR_DVECTOR for some
	uses in random.c.

	* ramap.c: whitespace change.

Tue Jan  4 14:21:35 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* options.c, objects.c, keywords.c, gc.c: Some redundant SCM_NIMP
 	removals from Dirk Hermann.

	* alist.c: Rename formals to match the parameter names in the
	documentation, updates to documentation.  Thanks Dirk Hermann!

2000-01-04  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* eval.c (SCM_CEVAL): Reverse order of
	scm_stack_checking_enabled_p and SCM_STACK_OVERFLOW_P
	(Thanks to Brad Knotwell.)

Mon Jan  3 08:30:02 2000  Greg Harvey  <Greg.Harvey@thezone.net> (applied --01/03/00 gjb)

	* gc.c (scm_debug_newcell): Added SCM_SETCAR of the newly
	allocated cell. 

	* pairs.h: Added a comment about the need for the SCM_SETCAR in
	SCM_NEWCELL macro.

Mon Jan  3 08:25:19 2000  Greg J. Badros  <gjb@cs.washington.edu>

	* dynl-vms.c, debug.c, coop-threads.c, backtrace.c, eval.c: More
 	SCM_NIMP tests that were redundant are now eliminated.  Patches
 	from Dirk Hermann applied by hand.

The ChangeLog continues in the file: "ChangeLog-1996-1999"