* numbers.c (isfinite): Define this macro if not defined: Return a

[bpt/guile.git] / libguile / ChangeLog

1999-03-22  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* keywords.c (scm_tc16_kw): Added for backward compatibility.
	Will be removed in next release.

	* Makefile.am (modinclude_HEADERS): Added kw.h;
	
	* kw.h: New file.

	* libguile.h: #include "libguile/kw.h"

1999-03-22  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* pairs.c (scm_set_car_x, scm_set_cdr_x): Return SCM_UNSPECIFIED.

1999-03-21  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* print.c (scm_isymnames): Added #@dispatch and #@hash-dispatch.

	* hashtab.c, hashtab.h (scm_hash_fold, scm_internal_hash_fold):
	Place the table argument last.

1999-03-19  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* modules.c: #include "procprop.h"
	(scm_system_module_env_p): Assume root environment is no lookup
	closure is found.

	* debug.c, eval.c, evalext.c, gdbint.c stacks.c:
	#include "modules.h".

	* modules.c, modules.h, eval.c, eval.h (scm_env_top_level,
	scm_top_level_env, scm_system_module_env_p): Moved to modules.c.

	* eval.c, eval.h (scm_top_level_lookup_closure): Removed.

1999-03-18  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* error.c (scm_wta): Pass SCM_LIST1 (arg) instead of SCM_EOL to
	scm_misc_error when pos is a string.  This allows for dispatching
	arbitrary error messages with one argument via SCM_ASSERT:
	SCM_ASSERT (<cond>, obj, "Undigestable object: %S", <subr>);

1999-03-17  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* list.c (scm_reverse): Report an error if given a circular list
	instead of filling memory.
	* list.c (scm_reverse_x): Check args.

1999-03-14  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	Most of this batch of changes is about how to deal with extended
	types when an object system is loaded into Guile.  A nice object
	system is capable of representing Guile's types as class objects.
	
	For example, we want a regular expression to be of class <regex>.
	But regular expressions are smobs which aren't under direct
	control of the object system, so there has to be some mechanism
	which informs the object system that a new class should be created
	which can represent the smob type.  I call this a "wrapper class".
	
	* objects.c: #include "smob.h";
	(scm_class_keyword): Removed.  (Class is automatically created by
	make_smob_classes.)
	(scm_smob_class): Array of smob classes indexed by smobnum.
	(scm_make_extended_class): "Plugin" function pointer for creation
	of wrapper classes for smob and struct types.
	(scm_class_of): Handle compiled closures.  (Currently regarded as
	<procedure>.);
	Use scm_smob_class to handle smob types;
	Handle scm_tc16_bigpos, scm_tc16_bigneg, and, scm_tc16_keyword
	through scm_smob_class;
	Handle structs.

	* smob.c (scm_newsmob): Also create a wrapper class if
	scm_smob_class has been initialized.

	* smob.h (SCM_TC2SMOBNUM): New macro for conversion between tc16
	type code and smobnum.

	* struct.c: #include "alist.h", "weaks.h", "hashtab.h";
	(scm_struct_table): Weak key table with auxilliary information for
	struct types.  Currently used for names and wrapper classes.
	(scm_struct_ihashq): Hash function for structs.
	(scm_struct_create_handle): Get/create entry in scm_struct_table.
	(scm_struct_vtable_name, scm_set_struct_vtable_name_x): Procedures
	for accessing names of vtables.  The record implementation in
	boot-9.scm currently uses the setter to record the name of record
	types.  When the object system is initialized, it can use this
	information to create wrapper classes with suitable names.
	(scm_init_struct): Allocate scm_struct_table.
	(scm_alloc_struct): Don't initialize scm_struct_i_tag here.
	(struct tags are a finite resource and we might want to restrict
	the use of tags to vtables only.  E.g., Goops only uses tags for
	classes.)
	(scm_make_struct): Use scm_struct_entity_n_extra_words instead of
	magic number 5.
	(scm_struct_vtable_tag): Use scm_struct_i_tag instead of magic
	number -1.

	* struct.h (SCM_STRUCT_TABLE_NAME, SCM_SET_STRUCT_TABLE_NAME,
	SCM_STRUCT_TABLE_CLASS, SCM_SET_STRUCT_TABLE_CLASS): New macros.
	Used for access of struct table entries.
	
	* hashtab.c, hashtab.h (scm_internal_hash_fold): New function.
	(scm_hash_fold): New procedure.  Used to process all entries in a
	hash table (in no particular order).

	Argh!  For the umpteenth time I got compilation errors because of
	the "intuitive" name `kw'.  This has to have an end:
	
	* Makefile.am, init.c, libguile.h, objects.c, root.h: Replaced
	"kw" --> "keywords" everywhere.
	(I doubt that this will cause big compatibility problems since the
	application interface is unaffected.)

	* keywords.c, keywords.h: Files renamed from kw.c, kw.h.

1999-03-12  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* srcprop.c (scm_set_source_property_x): Bugfix: Convert line and
	column inums to native form.

	Improvement of backtraces: Introduces a new stack narrowing
	specifier, #t, for the inner cut.  (See further in the comments in
	stacks.c:narrow_stack ().)
	
	* procprop.c, procprop.h (scm_sym_system_procedure): New symbol.
	(Used to flag certain system procedures which shouldn't turn up in
	backtraces.)

	* eval.c (scm_sym_system_module): New symbol.  (Used to flag
	modules which aren't "user" modules and the code of which
	shouldn't turn up in backtraces.)

	* eval.c, eval.h (scm_top_level_lookup_closure): New function:
	Extract the lookup closure from an environment.
	(scm_system_module_env_p): New function: Return non-#f if MODULE
	is a system module.

	* stacks.c: #include "eval.h"; #include "procprop.h";
	(narrow_stack): Handle new narrowing specifier #t.

	* debug.c (scm_procedure_name): Use name property in the default
	case.

	* gc.c, gc.h (scm_object_address): Renamed from scm_object_addr ().

	* objects.h (scm_si_redefined, scm_si_hashsets): Shifted.

	* eval.c, procs.c, procs.h, procprop.c: Renamed getter ->
	procedure throughout.

	* print.c (scm_iprin1): Removed extraneous space when printing
	procedure-with-setters.

	Entity and operator setter slots were introduced as a complement
	to the <procedure-with-setter> type in order to support entities
	and operators with setters in a reasonable and efficient way.
	* procs.c (scm_procedure, scm_setter): Handle entity and operator
	setter slots.

	* objects.h (SCM_OPERATOR_SETTER, SCM_ENTITY_SETTER): New macros.
	(struct scm_metaclass_operator): New setter slot.

	* gc.c (scm_gc_mark): Mark struct setter slot.

	* struct.c (scm_make_struct): Allocate one word more for
	entities and initialize the new slot.

	* struct.h (scm_struct_i_setter): New constant.

1999-03-08  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* objects.h (SCM_OBJ_CLASS_REDEF): New macro: Find class slots
	directly through the instance.

	* objects.c (scm_class_of): Use SCM_OBJ_CLASS_REDEF.

	* gc.c (scm_gc_sweep): Bugfix: Look for SCM_STRUCT_F_LIGHT flag at
	scm_struct_i_flags instead of scm_vtable_index_layout!

	* list.c (scm_list_star): New procedure.

1999-02-19  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>
	
	* debug.c (scm_init_debug): Added scheme level constant
	SCM_IM_DISPATCH.

1999-02-12  Jim Blandy  <jimb@savonarola.red-bean.com>

	* __scm.h (SCM_FENCE): Fix `asm volatile' warnings for EGCS.

	* gc.c (scm_gc_sweep): Properly properly record the size of a
	freed structure.  (Thanks to Greg Harvey.)

1999-02-07  Jim Blandy  <jimb@savonarola.red-bean.com>

	* gc.c (scm_gc_sweep): Properly record the size of a freed
	structure.  (Thanks to Michael Livshin.)

1999-02-06  Jim Blandy  <jimb@savonarola.red-bean.com>

	Readline paren matching from Greg Harvey, with modifications from
	Mikael Djurfeldt:
	
	* readline.c (in_readline, reentry_barrier_mutex): Make these
	static.
	
	* readline.c: #include <sys/time.h> and "iselect.h", so we can
	control how long we're paused, and threads will run while we're
	paused.
	(match_paren, find_matching_paren, init_bouncing_parens): New
	functions.
	(scm_init_readline): Call init_bouncing_parens.
	(scm_readline_opts): Add the bounce-parens option.
	* readline.h (SCM_READLINE_BOUNCE_PARENS): New readline option.
	(SCM_N_READLINE_OPTIONS): Adjust.

1999-02-06  Jim Blandy  <jimb@zwingli.cygnus.com>

	All the below are changes from Ken Raeburn, to get Guile to use
	const where it can.

	* chars.c (scm_lowers, scm_uppers, scm_charnames, scm_charnums),
	eval.c (s_expression, s_test, s_body, s_bindings, s_variable,
	s_clauses, s_formals): Variables now const.

	* eval.c (promsmob): Now const.
	* macros.c (macrosmob): Now const.
	* smob.c (scm_newsmob): Smobfuns argument now points to const.
	(freecell, flob, bigob): Now const.

	* dynl.c (scm_make_argv_from_stringlist, scm_coerce_rostring),
	error.c (scm_error, scm_syserror, scm_syserror_msg,
	scm_num_overflow, scm_out_of_range, scm_wrong_type_arg,
	scm_memory_error, scm_misc_error, scm_wta), macros.c
	(scm_make_synt), feature.c (scm_add_feature), filesys.c
	(scm_input_waiting_p), gc.c (scm_gc_start, scm_igc,
	scm_must_malloc, scm_must_realloc), gsubr.c (scm_make_gsubr),
	numbers.c (scm_num2dbl, scm_two_doubles, scm_num2long,
	scm_num2long_long, scm_num2ulong), 
	options.c (scm_options), posix.c (scm_convert_exec_args,
	environ_list_to_c), procs.c (scm_make_subr_opt, scm_make_subr),
	ramap.c (scm_ramapc), read.c (scm_flush_ws), socket.c
	(scm_sock_fd_to_port, scm_fill_sockaddr, scm_addr_vector), stime.c
	(setzone, restorezone, bdtime2c), strop.c (scm_i_index),
	strports.c (scm_mkstrport), symbols.c (scm_intern_obarray_soft,
	scm_intern_obarray, scm_intern, scm_intern0,
	scm_sysintern0_no_module_lookup, scm_sysintern, scm_sysintern0,
	scm_symbol_value0), unif.c (scm_aind, scm_shap2ra): Argument
	indicating calling subr, error message text, reason for error,
	symbol name or feature name are now pointer to const.
	* snarf.h (SCM_PROC, SCM_PROC1): String variables are now const.

	* procs.c (scm_init_iprocs): iproc argument now points to const.
	* pairs.c (cxrs): Now const.

	* chars.h, error.h, feature.h, filesys.h, gc.h, gsubr.h, macros.h,
	numbers.h, options.h, procs.h, ramap.h, read.h, smob.h,
	strports.h, symbols.h, unif.h: Update variable declarations and
	function prototypes for above changes.

	* dynl.c, dynl-dld.c, dynl-dl.c, dynl-shl.c (sysdep_dynl_link,
	sysdep_dynl_unlink, sysdep_dynl_func): Arguments FNAME, SUBR, and
	SYMB now point to const.

1999-01-30  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* print.c (scm_iprin1): Use scm_procedure_name instead of
	scm_procedure_property for compiled closures.

	* tags.h (scm_tc7_pws): New procedure type.  Four representations
	for procedure-with-setters were considered before selecting this
	one:

	1. A closure where the CODE and ENV slots are used to represent
	the getter and a new SETTER slot is used for the setter.  The
	original getter is stored as a `getter' procedure property.  For
	closure getters, the CODE and ENV slots contains a copy of the
	getter's CODE and ENV slots.  For subr getters, the CODE contains
	a call to the subr.

	2. A compiled closure with a call to the getter in the cclo
	procedure.  The getter and setter are stored in slots 1 and 2.
	
	3. An entity (i.e. a struct with an associated procedure) with a
	call to the getter in the entity procedure and the setter stored
	in slot 0.  The original getter is stored in slot 1.

	4. A new primitive procedure type supported in the evaluator.  The
	getter and setter are stored in a GETTER and SETTER slot.  A call
	to this procedure type results in a retrieval of the getter and a
	jump back to the correct eval dispatcher.

	Representation 4 was selected because of efficiency and
	simplicity.

	Rep 1 has the advantage that there is zero penalty for closure
	getters, but primitive getters will get considerable overhead
	because the procedure-with-getter will be a closure which calls
	the getter.
	
	Rep 3 has the advantage that a GOOPS accessor can be a subclass of
	<procedure-with-setter>, but together with rep 2 it suffers from a
	three level dispatch for non-GOOPS getters:

	  cclo/struct --> dispatch proc --> getter
	
	This is because the dispatch procedure must take an extra initial
	argument (cclo for rep 2, struct for rep 3).

	Rep 4 has the single disadvantage that it uses up one tc7 type
	code, but the plan for uniform vectors will very likely free tc7
	codes, so this is probably no big problem.  Also note that the
	GETTER and SETTER slots can live directly on the heap, using the
	new four-word cells.

	* procs.c, procs.h (SCM_PROCEDURE_WITH_SETTER_P, SCM_GETTER,
	SCM_SETTER): New macros.
	(scm_procedure_with_setter_p, scm_make_procedure_with_setter,
	scm_getter, scm_setter): New procedures.

	* eval.c, print.c (scm_iprin1): Added entries for scm_tc7_pws.

	* gc.c (scm_gc_mark, scm_gc_sweep): Added case labels for
	scm_tc7_pws.

	* objects.c, objects.h (scm_class_of,
	scm_class_procedure_with_setter): Added.

	* procprop.c (scm_i_procedure_arity), procs.c (scm_thunk_p): Added
	entry for scm_tc7_pws.

	* procs.c (scm_procedure_p): Added case label for scm_tc7_pws.

1999-01-28  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* evalext.c, evalext.h (scm_m_generalized_set_x): New memoizing
	macro.
	(scm_init_evalext): Call scm_make_gsubr for
	scm_m_generalized_set_x.

	* eval.c, debug.c, tags.h (SCM_IM_SET_X): Renamed from SCM_IM_SET.

	* eval.h: Declare scm_s_set_x, scm_sym_set_x;

	* eval.c: Renamed "set" --> "set_x" in various names for
	consistency of name correspondence between Scheme and C;
	Renamed scm_i_set_x --> scm_sym_set_x and made global.
	Renamed s_set_x --> scm_s_set_x and made global.

1999-01-26  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* random.c (scm_i_random_bignum): Made independent of endianness.

1999-01-22  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* eval.c (SCM_CEVAL): Added ENTER_APPLY in code for SCM_IM_APPLY.
	(Thanks to Eric Hanchrow.)

1999-01-21  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* objects.c, objects.h (SCM_CLASS_REDEF): Renamed from CLASS_REDEF.

	* random.c: Bugfix: Retrieve and store most significant 32 bits in
	different order if the machine is bigendian.
	(scm_init_random): Added safety check for bignum digit size.

1999-01-21  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* random.c, random.h (scm_i_make_rstate): New function: Makes
	scm_rstate from seed.
	(scm_copy_random_state, scm_seed_to_random_state): New functions.
	(scm_make_random_state): Removed.

	* random.c (scm_make_random_state): Use scm_i_make_rstate().

1999-01-21  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* random.c: Bugfix: Retrieve and store most significant 32 bits in
	different order if the machine is bigendian.
	(scm_init_random): Added safety check for bignum digit size.

1999-01-11  Roland Orre  <mdj@mdj.nada.kth.se>

	* sort.c (scm_merge, scm_merge_list_x): Bugfix: Place elements
	from first arg before equal elements in second arg in result.
	(scm_merge_list_step): Bugfix: Don't presume that arguments in a C
	function call are executed in a well defined order.

1999-01-11  Jim Blandy  <jimb@zwingli.cygnus.com>

	* alloca.c (alloca): Cast value returned by malloc.  (Thanks to
	Christian Lynbech.)

1999-01-11  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* random.c: Removed alloca includes.

1999-01-11  Jim Blandy  <jimb@zwingli.cygnus.com>

	* Makefile.in: This changed, for some reason I don't really
 	understand, when I ran automake in the top level directory.  This
 	may be contamination by Cygnus internal releases.  If you re-run
 	automake and this change gets reverted, don't worry about it.

1999-01-10  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* sort.c, sort.h: New files: Implement slib's and scsh's sort
 	interfaces.  Author: Roland Orre.

	* Makefile.am: Added sort.c, sort.h, sort.x.

	* init.c: #include "sort.h";
	(scm_boot_guile_1): Call scm_init_sort ().

	* numbers.h: Added #include "libguile/print.h".

	* numbers.c: Formatted according to Guile conventions;
	Renamed s_bignum --> scm_s_bignum.

	* random.c, random.h: New files: Random number support.
	Interface same as slib's.

	* Makefile.am: Added random.c, random.h, random.x.
	
	* init.c: #include "random.h";
	(scm_boot_guile_1): Call scm_init_random ().

1998-12-23  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* Makefile.am: New files: guardians.c, guardians.x, guardians.h

	* guardians.c, guardians.h (scm_make_guardian,
	scm_guardian_gc_init, scm_guardian_zombify, scm_guard,
	scm_get_one_zombie, scm_init_guardian): This is an implementation
	of guardians as described in R. Kent Dybvig, Carl Bruggeman, and
	David Eby (1993) "Guardians in a Generation-Based Garbage
	Collector" ACM SIGPLAN Conference on Programming Language Design
	and Implementation, June 1993 ftp://ftp.cs.indiana.edu
	/pub/scheme-repository/doc/pubs/guardians.ps.gz
	Author: Michael	N. Livshin.

	* gc.h (SCM_MARKEDP, SCM_NMARKEDP): New macros.

	* gc.c (scm_igc): Call scm_guardian_gc_init and
	scm_guardian_zombify.

	* init.c (scm_boot_guile_1): Call scm_init_guardian.

1998-12-19  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* macros.c (scm_makacro, scm_makmacro, scm_makmmacro): Added
	argument checking.

1998-12-15  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	Move the procedure slots of entities to invisible slots (so that
	we can have operator class objects which themselves are entities).
	* struct.h (scm_struct_i_proc, scm_struct_i_flags,
	SCM_STRUCTF_ENTITY): New constants.

	* struct.c (scm_make_struct): Allocate "invisible" room for
	procedures if SCM_STRUCTF_ENTITY is set in vtable.

	* gc.c (scm_gc_mark): Mark entity procedures.

        * struct.c, struct.h (scm_alloc_struct): Renamed from alloc_struct
        and made global.
        (scm_struct_init): Renamed from init_struct and made global.

	* objects.h (SCM_ENTITY, scm_entity): Removed.
	(SCM_ENTITY_PROC_0, SCM_ENTITY_PROC_0, SCM_ENTITY_PROC_0,
	SCM_ENTITY_PROC_3): Adjusted for new location of procedure slots.

Mon Dec 14 18:10:12 1998  Maciej Stachowiak  <mstachow@mit.edu>

	* snarf.h (SCM_SYNTAX): New macro to allow convenient declaration
 	of syntactic forms, similar to SCM_PROC.
	* debug.c, eval.c, evalext.c: use SCM_SYNTAX to declare all special
 	forms, and SCM_SYMBOL or SCM_GLOBAL_SYMBOL to delcare C variables
 	for the name symbols when needed.

1998-12-14  Jim Blandy  <jimb@zwingli.cygnus.com>

	* Makefile.in: Regenerated.

1998-12-05  Jim Blandy  <jimb@zwingli.cygnus.com>

	* pairs.h (SCM_NEWCELL): When DEBUG_FREELIST is defined, don't
	take the address of _into; it might be a register.  Just have
	scm_debug_newcell return the new cell.
	* gc.c (scm_debug_newcell): Just return the new cell, instead of
	taking the address of a place to store it as an argument.
	* gc.h (scm_debug_newcell): Change declaration.
	(Thanks to Greg Harvey.)

1998-12-08  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* ramap.c (array-copy-in-order!, array-map-in-order):  New names.
	Replaces old names serial-array-copy! and serial-array-map!.

	* evalext.c (map-in-order): New name.  Replaces serial-map.

1998-12-05  Jim Blandy  <jimb@totoro.red-bean.com>

	* smob.c (freeprint): New function.
	(freecell): Use it to print freed objects, for slightly easier
 	debugging.

1998-12-04  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* backtrace.c (display_frame): Made more robust.  Doesn't throw an
	error if no source properties can be found for a frame.  (Thanks
	to Christian Lynbech.)

1998-11-27  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* objects.h: Removed slots direct_supers and direct_slots from the
	definitions of the rudimentary classes described by objects.h.

	* objects.c, objects.h (scm_entity_p): New procedure.  Together
	with the predicates scm_procedure_p and scm_struct_p, this
	predicate makes it possible to differ between structs, entities
	and operators.

	* modules.c, modules.h (scm_resolve_module): New function.

1998-11-26  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* objects.h (SCM_METACLASS_STANDARD_LAYOUT,
	SCM_METACLASS_OPERATOR_LAYOUT): Removed slots `direct_supers' and
	`direct_slots'.

	* objects.c (scm_entity_p): New procedure.

	* procprop.c (scm_i_procedure_arity): Bugfix: Return correct value
	for asubrs, rpsubrs, lsubrs and lsubr_2s.

1998-11-26  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* procprop.h (scm_i_procedure_arity): Added declaration.

	* procprop.c (scm_i_procedure_arity): Made global; New code to
	handle operators and entities.
	(scm_procedure_property): No need to call scm_procedure_p since
	scm_i_procedure_arity now does all necessary type checking.
	Added #include "objects.h".

	* feature.c (scm_remove_hook_x): Bugfix: Changed reference to
	s_add_hook_x --> s_remove_hook_x.
	(scm_add_hook_x, scm_remove_hook_x): Hooks now takes arguments.
	Added #include "procprop.h"

	* feature.c, feature.h (scm_reset_hook_x): New procedure.
	(scm_make_hook): Optional argument defines number of arguments to
	the hook.
	(scm_make_named_hook): Take number of args as second arg.
	(scm_run_hook): Renamed from scm_run_hooks (old name defined in
	boot-9.scm for a while); First arg is the hook.  The rest are
	arguments passed on to the hook procedures.

1998-11-23  Jim Blandy  <jimb@zwingli.cygnus.com>

	* numbers.c (scm_logand, scm_logior, scm_logxor, scm_logtest,
 	scm_logbit_p): Do the computation in ulongs.  This is not as nice
 	as doing it in bignums, but at least it's good enough for
	manipulating flags in 32-bit words.  (Thanks to Jim Wilson.)
	
	* regex-posix.c (scm_regexp_exec): Reliably mark unmatched
	subexpressions.  (Thanks to Charbel Jacquin.)

1998-11-23  Mikael Djurfeldt  <mdj@kenneth>

	* feature.c, feature.h (scm_make_hook, scm_add_hook_x,
	scm_remove_hook_x, scm_run_hooks): Moved from ice-9/boot-9.scm.
	(scm_make_named_hook): New function.

	* feature.c: Added #include "eval.h".
	
	* modules.c (scm_make_module): Beautify the module.

1998-11-22  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* modules.c, modules.h: New files: C interface to modules.  (This
	is necessary in order to interface the object system to Guile
	properly.  The guts of these modules will be replaced by the new
	module system in the future.)
	
	* init.c: Added #include "modules.h"
	(scm_boot_guile_1): Call scm_init_modules.
	(invoke_main_func): Call scm_post_boot_init_modules.

	* Makefile.am: Added modules.c, modules.x, modules.h.

1998-11-22  Mikael Djurfeldt  <mdj@kenneth>

	* procs.c: #include "objects.h"
	(scm_procedure_p): Return #t also on structs which are operators.

	* objects.c (scm_init_objects): Renamed <standard-metaclass>,
	<operator-metaclass> and <entity-class> to <standard-class>,
	<operator-class> and <entity> in order to conform with STKlos
	naming conventions.

	* eval.c (SCM_CEVAL): Jump to badfun if trying to apply a struct
	which isn't an operator.
	(SCM_APPLY): Ditto, but jump to badproc.

1998-11-21  Mikael Djurfeldt  <mdj@kenneth>

	* eval.c (SCM_CEVAL): Allow structs implanted in code.
	Previously, structs implanted in code were interpreted as forms
	the operator of which was a gloc.  We solve this by checking for
	the zero in the emulated vcell in the struct vtable.  Since
	implanted structs always will look like forms with a gloc
	operator, execution will only be slowed down by maximally one
	extra test-and-branch per application.

	* evalext.c (scm_definedp): Removed check for isyms; Added a
	second optional argument: It is now possible to supply an
	evaluation environment in which to look for the symbol.

1998-11-19  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* readline.c (scm_init_readline): Set
	rl_basic_word_break_characters.  (Thanks to Ian Grant.)

	* coop.c (coop_condition_variable_wait): Removed
	(coop_condition_variable_wait_mutex): Folded logic of
	coop_mutex_unlock into coop_condition_variable_wait_mutex to
	prevent condvar signal lossage.  Previously, another thread could
	start to run after unlocking the mutex but before putting the
	current thread on the wait queue.  If that thread then would
	signal the first, the signal would be lost.  (Thanks to Christian
	Lynbech.)

1998-11-17  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* eval.c (SCM_CEVAL): Added missing case for cclo.  (Thanks to
	Christian Lynbech.)

1998-11-14  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* objects.c (scm_i_make_class_object): Renamed from
	make_class_object; exported; error checking moved to
	scm_make_class_object and scm_make_subclass_object.
	(scm_make_class_object, scm_make_subclass_object): Use
	scm_i_make_class_object.
	(scm_make_subclass_object): Let the subclass have same metaclass
	as the superclass.

1998-11-14  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* debug.c (scm_debug_options): Bugfix: Set the value of
	scm_stack_checking_enabled_p after setting debug options;
	#include "stackchk.h".  (Thanks to Richard Polton.)

1998-11-13  Radey Shouman  <rshouman@metro2000.com>

	* unif.c (scm_array_contents): removed unnecessary test for 0
	base.

1998-11-13  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* evalext.c, evalext.h (scm_m_sequence_to_list): Removed.
	Replaced by macro `collect' in boot-9.scm.

1998-11-10  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* eval.c (scm_copy_tree): Copy source properties if existent.

	* debug.c (scm_start_stack): Copy source when evaluating.  (If we
	don't, we may end up passing memoized source to a transformer.)

1998-11-10  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* stack.c (get_applybody): Help function which lookups the first
	body form of the apply closure.
	(read_frames): Prevent the source of the first form of the apply
	closure from being recorded.  This would only be confusing.

	* debug.h (SCM_SET_MACROEXP, SCM_CLEAR_MACROEXP, SCM_MACROEXPP):
	Replaces SCM_MACROFRAME, SCM_MACROFRAMEP.
	
	* eval.c (SCM_CEVAL): Use SCM_SET/CLEAR_MACROEXP.

	* stacks.c (read_frame): Bugfix: Removed lingering `else'
	statement.
	(read_frames): Use SCM_MACROEXPP.

1998-11-10  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* stacks.c (read_frames): Skip gsubr frames in backtraces.  (They
	don't contain interesting information since all arguments are
	present in the frame which applies the compiled closure anyway.);
	Skip the transformer application frames.

	* print.c (scm_iprin1): Print gsubrs as primitives.

1998-11-09  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* debug.h (SCM_MACROFRAME, SCM_MACROFRAMEP): New frame type.

	* eval.c (SCM_CEVAL): Mark macro frames at `handle_a_macro' so
	that we can identify these in a backtrace.  (This change doesn't
	introduce any significant speed penalty.)

	* eval.c: Added note about `serial-map' using scm_map.

	* read.c, read.h (scm_read_options, scm_read_opts): Removed
	readline options.  They should reside in their own options array.
	
	* readline.c, readline.h (scm_readline_options,
	scm_readline_opts): Moved readline options here.

1998-11-07  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* readline.c (scm_read_history, scm_write_history): Bugfix: Use
	SCM_ROCHARS instead of SCM_CHARS.

	* ports.c (scm_unread_string): Bugfixes: Check for SCM_STRINGP,
	not SCM_ROSTRINGP; use SCM_ROUCHARS instead of SCM_UCHARS.

1998-11-06  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* ports.h (SCM_CUC): #define as ~SCM_CRDY instead of 0x001fffffL.
	This is quite important since the latter clears the
	FPORT_READ_SAFE and FPORT_WRITE_SAFE flags causing flushes at
	every single character read...

1998-11-03  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* print.c (scm_iprin1): Removed suspect looking (and indeed
	malevolent) semicolon after test for user supplied closure print
	procedure.  (Thanks to Telford Tendys.)

	* list.c (scm_sloppy_memq): Removed sloppy_mem_check.
	(scm_memq, scm_memv, scm_member): Do argument checking *before*
	starting to search the list.  Removed call to sloppy_mem_check.

	* list.c, list.h (scm_delq1_x, scm_delv1_x, scm_delete1_x): New
	procedures: Same as scm_delq_x et al, but delete maximally one
	element.

	* options.c (scm_options, scm_init_options): GC-protect option
	values of type SCM.  (Thanks to Telford Tendys.)

1998-11-01  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* eval.c: Don't #define scm_lookupcar to scm_lookupcar1.  Instead
	make sure that there always is a "real" scm_lookupcar.

1998-11-01  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* read.c, read.h (scm_read_opts): New read options
	"history-length" and "history-file".
	(scm_read_options): Stifle history to history length.

	* readline.c (scm_read_history, scm_write_history): New procedures.

1998-10-31  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* eval.h (scm_macro_eval_x): Removed declaration.

	* eval.c (scm_s_expression, scm_s_test, scm_s_body,
	scm_s_bindings, scm_s_variable, scm_s_clauses, scm_s_formals):
	Renamed and made global.

	* eval.c, eval.h (SCM_EVALIM): Renamed from EVALIM.
	(SCM_XEVAL, SCM_XEVALCAR): Renamed from XEVAL, XEVALCAR.

	* evalext.c, evalext.h: New files.  Contain non-R5RS things
	having to do with evaluation.
	
	* evalext.c (scm_serial_map): New procedure: Version of `map'
	which guarantees that the procedure is applied to the lists in
	serial order.
	(scm_m_sequence_to_list): New syntax: Version of `begin' which
	returns a list of the results of the body forms instead of the
	result of the last body form.
	(scm_definedp, scm_m_undefine): Moved from eval.c

	* evalext.h (scm_m_sequence_to_list): Added declaration.

	* macros.c, macros.h: New files.
	(scm_procedure_to_syntax, scm_procedure_to_macro,
	scm_procedure_to_memoizing_macro, scm_macro_p, scm_macro_type,
	scm_macro_name, scm_macro_transformer): Moved from eval.c
	(scm_make_synt): Moved from eval.c

	* procs.c, procs.h (scm_procedure_documentation): Moved from eval.c.

	* init.c (scm_boot_guile_1): Added calls to scm_init_macros and
	scm_init_evalext.

	* Makefile.am: Added evalext.c, evalext.h, macros.c, macros.h.

	* debug.c, print.c: Added #include "macros.h".

1998-10-29  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	This change adds the ability to use `unread-char' multiple times
	without interspersed reads and adds the new procedure
	`unread-string'.  The change is optimized for the common case of
	unreading a single character.  This is also the reason behind the
	choice to store characters in the port itself: in most cases no
	extra malloc is required.

	The amount of code in some macros in ports.h has been increased to
	the extent that they would fit better as C functions, but, since
	this code belongs to the port representation, such functions
	should be placed in ports.c which would cause calls back and forth
	between ports.c and genio.c.  That is not good for performance.
	Also, keeping them as macros allows the compiler to do some
	optimizations which are needed to make the current interface
	(SCM_CRDYP, SCM_CGETUN, SCM_TRY_CLRDY) efficient.

	One benchmark (Guile startup time) indicates an increase of
	loading speed of 1%.  Another (reading (using `read') boot-9.scm
	10 times) shows no change in performance.

	(Caveat: Since Gary is redesigning I/O anyway, no big efforts were
	made to find a beautiful solution.)
	
	* ports.h (SCM_CLRDY, SCM_CUNGET, SCM_CGETUN): Rewritten.
	(SCM_TRY_CLRDY): New macro: Only clear the first unread
	character. (SCM_CLRDY clears all.)
	(SCM_N_READY_CHARS): New macro: Returns number of unread
	characters in a port.  Returns wrong answer if SCM_CRDYP is false.
	(struct scm_port_table): New fields: `entry' contains port table
	index, `cp' points to last unread char, `cbuf' is the buffer for
	unread chars, `cbufend' points after end of the character buffer.

	* ports.h, ports.c (scm_unread_string): New procedure.
	(scm_grow_port_cbuf): New function.

	* ports.c (scm_add_to_port_table, scm_remove_from_port_table):
	Handle new fields.
	(scm_generic_fgets), fports.c (local_fgets): Use a loop
	to read unread characters.  Use SCM_TRY_CLRDY instead of
	SCM_CLRDY.
	
	* ioext.c (scm_ftell): Use SCM_N_READY_CHARS to correct position.

	* genio.c (scm_getc): Use SCM_TRY_CLRDY instead of SCM_CLRDY.

	* genio.c, genio.h (scm_ungets): New function.

	* genio.c (scm_puts): Removed mysterious TRANSCRIPT_SUPPORT code
	sections.

1998-10-28  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* threads.h (scm_thread_sleep, scm_thread_usleep): Fixed
	declarations. (Thanks to Russ McManus.)

1998-10-26  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* numbers.c (num2long): As a software archeologist, I'm proud of
	this finding! :) Preliminary dating suggests an almost 4 year old
	remnant from the SCM ancestor.  The sample has been removed from
	the finding site and is now safely stored in the repository.

	* numbers.h: Removed prototype for num2long.

	* unif.c (scm_array_set_x): Use scm_num2long instead of num2long.

	* gh_data.c (gh_scm2doubles): Make it possible to pass result
	array as second arg.
	(gh_chars2byvect, gh_shorts2svect, gh_floats2fvect, gh_scm2chars,
	gh_scm2shorts, gh_scm2longs, gh_scm2floats): New functions.
	* gh.h: Updated and added prototypes.

	* gh_data.c (gh_ints2scm): Handle integers outside INUM limits.

1998-10-24  Jim Blandy  <jimb@zwingli.cygnus.com>

	* gc.h: Remove all uses of SCM_P.  (Thanks to Richard Polton.)

	* guile-snarf.in: Never generate an empty file.  (Thanks to
	Richard Polton.)

	* gh.h (gh_enter, gh_new_procedure0_0, gh_new_procedure0_1,
 	gh_new_procedure0_2, gh_new_procedure1_0, gh_new_procedure1_1,
 	gh_new_procedure1_2, gh_new_procedure2_0, gh_new_procedure2_1,
 	gh_new_procedure2_2, gh_new_procedure3_0, gh_new_procedure4_0,
 	gh_new_procedure5_0): Specify argument types, to appease C++
	compilers.  (Thanks to Brad Bowman.)

	Bug reports from Russ McManus:
	* guile-snarf.in: If the CPP environment variable is set, use that
	as the C preprocessor, instead of the preprocessor autoconf
	found.
	* snarf.h (SCM_PROC): Cast the function pointer passed to
	scm_make_gsubr, to satisfy C++.

	* gh_eval.c (gh_eval_str_with_catch, gh_eval_file_with_catch):
	Use the handler passed, instead of ignoring it and using
	gh_standard_handler.  (Thanks to Etienne Bernard.)
	
1998-10-20  Jim Blandy  <jimb@zwingli.cygnus.com>

	* ports.h: Put text after #endif in comment.  (Thanks to Nicolas
	Neuss.)

1998-10-19  Jim Blandy  <jimb@zwingli.cygnus.com>

	* __scm.h, alist.c, async.c, async.h, backtrace.h, chars.c,
 	continuations.c, debug.c, debug.h, dynl-dl.c, dynl.c, dynl.h,
 	dynwind.c, dynwind.h, eq.c, error.c, error.h, eval.c, eval.h,
 	feature.c, filesys.c, filesys.h, fports.c, fports.h, gc.c, gc.h,
 	genio.c, genio.h, gh.h, gh_data.c, gsubr.c, gsubr.h, hash.c,
 	hashtab.c, init.c, init.h, ioext.c, ioext.h, kw.c, libguile.h,
 	list.c, list.h, load.c, load.h, mallocs.c, markers.c,
 	mit-pthreads.c, net_db.c, numbers.c, numbers.h, options.c,
 	ports.c, ports.h, posix.c, posix.h, print.c, print.h, procprop.c,
 	procprop.h, procs.c, procs.h, ramap.c, ramap.h, regex-posix.c,
 	regex-posix.h, root.c, root.h, scmsigs.c, scmsigs.h, script.c,
 	script.h, simpos.c, simpos.h, smob.c, smob.h, snarf.h, socket.c,
 	srcprop.c, stackchk.c, stackchk.h, stacks.c, stime.c, stime.h,
 	strings.c, strings.h, strports.c, struct.c, struct.h, symbols.c,
 	symbols.h, tags.h, threads.c, throw.h, unif.c, variable.c,
 	vectors.c, vectors.h, version.h, vports.c, weaks.c: Update
	copyright years.

	* script.c (scm_compile_shell_switches): Define
	use-emacs-interface in the root module, so the repl code can see
	it.  See today's change to top-repl in ice-9/boot-9.scm.

	* filesys.c (set_element, get_element): Make sure that `element'
	is a cell before applying SCM_FPORTP to it.  (Thanks to Jost
	Boekemeier and Jorgen "forcer" Schaefer.)

1998-10-18  Jim Blandy  <jimb@zwingli.cygnus.com>

	* chars.c (scm_charnames): In ASCII character name table, make
	newlines print as #\newline by default, not #\nl.

	* Makefile.am (libguile_la_SOURCES, BUILT_SOURCES): Put these in
	alphabetical order.  Oh thrills.  But it helps me know how far
	along in the compilation I am. 
	* Makefile.in: Regenerated.

1998-10-18  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* unif.c (scm_raprin1): Changed print syntax for byte vectors from
	#bytes(...) to #y(...), and syntax for short vectors from
	#short(...) to #h(...).  This may seem nutty, but, like the other
	uniform vectors, byte vectors and short vectors want to have the
	same print and read syntax (and, more basic, want to have read
	syntax!).  Changing the read syntax to use multiple characters
	after the hash sign breaks with the conventions used in R5RS and
	the conventions used for the other uniform vectors.  It also
	introduces complexity in the current reader, both on the C and
	Scheme levels.  (The Right solution is probably to change the
	syntax and prototypes for uniform vectors entirely.)

1998-10-17  Jim Blandy  <jimb@savonarola.red-bean.com>

	Don't use local_fgets on sockets; ftell doesn't work on sockets.
	(Thanks to Jorgen "forcer" Schaefer.)
	* ports.h (SCM_NOFTELL): New flag.
	* fports.c (local_fgets): If it's set, use the generic fgets.
	* socket.c (scm_socket): Set SCM_NOFTELL on the ports we produce.

1998-10-17  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* script.c (scm_compile_shell_switches): Add handling of -q switch
	(inhibit loading of user init file).
	(scm_shell_usage): Add usage text for -q switch.
	(scm_compile_shell_switches): Always load user init file first if
	it is loaded at all.

1998-10-16  Jim Blandy  <jimb@zwingli.cygnus.com>

	* stime.c: The CPP hair to determine a value for CLKTCK is weird,
	and is now broken under NetBSD.  I can't fathom what it's trying
	to do, so I've replaced it with something that I do understand,
	which seems to work, and which isn't broken on NetBSD.  "Progress?
	You Decide."  (Thanks to Perry Metzger.)

	* regex-posix.c (scm_regexp_exec): Add a cast to remove a
	signed/unsigned comparison.

1998-10-15  Jim Blandy  <jimb@zwingli.cygnus.com>

	Warning fixes from Greg Harvey:
	* unif.c (scm_array_set_x): initializer for pos
	* throw.c (scm_ithrow): added initializer for jmpbuf (SCM_UNDEFINED)
	* struct.c (scm_struct_ref, scm_struct_set_x): Added
	initializers for field_type, since EGCS so desparately wants to
	play dumb
	* debug.h (scm_make_gloc, scm_gloc_p, scm_make_iloc, scm_memcons,
	scm_mem_to_proc, scm_proc_to_mem, scm_debug_hang): Added prototypes
	when GUILE_DEBUG is defined.
	* dynwind.h (scm_wind_chain): Same.
	* ports.h (scm_pt_size, scm_pt_member): Same.
	* print.h (scm_current_pstate): Same.
	* procs.h (scm_make_cclo): Same.

1998-10-14  Jim Blandy  <jimb@zwingli.cygnus.com>

	Handle short and long long uniform arrays properly.  (Thanks to
	Clark McGrew.)
	* ramap.c (scm_ra_matchp, scm_array_fill_int, scm_array_index_map_x,
 	raeql_1): Add cases for scm_tc7_svect (short vectors) and
	scm_tc7_llvect (long long vectors).

	Change the way libguile and boot-9.scm are timestamped, to try to
	get rid of these spurious mismatch warnings.  Now both
	libguile/versiondat.h and ice-9/version.scm are generated directly
	by the configuration process, rather than having version.scm
	generated directly, and libguile/versiondat.h generated by the
	Makefile, which is generated by configure.  It seems that
	sometimes the Makefile would change, but versiondat.h depends on
	Makefile.in, not Makefile, so it wouldn't get rebuilt.
	* Makefile.am (versiondat.h): Target removed; this is generated
	directly by the configure script now.
	(BUILT_SOURCES): Remove versiondat.h.
	* versiondat.h.in: New file, transformed by the configure script
	into versiondat.h.
	* Makefile.in: Regenerated.

	* (__scm.h, alist.c, eval.c, feature.c, gsubr.c, numbers.c,
 	options.c): Rename RECKLESS -> SCM_RECKLESS, CAUTIOUS ->
	SCM_CAUTIOUS; this way, 1) there's only one version of each flag
	to define (we used to have both RECKLESS and SCM_RECKLESS), and 2)
	if we want to use them in a header file some day, we can.  (Thanks
	to Michael Livshin.)

	* stime.c (scm_get_internal_real_time): Do the arithmetic using
 	SCM numbers, so we won't have rollover problems; the range of a
 	signed long in milliseconds is about 25 days.  (Thanks to Karl
 	Hegbloom.)

	Don't redefine sleep and usleep; fix this problem now.
	* coop.c (sleep, usleep): Remove declarations; we don't use or
	redefine these any more.
	(scm_thread_usleep, scm_thread_sleep): New functions which do the
	job of usleep and sleep in a thread-friendly way.  We can use
	these in the rest of Guile.  Define versions for systems both with
	and without iselect.
	* threads.h (scm_thread_usleep, scm_thread_sleep): New declarations.
	* scmsigs.c (usleep): Clean up oddities declaring usleep; since
 	we're just using it, not redefining it, we can use a K&R style
 	declaration here.
	(sleep): Declare this, too, if the system hasn't.
	(scm_sleep, scm_usleep): Use scm_thread_sleep and
	scm_uthread_sleep if they're available; otherwise, just call the
	system functions.
	* scmconfig.h.in: Regenerated.

	* coop.c (scm_thread_sleep): Make `slept' an unsigned long.

	* coop.c (coop_sleephelp): Remove ANSI #ifdef hair.

1998-10-12  Jim Blandy  <jimb@zwingli.cygnus.com>

	* threads.c: Doc fix.

	The argument type of usleep varies from system to system,
	as does the return type.  We really shouldn't be redefining usleep
	at all, but I don't have time to clean that up before the 1.3.
	release.  It's on the schedule for afterwards.  (Thanks to Julian
	Satchell.)
	* coop.c (usleep): Use USLEEP_ARG_TYPE in prototype and
	definition.
	* scmsigs.c (usleep): Use USLEEP_ARG_TYPE in prototype.
	* scmconfig.h: Regenerated.

	* simpos.c (scm_software_type): Procedure deleted.  This isn't the
	right way to handle system variation.  Autoconf's approach is the
	way and the light.
	* simpos.h (scm_software_type): Declaration deleted.

	* script.c (scm_find_executable): Don't test if unix is #defined
	here; first, NetBSD doesn't #define it, and second, it's the wrong
	way to go about these things.  (Thanks to Perry Metzger.)
	(dld_find_executable): Delete this MSDOS support code.  This isn't
	the way we want to support this; it needs to be rethunk at a
	larger scale.

	* genio.c (scm_do_read_line): Don't just politely check that the
	line was either non-empty or EOF.  Abort if it's empty and not
	EOF.

1998-10-11  Jim Blandy  <jimb@zwingli.cygnus.com>

	* scmconfig.h.in: Regenerated.

	* libguile.h: Don't omit the dynamic linking functions.  (Thanks
	to Greg Badros.)

	* genio.c (scm_do_read_line): Count lines correctly when the file
	doesn't end in a newline.

1998-10-10  Jim Blandy  <jimb@zwingli.cygnus.com>

	* genio.c (scm_do_read_line): Maintain the line count correctly.
	(Thanks to Harvey J. Stein and Greg Harvey.)

	* gc.c (scm_return_first): Remove #ifdef __STDC__ garbage; Guile
	requires ANSI now.

	* numbers.c (big2str): Protect t from garbage collection until
	we're done.  (Thanks to Gary Houston.)

1998-10-09  Jim Blandy  <jimb@zwingli.cygnus.com>

	* Makefile.am (libguile_la_LDFLAGS): Increment shared library
	version number.
	* Makefile.in: Regenerated.

	* fports.h (scm_setbuf0, scm_setvbuf, scm_setfileno,
	scm_evict_ports, scm_open_file, scm_stdio_to_port): Get rid of
	SCM_P macro.

	Do magic to mix reads and writes on stdio FILE-based ports.
	(Thanks to Christian Lynbech.)
	* fports.c (FPORT_READ_SAFE, FPORT_WRITE_SAFE, FPORT_ALL_OKAY):
	New macros.
	(pre_read, pre_write): New functions.
	(local_fgetc, local_fgets, local_ffwrite, local_fputc,
	local_fputs): Call them.
	(local_fflush): Mark the port as ready for reading and writing.
	(scm_stdio_to_port): Set the FPORT_READ_SAFE, FPORT_WRITE_SAFE
	flags on new port objects.  This might not be accurate --- who
	knows what state the FILE * is in when we get it --- but it won't
	do extraneous calls to fflush or fseek, so it's no worse than the
	behavior before this change.
	* ports.h: Add comment.

	Centralize the creation of port objects based on stdio FILE * in
	fports.c; don't just throw them together anywhere.
	* fports.c (scm_stdio_to_port): Make NAME a SCM value, which is
	what the rest of Guile wants.  Don't set the revealed count;
	that's only appropriate for stdin, stdout, stderr.
	(scm_standard_stream_to_port): This function does set the revealed
	count.
	* init.c (scm_init_standard_ports): Use	scm_standard_stream_to_port,
	not scm_stdio_to_port.
	* filesys.c (scm_open): Call scm_stdio_to_port; don't write it out.
	* fports.c (scm_open_file): Same.
	* posix.c (scm_pipe): Same.
	* socket.c (scm_sock_fd_to_port): Same.
	* ioext.c (scm_fdopen): Same.
	(scm_freopen): Moved from here to...
	* fports.c (scm_freopen): ... here.  This is really something that
	munges the internals of an fport, so it should go here.
	* fports.h (scm_stdio_to_port): Adjust prototype.
	(scm_standard_stream_to_port, scm_freopen): New protoypes.
	* ioext.h (scm_freopen): Prototype removed.

	* filesys.c (set_element, get_element): This can work on both pipe
	and file ports, so use SCM_FPORTP to typecheck, instead of testing
	for scm_tc16_fport.

	* scmconfig.h.in: Regenerated.

	Change the definition of the functions in scm_ptobfuns so that
	they get passed the port object, not the port's stream.
	* ports.h (scm_ptobfuns): Rename all `stream' arguments to `port'.
	* gc.c (scm_gc_sweep): Pass the port itself to the free function.
	* genio.c (scm_putc, scm_puts, scm_lfwrite, scm_fflush, scm_getc):
	Pass the port itself to the scm_ptobs function.
	* ports.c (scm_close_port, scm_force_output, scm_flush_all_ports,
	scm_generic_fgets): Same.
	(putc_void_port, puts_void_port, write_void_port, flush_void_port,
	getc_void_port, fgets_void_port, close_void_port): Just change the
	argument names; these functions don't really do anything.
	* fports.c (local_fgetc, local_fgets, local_fclose, local_fflush,
	local_fputc, local_fputs, local_ffwrite, local_pclose): Take the
	port as an argument, and use SCM_STREAM to get the stdio FILE *.
	Also, use prototyped definitions, and get rid of the extra
	declarations.
	(scm_fptob, scm_pipob): We don't need casts here any more.
	* strports.c (prinstpt): Use prototype declarations.
	(stputc, stwrite, stputs, stgetc): Take the port as an argument,
	and use SCM_STREAM to get the string info.  Also, use prototyped
	definitions, and get rid of the extra declarations.
	* vports.c (sfputc, sfwrite, sfputs, sfflush, sfgetc, sfclose,
	noop0): Same.

	* ports.h (scm_ptobfuns): Replace uses of SCM_P with a straight
	prototype; it's okay (preferred, even!) to use ANSI C in Guile.
	
	* fports.c (local_fgetc, local_fgets): Renamed from scm_fgetc and
	scm_fgets, for consistency.
	(scm_fptop, scm_pipob): References updated.

1998-10-08  Jim Blandy  <jimb@zwingli.cygnus.com>

	Include the source location in error messages for scripts.
	* init.c (scm_boot_guile_1): Use scm_internal_lazy_catch, so the
	stack is still there when we catch the error.
	* throw.c (handler_message): If we are handling an error with a
	message, then put together the proper arguments and call
	scm_display_error, instead of scm_display_error_message.  That
	displays source location, if it can find it.

1998-10-07  Jim Blandy  <jimb@zwingli.cygnus.com>

	* gc.c (scm_unprotect_object): Change this so that calls to
	scm_protect_object and scm_unprotect_object nest properly.
	(scm_protect_object): Doc fixes.

	* strings.c (scm_string_set_x): Require the argument to be a
	writable string, not a substring or a symbol.
	* strings.h (SCM_RWSTRINGP, SCM_NRWSTRINGP): New predicates.
	(Thanks to John Redford and Charbel Jacquin.)
	
	* scmconfig.h.in: Regenerated; ../acconfig.h has changed.

1998-10-07  Jim Blandy  <jimb@totoro.red-bean.com>

	* eval.c (safe_setjmp): Remove this misunderstanding.
	(ENTER_APPLY, SCM_CEVAL, SCM_APPLY): Replace with references to
	ordinary setjmp.

1998-10-06  Jim Blandy  <jimb@zwingli.cygnus.com>

	* libguile.h: Mark these as C declarations, for compilation by C++
	compilers.

	* snarf.h (SCM_PROC, SCM_PROC1): Remove very odd code in #ifdef
 	__cplusplus clause.  I seriously doubt this ever worked the way
 	the author seems to have intended.

1998-10-05  Jim Blandy  <jimb@zwingli.cygnus.com>

	Utterly needless cleanups to hopelessly messy code.
	* ports.c: Doc fixes.
	(scm_fflush): Moved to ...
	* genio.c (scm_fflush): ... here, amongst all the other port
	method invocation functions.
	* genio.h, ports.h: The prototype moves too.

1998-10-04  Jim Blandy  <jimb@zwingli.cygnus.com>

	* backtrace.c (display_error_body): The current frame does not
 	always have a parent frame; consider a function called directly
 	from the MAIN_FUNC passed to scm_boot_guile.  (Thanks to Maciej
 	Stachowiak.)

	* alloca.c (alloca): Undo yesterday's changes, and simply call
 	malloc directly for storage, and abort if we don't get what we
 	want.  The situation is much simpler --- just call malloc.  Emacs
 	has bizarre/evil requirements (signal handlers might malloc unless
 	you set this global flag, so you have to set the flag around all
 	calls to malloc) which we are certainly not going to conform to,
 	so we can do the simple and obvious thing.

	* coop.c (coop_condition_variable_wait): Make this function
	static.  It's only useful internally --- you should never just
	wait on a condition variable.
	* coop-defs.h (coop_condition_variable_wait): Delete prototype.

1998-10-03  Jim Blandy  <jimb@zwingli.cygnus.com>

	* unif.c (scm_array_set_x): Accept any kind of number as an
	element for a uniform vector of doubles.  This is more consistent
	with Scheme's view of numbers.  (Thanks to Miroslav Silovic.)

	* alloca.c: Use scm_must_malloc to obtain storage.  Hopefully this
	works; I can't conveniently test it myself.  (Thanks to Dvid
	Tillman for the bug report.)

	* init.c: Doc fixes.

	* init.c (invoke_main_func): Load the startup files (boot-9.scm)
 	from here, not from scm_compile_shell_switches (which is a pretty
 	dumb place to do it).
	(scm_load_startup_files): New function.
	(scm_ice_9_already_loaded): Variable moved to here from script.c.
	* script.c (scm_compile_shell_switches): Don't load the startup
	files here.
	(scm_ice_9_already_loaded): Variable moved.
	* init.c (scm_load_startup_files): Prototype for new function.
	* gh_init.c (gh_enter): Doc fix.

1998-10-03  Jim Blandy  <jimb@savonarola.red-bean.com>

	Some anti-warning changes from Greg Harvey.
	* gh_data.c (gh_scm2doubles): Initialize m, to avoid compiler
	warnings when it doesn't understand our NORETURN declarations in
	error.h.
	* posix.c (scm_mknod): Similar.

1998-10-03  Jim Blandy  <jimb@zwingli.cygnus.com>

	* posix.c (scm_getpwuid): If we can't find an entry, return our
 	own message, instead of using scm_syserror --- the getpwMUMBLE
 	functions don't set `errno' to anything interesting.

1998-10-03  Jim Blandy  <jimb@savonarola.red-bean.com>

	Get rid of warnings from the cooperative threading system.
	* threads.h (scm_single_thread_p, scm_yield,
	scm_call_with_new_thread, scm_join_thread, scm_make_mutex,
	scm_lock_mutex, scm_unlock_mutex, scm_make_condition_variable,
	scm_wait_condition_variable, scm_signal_condition_variable): Add
	prototypes for these Scheme-visible functions.
	* coop-defs.h (coop_next_runnable_thread,
	coop_wait_for_runnable_thread_now, coop_wait_for_runnable_thread):
	Prototypes for these here, even though they're from iselect.c.
	(coop_condition_variable_wait, coop_join): Add prototypes.
	* coop-threads.c (scm_threads_free_thread, scm_threads_free_mutex,
	scm_threads_free_condvar): Make these smob functions static.
	* coop-threads.h (coop_init): Give this a real prototype.
	* coop.c: #include <unistd.h>, if we have it, for `usleep' and `sleep'.
	(coop_next_runnable_thread): No need to provide prototype; it's in 
	coop-defs.h.

	* scmconfig.h.in: .detarenegeR

	* iselect.c, threads.c: Doc fixes.

1998-10-03  Jim Blandy  <jimb@zwingli.cygnus.com>

	* Makefile.in: Regenerated with a patched automake, to get
	dependency generation right when using EGCS.

	* inet_aton.c (inet_aton): Add prototype, to remove compiler
 	warning. (Thanks to Robert Pluim.)

	* inet_aton.c (inet_aton): Reassure the compiler that the
 	arguments to the <ctype.h> macros are all unsigned characters, not
 	signed characters.

1998-10-03  Jim Blandy  <jimb@savonarola.red-bean.com>

	Getting rid of more warnings...
	* iselect.c: Test for MISSING_BZERO_DECL, not DECLARE_BZERO; see
	today's change to ../configure.in.
	* scmsigs.c: Test for MISSING_USLEEP_DECL, not DECLARE_USLEEP.
	* scmconfig.h.in: Regenertaded.de.,.__
	* stime.c (strptime): Declare this, #ifdef MISSING_STRPTIME_DECL.
	(scm_localtime, scm_mktime): Use a const char * to manipulate the
	time zone name.

	* readline.c: Doc fix.
	(rl_cleanup_after_signal, rl_free_line_state): Make these static.
	* readline.h (scm_filename_completion_function): Add prototype.
	(scm_init_readline): Make this into a prototype.

	* readline.c (scm_filename_completion_function): Use SCM_PROC to
	declare this, instead of calling scm_make_subr manually.

1998-10-02  Jim Blandy  <jimb@savonarola.red-bean.com>

	* readline.h (scm_init_readline): Add prototype for this.
	(scm_init_readline): Make this a real prototype.

1998-09-30  Jim Blandy  <jimb@zwingli.cygnus.com>
	
	Warning fixes from Maciej Stachowiak:
	* backtrace.h (scm_display_application, scm_backtrace): Add
	prototypes.
	* debug.c (scm_m_start_stack): Make this function static.
	* fluids.h (scm_fluid_p): Add prototype.
	* procprop.c (scm_i_procedure_arity): Make this function static.
	* regex-posix.c (scm_regexp_error_msg): Make this function static.
	* regex-posix.h (scm_init_regex_posix): Use prototype, not K&R decl.
	* root.h (scm_dynamic_root): Add external prototype.
	* scmsigs.h (scm_usleep): Add external prototype.
	* script.h (scm_init_script): Use prototype, not K&R decl.
	* stacks.h (scm_stack_id): Add external prototype.
	* symbols.h (scm_sysintern0_no_module_lookup): Add external prototype.

1998-09-30  Mark Galassi  <rosalia@cygnus.com>

	* gh.h: took out the definitions of vset and vref, since they are
	replaced by the proper vector routines that correspond to the R4RS
	procedures.

1998-09-29  Jim Blandy  <jimb@totoro.red-bean.com>

	* snarf.h (SCM_VCELL_INIT, SCM_GLOBAL_VCELL_INIT): New macros;
 	these are analogous to SCM_VCELL and SCM_GLOBAL_VCELL but take a
 	third argument, a C expression that should result in a SCM value,
 	which is used to initialize the variable. Reimplemented
 	SCM_CONST_LONG in terms of SCM_VCELL_INIT.  (Thanks to Maciej
	Stachowiak.)

	* version.h (scm_libguile_config_stamp): Add prototype.
	(From Maciej Stachowiak.)

1998-09-26  Jim Blandy  <jimb@zwingli.cygnus.com>

	* eval.c (scm_force): Assert that x is SCM_NIMP to fix segv when
	(force 9) is tried.  (Thanks to Karl M. Hegbloom.) 

1998-09-06  Jim Blandy  <jimb@zwingli.cygnus.com>

	* print.c (scm_iprin1): Rather than having one i, and using it in
	several places, declare a fresh i local to each block where it is
	used, and give it a signedness appropriate to its use in each case.
	(scm_iprlist): Same.

	* print.c (scm_iprin1): Add cast to avoid unsigned/signed
	comparison warnings.

	* print.c (ENTER_NESTED_DATA): Make i an unsigned long, to avoid
	signed/unsigned clashes.

	* posix.h (scm_tmpnam): Added prototype.

	* objects.h (scm_set_object_procedure_x, scm_make_class_object,
 	scm_make_subclass_object): Add external prototypes.

	* numbers.c (scm_mkbig): Add cast, and note that signed/unsigned
 	comparison is okay here.

	* numbers.c (scm_istr2int): Add cast; len is known to be positive.

	* numbers.c (scm_bigcomp): Clarify logic, and avoid relying on the
	(true, but confusing) fact that -1 == ((unsigned) 0 - 1).

	* numbers.c (scm_adjbig): Make nsiz an scm_sizet, to avoid mixing
 	signed/unsigned.

	* load.c (swap_port): Make this function static.

	* load.c (scm_search_path): Make max_path_len and max_ext_len
 	unsigned, since they're compared against string sizes.

	* load.c (init_build_info): Make i unsigned.

	* ioext.h (scm_read_line): Add prototype.
	
	* hashtab.c (scm_hash_fn_get_handle, scm_hash_fn_create_handle_x,
 	scm_hash_fn_remove_x): Make hash bucket index local variable k
 	unsigned.  Use scm_ulong2num to pass it to SCM_ASSERT as
 	accurately as possible.

	* gh_data.c (gh_set_substr): Add casts to avoid signed/unsigned
	comparisons, and range checking to make sure those casts are
	harmless.

	* stackchk.h (SCM_STACK_OVERFLOW_P): Change definition to avoid
	signed/unsigned comparisons.

	* smob.c (scm_numsmob): Make this an int, not an scm_sizet, to
	avoid signed/unsigned comparisons.
	* smob.h (scm_numsmob): Change extern declaration to match.

	* ports.c (scm_numptob): Make this an int, not an scm_sizet, to
	avoid signed/unsigned comparisons.
	* ports.h (scm_numptob): Change extern declaration to match.
	(scm_current_load_port, scm_set_port_line_x,
	scm_set_port_column_x): New prototypes.

	* gsubr.c (GSUBR_TEST): Don't #define this.  Nobody's using the
 	test code, and it causes warnings.

	* gh.h (gh_int2scmb, gh_uniform_vector_length,
 	gh_uniform_vector_ref): Added prototypes.

	* Makefile.am (libguile_la_SOURCES): Include Scheme-level
	debugging support unconditionally.  That's backtrace.c, stack.c,
	debug.c, and srcprop.c.
	(EXTRA_libguile_la_SOURCES): Omit those from here.
	* Makefile.in: Regenerated.

1998-08-21  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* options.c (scm_options): Bugfix: Allow empty list of options!

	* debug.c, debug.h (scm_single_step): Removed.
	(scm_with_traps): New procedure.  This procedure could easily be
	written in Scheme but needs to be highly optimized.

	* eval.h, eval.c: New evaluator trap flag: SCM_TRAPS_P.

	* eval.c (SCM_CEVAL, SCM_APPLY): Removed resetting of trap flags.
	Check SCM_TRAPS_P before trapping.

1998-07-30  Jim Blandy  <jimb@zwingli.cygnus.com>

	Changes to avoid signed/unsigned comparison warnings.
	* gc.c (scm_mtrigger, scm_heap_size): Make these unsigned longs.
	(scm_gc_sweep): Make n and j local to the blocks they're used in,
	so they can have appropriate types for each application.  Make i
	signed.  Use initializers in some spots.  I'll probably pay for
	all this tweaking.
	(scm_must_malloc, scm_must_realloc): Use scm_sizet for size args.
	(scm_must_realloc): Make nm unsigned.
	(init_heap_seg): Make new_seg_index and n_new_objects signed.
	(scm_init_storage): Use prototype-style definition, and make the
	argument unsigned.
	* gc.h (scm_heap_size, scm_mtrigger, scm_must_malloc,
	scm_must_realloc, scm_init_storage): Adjust prototype accordingly.
	
	* filesys.c (scm_readlink): Make local vars rv and size ints, to
	avoid signed/unsigned comparison warnings, and because the return
	value of readlink may be -1.  Don't bother casting the third
	argument to readlink.

	* filesys.c (scm_dirname, scm_basename): Move these to their own
	page, at the end of the file.
	* filesys.h (scm_dirname, scm_basename): Add prototypes for these.

	* eval.h (scm_eval_options_interface): Add external prototype for this.
	* eval.c (scm_eval_options_interface): Use prototype-style def'n.
	
	* eval.c (scm_lookupcar1): Make this static.

	* dynl.h (scm_registered_modules, scm_clear_registered_modules):
	Make these prototype declarations, not K&R-style.

	* chars.c (scm_tables_prehistory): Add cast, to remove signed/
 	unsigned comparison warning.

	* appinit.c: File removed.  It had a single function in it, empty,
 	whose reason for existence is explained in no documentation or
 	comment.  I think it's there as a default for some Tcl-style
	initialization, but Tcl abandoned that approach a while ago.
	* Makefile.am (libguile_la_SOURCES): Remove appinit.c.
	(BUILT_SOURCES): Remove appinit.x.
	* Makefile.in: Regenerated.

1998-07-29  Jim Blandy  <jimb@zwingli.cygnus.com>

	* Makefile.in: Regenerated using the last public version of
	automake, not the hacked Cygnus version.

1998-07-28  Jim Blandy  <jimb@zwingli.cygnus.com>

	Remove Totoro kludge.
	* Makefile.in, scmconfig.h.in: Regenerated.
	* init.c, readline.c: Don't check if TOTORO is #defined.

1998-07-26  Jim Blandy  <jimb@zwingli.cygnus.com>

	* Makefile.am: Adjust for new thread configuration system.
	(INCLUDES): Include the value of THREAD_CPPFLAGS.
	(guile_LDADD, check_ldadd): THREAD_LIBS_LOCAL has been renamed from
	THREAD_LIBS.
	(THREAD_LIBS): Definition deleted; automake will generate such
	things automatically.
	* Makefile.in: Regenerated.

1998-07-23  Jim Blandy  <jimb@zwingli.cygnus.com>

	Simplify smob and port marking; set the mark bit in the generic
	marking code, and make marker routines only responsible for
	turning up outgoing pointers.
	* gc.c (scm_gc_mark): Set the mark bit on ports and smobs here,
	before calling the marking function.  Don't call the marking
	function if it's zero.
	* markers.c (scm_mark0): Just return #f.  This function isn't
	necessary at all now, but it's harmless to call it.  We'll leave
	it in so other folks' code doesn't croak at link time.
	(scm_markcdr): Don't call SCM_SETGC8MARK.
	* async.c (mark_async): Don't call SCM_SETGC8MARK.
	* dynl.c (mark_dynl_obj): Same.
	* root.c (mark_root): Same.
	* srcprop.c (marksrcprops): Same.
	* unif.c (markra): Same.
	* variable.c (scm_markvar): Same.
	* ports.c (scm_markstream): Same.
	(void_port_ptob): Specify zero for our marking function.
	* debug.c (debugobjsmob): Same.
	* dynwind.c (guardsmob): Same.
	* filesys.c (dir_smob): Same.
	* fluids.c (fluid_smob): Same.
	* fports.c (scm_fptob, scm_pipob): Same.
	* mallocs.c (mallocsmob): Same.
	* regex-posix.c (regex_t_smob): Same.
	* smob.c (freecell, flob, bigob): Same.
	* threads.c (thread_smob, mutex_smob, condvar_smob): Same.
	* throw.c (jbsmob, lazy_catch_funs): Same.

1998-07-17  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* eval.c (scm_copy_tree): Reverted last change: `eval' uses
	scm_copy_tree on code in order not to let memoized code to leak
	out.  Thus, scm_copy_tree needs to copy vectors as well since
	quasiquote can introduce evaluated code also inside vector
	constants.

1998-07-17  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* eval.c (scm_copy_tree): Removed ability to copy vectors.

	* gh_data.c, gh.h (gh_ints2scm, gh_longs2ivect,
	gh_ulongs2uvect): New procedures.  (Complements gh_doubles2scm and
	gh_doubles2dvect.)

	* unif.c: Say that ivect and uvect are of type signed and unsigned
	long instead of int in commentary so that it correctly describes
	the implementation.

1998-07-12  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* stime.c: Removed declaration of strptime.  (It should be
	declared by the system headers.  If it turns out that some systems
	don't, we'll handle that then.)  (Thanks to Greg Troxel.)

	* stime.h: Renamed TIMEH --> STIMEH

	* backtrace.c (scm_display_error, scm_display_backtrace): In order
	to increase portability, don't use structure assignment.
	(Thanks to Nicolas Neuss.)

	* iselect.c: Use LONG_MAX instead of ULONG_MAX for increased
	portability.
	(finalize_fd_sets): Added empty statement after last case label.
	(Thanks to Nicolas Neuss.)

	* gc.c (scm_igc): Changed //-comment into /*-comment.  (Thanks to
	Nicolas Neuss.)

Sat Jul 11 22:08:21 1998  Mikael Djurfeldt  <mdj@totoro.red-bean.com>

	* init.c, readline.c: OK, I won't have these readline.x bug
 	reports anymore.  We've had them since April.  The current reason
	is a completely unintelligible failure of totoro.red-bean.com to
	do the test for rl_getc_function in libreadline correctly.  This
	kludge overrides the test if we're on totoro so that the snapshot
	generation process can work.

	* readline.c: Define a strdup replacement if not existent on system.

1998-07-12  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* vectors.c, vectors.h (scm_make_vector): Removed third argument.
	This change makes scm_make_vector R5RS compatible.  We cannot keep
	the third argument since people want to be able to deduce the form
	of the C function call only by looking at R5RS.  (At the same time
	we have removed some unnecessary complexity!)

	* eval.c, filesys.c, fluids.c, gc.c, gh_data.c, init.c, kw.c,
	net_db.c, posix.c, print.c, regex-posix.c, scmsigs.c, socket.c,
	stime.c, symbols.c, unif.c, vectors.c, weaks.c: Removed third
	argument in call to scm_make_vector.
	
1998-07-02  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* numbers.h (SCM_NUM2DBL): New macro.  Complements SCM_NUMBERP.
	This macro is useful in applications.

1998-06-21  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* load.c (scm_internal_parse_path): Renamed from scm_parse_path.
	(scm_parse_path, scm_search_path): New Scheme level procedures.

	* load.h (scm_internal_parse_path, scm_parse_path,
	scm_search_path): Declared.

1998-06-18  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* filesys.c (dirname, basename): New procedures.

	* init.c (scm_boot_guile_1): Removed condition around
	scm_init_options.

	* dynwind.c: #include "genio.h"; #include "smob.h"; Implemented a
	new data type (guards) for representation of C level guards and
	data on the wind chain.
	(scm_internal_dynamic_wind): New function.

	* dynwind.h: Declare scm_internal_dynamic_wind.
	
	* root.h (scm_root_state): Added scm_cur_loadp.

	* root.c (mark_root): Added comment about cur_loadp.

	* load.c: #include "dynwind.h";
	(scm_primitive_load): Use scm_inner_dynamic_wind to update
	scm_cur_loadp.

	* init.c (scm_init_standard_ports): Initialize scm_def_loadp.

	* ports.c (current-load-port): New procedure.

1998-06-09  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* ioext.c (scm_isatty_p): Accept any kind of data as argument.  If
	not a tty, return #f.

	* regex-posix.c (scm_regexp_exec): Free malloced memory.  (Thanks
	to Julian Satchell and Roland Kaufmann.)

	* gh.h (gh_memv, gh_member): Fixed typos in macro definitions.
	(gh_write):  Added declaration.  (Thanks to Eiichi Takamori.)

1998-06-07  Mikael Djurfeldt  <mdj@barbara.nada.kth.se>

	* debug.h, debug.c (scm_start_stack): New function.  Implements
	the guts of old scm_m_start_stack.

	* debug.c (scm_m_start_stack): Use scm_start_stack.

	* init.c (scm_start_stack, scm_restart_stack): Renamed to
	start_stack and restart_stack.  (These have static scope.)

1998-05-19  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* readline.c (rl_cleanup_after_signals, rl_free_line_state): New
	readline functions to come in release 2.3.  (Thanks to Chet
	Ramey.)
	(handle_errors): Use the above functions.

1998-05-12  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* readline.c: Improvements for readline support: Handle errors
	better; Implement before-read-hook.

1998-05-11  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* init.c (scm_boot_guile_1), readline.c: Test for
	HAVE_RL_GETC_FUNCTION instead of HAVE_LIBREADLINE.  (Need to
	assure that we have version >= 2.1.)

1998-05-11  Mikael Djurfeldt  <mdj@kenneth>

	* readline.c (scm_readline): Defer interrupts while we're calling
	readline.

	* readline.c (scm_add_history): Bugfix: Do strdup before giving
	away the string to add_history.
	(completion_function): Do completion for readline. (Thanks to
	Andrew Archibald.)
	(scm_filename_completion_function): New procedure: Filename
	completer.
	(current_input_getc): New function.  Use this one instead of
	standard getc from readline.

	* throw.c, throw.h (scm_handle_by_throw): New function: This
	handler throws errors to next handler on the dynwind chain.

1998-05-09  Mikael Djurfeldt  <mdj@kenneth>

	* scmsigs.c (scm_usleep): Bugfix: Don't define j if it isn't
	used.

1998-05-03  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* procprop.c (scm_i_procedure_arity): New function.  Returns arity
	of procedure.
	(scm_procedure_properties): Modified to return arity together with
	other properties.
	(scm_procedure_property): Added the read-only property `arity'.
	(scm_set_procedure_property_x): It is an error to set the `arity'
	property.

	* gsubr.h, gsubr.c: Moved macros from gsubr.c to gsubr.h and added
	prefix SCM_; Made f_gsubr_apply global and added prefix scm_.

	* procprop.h (scm_sym_arity): New symbol.

	* objects.c (scm_set_object_procedure_x): New procedure: Use this
	to set the dispatch procedure of an operator or entity object.

	* objects.h (SCM_METACLASS_OPERATOR_LAYOUT, SCM_ENTITY_LAYOUT):
	Made procedure slots read-only.

	* eval.c (SCM_CEVAL): Moved scm_tc7_contin case above
	scm_tcs_cons_gloc case in zero args switch; Fixed args
	construction for operators in scm_tcs_cons_gloc case in two args
	switch.

1998-05-02  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* fluids.c: Removed use of assert.h (in order to avoid
	__eprintf).

	* Makefile.am (libguile_la_LDFLAGS): Added -export-dynamic.

	* dynl.c (maybe_drag_in_eprintf): Disabled through #ifdef 0.

	* eval.c (SCM_CEVAL): Do more thorough argument checking.  This
	change makes the evaluator safer at the cost of evaluation speed.
	It handles the case when the user has added a non-immediate
	improper end of the application form, e.g., `(+ 0 . x)'.
	(Earlier only cases like `(+ 0 . 0)' were handled.)  I've tried to
	minimize the extra cost as much as possible.  The new code is
	enclosed in #ifdef CAUTIOUS regions.  NOTE: This also fixes the
	problem with structs planted directly in the code (e.g. by a
	macro).  This no longer causes segmentation fault.  (Thanks to
	Eric Hanchrow.)
	
	* eval.c, eval.h (scm_eval_args, scm_deval_args): Take one extra
	arg `proc' in order to be able to throw errors; New argument
	checking code.

	* Removed extra #include "debug.h"

1998-04-25  Mikael Djurfeldt  <mdj@kenneth>

	* scmsigs.c: Declare usleep as returning void on some systems.
	(scm_usleep): Return SCM_INUM0 on those systems. (Thanks to Julian
	Satchell.)

	* coop.c (usleep): Return void on some systems.

1998-04-20  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* Makefile.am (libguile_la_LDFLAGS): Removed redundant -rpath.

	* coop.c: Changed return type of usleep to int.

	* scmsigs.c (scm_usleep): New procedure; Declare usleep if it
	isn't found in the OS.

	* iselect.h: #define scm_internal_select select if GUILE_ISELECT
	isn't enabled.  (Thought that I had made this change ages ago...)

	* iselect.c: Declare bzero if not defined by OS.

1998-04-19  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* dynl.c (scm_must_free_argv): Fixed memory leak due to negated
	condition.  (Thanks to John Tobey.)

	* continuations.c (scm_make_cont), debug.c (scm_make_memoized,
	scm_make_debugobj), eval.c (scm_makprom): SCM_DEFER/ALLOW_INTS -->
	A section.

	* __scm.h: Start the long-term project of moving to POSIX threads.
	Phase 1: Classification of all critical sections.
	(SCM_ENTER_A_SECTION, SCM_EXIT_A_SECTION): New macros: Delimiters
	for A sections.  (See comments in __scm.h for details.)

	* dynl.c: Only check that HAVE_DLOPEN is defined before loading
	dynl-dl.c; Test on HAVE_LIBDLD instead of HAVE_DLD.

1998-04-18  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* Makefile.am (BUILT_SOURCES): Added cpp_err_symbols.c,
	cpp_sig_symbols.c, libpath.h and versiondat.h to BUILT_SOURCES
	(libpath.h, versiondat.h): Replaced dependency on Makefile with
	dependencies on $(srcdir)/Makefile.in
	$(top_builddir)/config.status in order to avoid circularity.

	* script.c (scm_compile_shell_switches): Bugfix: Don't discount i
	from argc if argc was 0 initially.

	* Makefile.am (Makefile.am): Replaced THREAD_LIBS --> GUILE_LIBS
	in generation of libpath.h.

1998-04-15  Mikael Djurfeldt  <mdj@nada.kth.se>

	* Makefile.am (libguile_la_LDFLAGS): Bumped version number of
	libguile from 2 to 3.

1998-04-14  Mikael Djurfeldt  <mdj@nada.kth.se>

	* Makefile.am: Added .x-dependencies to variable BUILT_SOURCES.

1998-04-13  Mikael Djurfeldt  <mdj@kenneth>

	* ports.c (scm_port_line, scm_set_port_line_x, scm_port_column,
	scm_set_port_column_x, scm_port_filename,
	scm_set_port_filename_x): Removed optional arguments.  Added
	proper argument checking.

	* eval.c, eval.h, coop.c (scm_eval_stack, SCM_EVAL_STACK): Measure
	stack size in machine words.

	* unif.c (scm_uniform_vector_ref, scm_cvref, scm_array_set_x,
	rapr1): Use SCM_UCHARS instead of SCM_CHARS for strings.  (Thanks