path: root/ChangeLog

2009-09-29  Kaz Kylheku  <kkylheku@gmail.com>

	Version 013

	Some minor garbage collection issues fixed.

	Infinite looping bug fixed.

	New @(trailer) directive.

	* extract.y (match_files): Implemented trailer directive.
	* extract.l (version): Bump to 013.
	* lib.h (trailer): Declaration added.
	* lib.c (trailer): External definition added.
	(obj_init): Initializer trailer with interned symbol.
	* txr.1: Documented @(trailer) and bumped version to 013.

2009-09-29  Kaz Kylheku  <kkylheku@gmail.com>

	Looping bug fixed. Certain directives could cause an infinite
	loop if the query has run out of data.

	* extract.y (match_files): The semantics of the first_file_parsed
	argument changes a little bit. Previously, if nil was passed,
	a new lazy stream would be opened for the first file.
	But this is ambiguous because nil also means empty list;
	sometimes when we recurse into match_files, the data has ran
	out and this argument is thus nil. Now, that argument must be
	the symbol t in order to mean ``open the first file''.
	If the argument is nil, it unambiously means ``we are at the end of the
	current file; don't open anything''.
	(extract): The initial call to match_files now passes
	the symbol t for the first_file_parsed argument.

2009-09-29  Kaz Kylheku  <kkylheku@gmail.com>

	Fixing some gc issues. The test cases were found to bomb
	with an assertion when run with --gc-debug enabled,
	due to a garbage-collected object still being used.
	This was due to the way the main function was structured.

	Also, the stack ``top'' terminology in the gc was stupidly wrong. Leaf
	function frames are at the stack top, and main is near the bottom.
	I was thinking of the ``top caller''.

	* Makefile (TXR_DBG_OPTS): New variable.
	Tests are now run with --gc-debug, which makes them slower,
	but has much greater chance of trapping gc problems.
	* extract.l (main): Two variables are now used for determining the
	stack bottom. We don't know in which order the compiler
	places local variables into a stack frame.  (This is a separate
	question from that of the direction of stack growth).
	The call to the init function is now done right away.
	The argument processing section of main does some processing
	with GC objects, but the init function was being called afterward,
	before the list of interned symbols is protected from garbage
	collection! So with --gc-debug turned on, parts of the interned
	symbol list were being garbage collected (since the variable
	has not yet been added to the set of root pointers, which is
	done in the init function).  Also, the use of an unknown --long-option
	is diagnosed properly now.
	* gc.c (gc_stack_top): Renamed to gc_stack_bottom, and converted from
	extern to static.
	(mark): Follows rename of gc_stack_top to gc_stack_bottom.
	(sweep): Eliminated the freed variable for counting freed objects,
	and the associated debug message, which was not useful.
	Commented why the free list is managed differently when dbg
	is turned on.
	(gc_init): New function.
	* gc.h (gc_stack_top): Declaration removed.
	(gc_init): Declaration added.
	* lib.c (min): New macro.
	(init): Takes two additional arguments which are used to
	determine the stack bottom. The function first determiens
	whether the stack grows up or down. Then it takes the
	greater or smaller of the two potential stack top pointers,
	based on that. The result is passed go gc_init.
	* lib.h (init): Declaration updated.

2009-09-28  Kaz Kylheku  <kkylheku@gmail.com>

	Version 012

	Semantics change of @(until) in @(collect) and @(coll).

	Minor fixes.

	* extract.y (match_line, match_files): The until clauses
	continue to be processed after the main clauses of the collect
	or coll (to see the bindings), but are processed before
	the collection occurs, so that the until will veto
	the bindings of the last iteration.  Moreover, the data
	positions stays where it is when this happens, and no
	arrangement is made to match the until material again.

	* txr.1: Tried to document the change.


2009-09-27  Kaz Kylheku  <kkylheku@gmail.com>

	* txr.1: following proofread, fixed various escaping problems and
	instances of missing text.

2009-09-26  Kaz Kylheku  <kkylheku@gmail.com>

	* lib.c (equal): Bugfixes: wrong fallthrough of FUN case.
	VEC case must return nil, not break.

2009-09-26  Kaz Kylheku  <kkylheku@gmail.com>

	Preparation for some sorting support.

	* extract.y (merge): Renamed to weird_merge.
	(map_leaf_lists): New functino.
	(match_file): Follow weird_merge rename.
	* lib.c (all_satisfy, none_satisfy, string_lt, do_bind2other,
	bind2other, merge, do_sort, sort): New functions.
	* lib.h (all_satsify, none_satisfy, string_lt,
	bind2other, sort): Declared.

2009-09-25  Kaz Kylheku  <kkylheku@gmail.com>

	Version 011

	New @(maybe) clause optionally matches (does not fail if none of
	its clauses match anything).

	New blocks feature: allows a query or subquery to be
	abruptly terminated by invoking an exit to a named or anonymous
	block. @(collect) and @(skip) have implicit anonymous blocks now.

	The @(skip) directive takes a numeric argument now, which limits
	how many lines are searched.

	* Makefile, extract.l, extract.y, extract.h, gc.c, gc.h, lib.c, lib.h,
	regex.c, regex.h, txr.1, unwind.c, unwind.h: Copyright notice and
	license text updated or added, and version bumped up to 011.
	* tests/001/query-1.txr, tests/001/query-2.txr, tests/001/query-3.txr,
	tests/002/query-1.txr: Assigned to public domain.

2009-09-25  Kaz Kylheku  <kkylheku@gmail.com>

	New features:
	- named blocks;
	- maybe clause;
	- optional iteration bound on skip.

	* extract.y: includes added: "unwind.h", <setjmp.h>.
	(MAYBE, OR): New grammar tokens.
	(maybe_clause): New nonterminal grammar symbol.
	(expr): A NUMBER can be an expression now, so that @(skip 42)
	is valid syntax.
	(match_files): Support for numeric argument in skip directive
	to bound the search to a maximum number of lines.
	Anonymous block established around skip.
	New directives implemented: maybe, block, accept and fail.
	Anonymous block established around collect.
	* txr.1: Documentation updated with new features.
	* Makefile: new object file unwind.o, and associated rules.
	* extract.l (yybadtoken): New cases for MAYBE and OR.
	(grammar): Likewise.
	* lib.c (block, fail, accept): New symbol variables.
	(obj_init): New symbols interned.
	* lib.h (block, fail, accept): Declared.
	(if2, if3): Macros fixed so test expression is not compared to nil,
	but implicitly tested as boolean.
	* unwind.c, unwind.h: New source files.

2009-09-24  Kaz Kylheku  <kkylheku@gmail.com>

	Stability fixes.

	* extract.y (match_files): Fixed invalid string("-") to
	string(chk_strdup("-")) which caused a freeing of
	a non-malloced string at gc finalization time.
	* regex.c (nfa_state_shallow_free): New function: does not
	free satellite objects, just the structure itself.
	(nfa_combine): Use nfa_state_shallow_free instead of nfa_state_free,
	because the merged state inherits ownership of objects from the state
	being spliced out.
	(nfa_state_set): Fix lack of initialization of s.visited member of the
	state structure.

2009-09-24  Kaz Kylheku  <kkylheku@gmail.com>

	Version 010

	A file specs can start with $, which means read a directory.

	Data sources are not into memory at once, but on demand,
	which can reduce memory for many queries.

	Regular expressions are now compiled once, when the
	query is parsed.

	Character escapes are now supported in regular expressions,
	and as a special syntax.

	* extract.l (version): Bumped to 010.
	(grammar): 8 and 9 are not octal digits; handle all regex
	backslash escaping in lexical grammar.
	* extract.y (grammar): Get rid of backslash handling from
	regex grammar. Lexer returns a REGCHAR for every escaped
	item. In situations where an operator character is implicily
	literal, like * in a character class, we use the grammar
	to include that alongside REGCHAR. Bugfixes: the character ], when not
	closing a class, is not a syntax error but stands for itself;
	the character - stands for itself outside of character class;
	the | character is literal in a character class.
	* txr.1: Updated version. Documented character escapes.

2009-09-24  Kaz Kylheku  <kkylheku@gmail.com>

	Lazy stream list improvement: no extra NIL element caused
	by end-of-file. Requires push-back support in streams.
	To avoid introducing a new structure member into streams,
	we extend the semantics of the label member, and rename
	it to label_pushback.

	* lib.c (stdio_line_stream, pipe_line_stream,
	dirent_stream): Follow rename of struct stream member;
	assert that label is an atom.
	(stream_get): Check pushback stack first and get item from there.
	(stream_pushback): New function.
	(lazy_stream_func): Pull one more item from the stream and
	use /that/ to decide whether to continue the lazy stream.
	The extra item is pushed back, if valid.
	(lazy_stream_cons): Simplified: no hack involving regular cons. Starts
	the induction by peeking into the stream. If something is there, it is
	pushed back, and a lazy cons is constructed which will fetch it.
	(obj_print): Made aware of the pushback, which must be skipped
	to get to the terminating label.
	* lib.h (struct stream): Member renamed from label to label_pushback.
	(stream_pushback): New function declaration.

2009-09-23  Kaz Kylheku  <kkylheku@gmail.com>

	Escape syntax in regexes, and text. The
	standard seven character escapes are supported,
	namely \a, \b, \t, \n, \v, \f, and \r,
	as well as hex and octal escapes, plus
	the code \e for ASCII ESC.

	* extract.l (char_esc, num_esc): New functions.
	(grammar): New lex cases.
	* lib.c (obj_print): Support all character escapes in printing.
	Bugfix: backslash printed as two backslashes, not one.

2009-09-23  Kaz Kylheku  <kkylheku@gmail.com>

	* tests/002/query-1.txr: Modified to use $ to scan thread
	subdirectories.
	* tests/002/query-1.expected: Updated.

2009-09-23  Kaz Kylheku  <kkylheku@gmail.com>

	New COBJ type for wrapping arbitrary C objects into the
	Lisp-like framework. Compiled regexes are objects now.
	Regexes in a query are now compiled just once.

	* extract.y (grammar): Regexes compiled while parsing.
	(match_line): Modify with respect to the abstract syntax
	tree change, and the interface changes in the match_regex,
	and search_regex functions.
	* gc.c (mark_obj, finalize): Handle marking and finalization
	of COBJ objects.
	* lib.c (typeof, equal, obj_print): Handle COBJ.
	(cobj, cobj_print_op): New functions.
	* lib.h (type_t): New enum element, COBJ.
	(struct cobj, struct subj_ops): New types.
	(union obj): New member, co.
	(cobj, cobj_print_op): New functions declared.
	* regex.c (regex_equal, regex_destroy, regex_compile, regex_nfa): New
	functions.
	(regex_obj_ops): New static struct.
	(search_regex, match_regex): Interface change. Regex arguments
	are now compiled regexes. Functions won't handle raw regexes.
	* regex.h (regex_compile, regex_nfa): New functions declared.

2009-09-23  Kaz Kylheku  <kkylheku@gmail.com>

	New feature: file specs that start with $ read directories.
	Reading from an ``ls'' pipe is too slow.

	Streams and lazy conses implemented. Lazy conses allow us to treat a
	file or other kind of stream exactly as if it were a list.
	We can use car and cdr, etc. But only the parts of the list
	that we actually touch are instantiated on-the-fly by
	reading from the underlying stream.

	* extract.l: inclusion of <dirent.h> added.
	* extract.l: inclusion of <dirent.h> added.
	* extract.y (fpip_closedir): new enumeration in struct fpip,
	and fpip_noclose removed.
	(complex_open): Check for leading $, use opendir.
	(complex_open_failed): New function.
	(complex_close): Handle fpip_closedir case. Not closing
	stdin and stdout is handled by explicit comparison now.
	(complex_snarf): New function, constructs stream of
	a suitable type, over object returned from complex_close,
	wraps it in a lazy list.
	(match_files): Use complex_snarf instead of snarf to get a lazy list.
	* gc.c: Handle LCONS and STREAM cases.
	* lib.c (stream_t, lcons_t): New variables holding symbols.
	(typeof, equal, obj_print): Handle LCONS and STREAM.
	(car, cdr, car_l, cdr_l, consp, atom, listp): Rewritten to handle
	LCONS.
	(chk_strdup, stdio_line_read, stdio_line_write, stdio_close
	stdio_line_stream, pipe_close, pipe_line_stream,
	dirent_read, dirent_close, dirent_stream,
	stream_get, stream_put, stream_close,
	make_lazycons, lazy_stream_func, lazy_stream_cons): New functions.
	(stdio_line_stream_ops, pipe_line_stream_ops,
	dirent_stream_ops): New static structs.
	(obj_init): Intern new symbols lstream, lcons, and dir.
	* lib.h (type_t): New enum members STREAM and LCONS.
	(struct stream, struct stream_ops, struct lazy_cons): New types.
	(union obj): New members sm and lc.
	(chk_strdup, stdio_line_stream, pipe_line_stream,
	dirent_stream, stream_get, stream_put, stream_close,
	lazy_stream_cons): New function declarations.
	* regex.c: inclusion of <dirent.h> added

2009-09-23  Kaz Kylheku  <kkylheku@gmail.com>

	Version 009

	User-friendly error messages from parser.
	Fixed -q option.

	* extract.l (version): Bumped to 009.
	* txr.1: Updated version.

2009-09-22  Kaz Kylheku  <kkylheku@gmail.com>

	* Makefile (LIBLEX): New variable.
	Refer to lex library as -lfl, using variable
	that can be overridden.

2009-09-22  Kaz Kylheku  <kkylheku@gmail.com>

	* extract.h (yybadtoken): New function declaration.
	* extract.l (yybadtoken): New function.
	(main): Fixed -q option.
	* extract.y (grammar): Lots of new error productions, some
	phrase rules refactored, resulting in much more user-friendly
	error diagnosis.
	* txr.1: -q option semantics clarified.