Implement print-circle, allowing recursive and circular structures to be read. src/ChangeLog addition: 2011-09-04 Aidan Kehoe <kehoea@parhasard.net> * alloc.c: * alloc.c (ALLOC_FROB_BLOCK_LISP_OBJECT_1): * alloc.c (ALLOC_FROB_BLOCK_LISP_OBJECT): * alloc.c (cons_print_preprocess): * alloc.c (vector_print_preprocess): * alloc.c (vector_nsubst_structures_descend): * alloc.c (Fmake_symbol): * alloc.c (UNMARK_symbol): * alloc.c (sweep_symbols): * alloc.c (reinit_alloc_objects_early): * alloc.c (reinit_alloc_early): * bytecode.c: * bytecode.c (compiled_function_print_preprocess): * bytecode.c (compiled_function_nsubst_structures_descend): * bytecode.c (set_compiled_function_arglist): * bytecode.c (set_compiled_function_interactive): * bytecode.c (bytecode_objects_create): * chartab.c: * chartab.c (print_preprocess_mapper): * chartab.c (nsubst_structures_mapper): * chartab.c (char_table_nsubst_structures_descend): * chartab.c (chartab_objects_create): * elhash.c: * elhash.c (nsubst_structures_map_hash_table): * elhash.c (hash_table_nsubst_structures_descend): * elhash.c (print_preprocess_mapper): * elhash.c (hash_table_print_preprocess): * elhash.c (inchash_eq): * elhash.c (hash_table_objects_create): * elhash.c (syms_of_elhash): * elhash.h: * emacs.c (main_1): * fns.c: * fns.c (check_eq_nokey): * fns.c (Fnsubst): * fns.c (syms_of_fns): * lisp.h: * lisp.h (struct Lisp_Symbol): * lisp.h (IN_OBARRAY): * lisp.h (struct): * lisp.h (PRINT_PREPROCESS): * lread.c (read1): * lrecord.h: * lrecord.h (struct lrecord_implementation): * lrecord.h (DEFINE_DUMPABLE_MODULE_LISP_OBJECT): * print.c: * print.c (PRINT_CIRCLE_LIMIT): * print.c (print_continuous_numbering_changed): * print.c (print_prepare): * print.c (print_finish): * print.c (Fprin1_to_string): * print.c (print_cons): * print.c (print_preprocess_inchash_eq): * print.c (print_preprocess): * print.c (print_sort_get_numbers): * print.c (print_sort_compare_ordinals): * print.c (print_gensym_or_circle): * print.c (nsubst_structures_descend): * print.c (nsubst_structures): * print.c (print_internal): * print.c (print_symbol): * print.c (vars_of_print): * rangetab.c: * rangetab.c (range_table_print_preprocess): * rangetab.c (range_table_nsubst_structures_descend): * rangetab.c (rangetab_objects_create): * rangetab.c (syms_of_rangetab): * symbols.c: * symbols.c (symbol_print_preprocess): * symbols.c (Fintern): * symbols.c (Funintern): * symbols.c (reinit_symbol_objects_early): * symbols.c (init_symbols_once_early): * symsinit.h: Implement print-circle, printing circular structures in a readable fashion, and treating them appropriately on read. This is by means of two new object methods, print_preprocess (detecting circularities), and nsubst_structures_descend (replacing placeholders with the read objects). Expose the substitution to Lisp via #'nsubst and its new :descend-structures keyword. Store information as to whether symbols are interned in obarray or not in their header, making checking for keywords and uninterned symbols (and thus printing) cheaper. Default print_gensym to t, as Common Lisp does, and as a more-than-decade old comment suggests. lisp/ChangeLog addition: 2011-09-04 Aidan Kehoe <kehoea@parhasard.net> * bytecomp.el (byte-compile-output-file-form): * bytecomp.el (byte-compile-output-docform): Bind print-circle, print-continuous-numbering in these functions, now those variables are available. * lisp.el (forward-sexp): * lisp.el (backward-sexp): Recognise leading #N= as being part of an expression. tests/ChangeLog addition: 2011-09-04 Aidan Kehoe <kehoea@parhasard.net> * automated/lisp-reader-tests.el: * automated/lisp-tests.el (literal-with-uninterned): * automated/symbol-tests.el (foo): Test print-circle, for printing (mutually-)recursive and circular structures. Bind print-continuous-numbering where appropriate.

--- a/lisp/lisp.el	Sun Aug 28 10:34:54 2011 +0100
+++ b/lisp/lisp.el	Sun Sep 04 19:51:35 2011 +0100
@@ -69,7 +69,8 @@
     (let* ((diff (- (point) (point-min)))
 	   (subject (buffer-substring (- (point) (min diff 3))
 				      (1+ (point))))
-	   (matched (string-match "#s(\\|#r[uU]\\{0,1\\}\"" subject)))
+	   (matched (string-match "#[0-9]+=\\|#s(\\|#r[uU]\\{0,1\\}\""
+				  subject)))
       (if matched
 	(goto-char (1+ (- (point) (- (length subject) matched))))))))