(derives_compute): Do not subtract NTOKENS from
the pointer DSET returned by malloc; this isn't portable.
Instead, always use DSET[i - NTOKENS] rather than DSET[i].
Similarly for DERIVES.

(derives_compute): Do not bother invoking
int_of_rule_number, since rule numbers are integers.

All uses of XCALLOC, XMALLOC, and XREALLOC changed to use new macros
CALLOC, MALLOC, REALLOC.  All uses of XFREE changed to free.

All uses of XCALLOC, XMALLOC, and XREALLOC changed to use new macros
CALLOC, MALLOC, REALLOC.  All uses of XFREE changed to free.

All uses of XCALLOC, XMALLOC, and XREALLOC changed to use new macros
CALLOC, MALLOC, REALLOC.  All uses of XFREE changed to free.

Don't subtract from the value that malloc returns; it's not portable.

(AC_CHECK_TYPES): Check for uintptr_t.

Change identifier spellings to avoid collisions with names
that are reserved by POSIX.
Fix misspelling in names.
Sort include-file uses.

(Torturing the Scanner): Rename value_t -> value,
value_t_as_yystype -> value_as_yystype.

(_AT_DATA_CALC_Y, Simple LALR Calculator):
Rename value_t -> value.

(G_EXPAND): Renamed from G_STRETCH.
(G_LONG_STRAIGHT_PHASE): Renamed from G_STRAIGHT_PHASE.
(N_EXPAND): Renamed from N_STRETCH.
(G_NODE_ALIGNMENT): Renamed from G_NODE_ALIGNEMENT.

(enum color): Renamed from enum color_e.
(enum textmode): Renamed from enum textmode_e.
(enum shape): Renamed from enum shape_e.
(struct colorentry): Renamed from struct colorentry_s.
(struct classname): Renamed from struct classname_s.
(struct infoname): Renamed from struct infoname_s.
(enum layoutalgorithm): Renamed from enum layoutalgorithm_e.
(enum decision): Renamed from enum decision_e.
(enum orientation): Renamed from enum orientation_e.
(enum alignment): Renamed from enum alignment_e.
(enum arrow_mode): Renamed form enum arrow_mode_e.
(enum crossing_type): Renamed from enum crossing_type_e.
(enum view): Renamed from enum view_e.
(struct node): Renamed from struct node_s.
(node): Renamed from node_t.
(enum linestyle): Renamed from enum linestyle_e.
(enum arrowstyle): Renamed from enum arrowstyle_e.
(struct edge): Renamed from struct edge.
(edge): Renamed from edge_t.
(struct graph): Renamed from struct graph_s.
(graph): Renamed from graph_t.
(struct node_s.expand): Renamed from struct node_s.stretch.
(struct graph_s.expand): Renamed from struct graph_s.stretch.
(enum_alignment_e): Renamed from enum_alignement_e.

Adjust to renamings.

(base_number): Renamed from base_t.

(vector_number): Renamed from vector_number_t.
(action_number): Renamed from action_t.
(VECTOR_NUMBER_MAX, VECTOR_NUMBER_MIN): Remove; unused.
(BASE_MAXIMUM): Renamed from BASE_MAX.
(BASE_MINIMUM): Renamed from BASE_MIN.
(ACTION_MAX): Remove; unused.
(ACTION_NUMBER_MINIMUM): Renamed from ACTION_MIN.

Do not include <errno.h>.
(TAB_EXT): Renamed from EXT_TAB.
(OUTPUT_EXT): Renamed from EXT_OUTPUT.
(time_report): Remove; decl is now in lib/timevar.h.

(symbol_number): Renamed from symbol_number_t.
(struct symbol): Renamed from struct symbol_s.
(symbol): Renamed from symbol_t.

(symbol_list): Renamed from symbol_list_t.
(struct symbol_list): Renamed from struct symbol_list_s.

(state_number): Renamed from state_number_t.
(struct state): Renamed from struct state_s.
(state): Renamed from state_t.
(transitions): Renamed from transitions_t.  Unused (and
misspelled) transtion_s tag removed.
(errs): Renamed from errs_t.  Unused errs_s tag removed.
(reductions): Renamed from reductions_t.  Unused tag
reductions_s removed.
(STATE_NUMBER_MAXIMUM): Renamed from STATE_NUMBER_MAX.
Include "gram.h".

Sort include files.

(<INITIAL>{int}): Use set_errno and get_errno instead of errno.
(<SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>\\x[0-9abcdefABCDEF]+): Likewise.
(handle_action_dollar, handle_action_at): Likewise.
(obstack_for_string): Renamed from string_obstack.

(relation_node): Renamed from relation_node_t.
(relation_nodes): Renamed from relation_nodes_t.
(relation): Renamed from relation_t.

(braced_code): Renamed from braced_code_t.
Remove brace_code_e tag.

(static_graph): Renamed from graph.

(<chars>): Renamed from <string>.

(output_skeleton): Use get_errno instead of errno.

(struct rule_list): Renamed from struct rule_list_s.
(rule_list): Renamed from rule_list_t.

(muscle_entry_t): Remove; moved to muscle_tab.c and renamed to muscle_entry.

(muscle_entry): Renamed from muscle_entry_t,
and moved here from muscle_tab.h.

(location): Renamed from location_t.

Adjust to type renaming.

(goto_number): Renamed from goto_number_t.
(GOTO_NUMBER_MAXIMUM): Renamed from GOTO_NUMBER_MAX.

(struct goto_list): Renamed from struct goto_list_s.
(goto_list): Renamed from goto_list_t.

(item_number): Renamed from item_number_t.
(rule_number): Renamed from rule_number_t.
(struct rule_s): Remove the "rule_s" part; not used.
(rule): Renamed from rule_t.
(rule_filter): Renamed from rule_filter_t.
(ITEM_NUMBER_MAX, ITEM_NUMBER_MIN, RULE_NUMBER_MAX):
Remove; unused.

(enum trace): Renamed from enum trace_e.
(enum report): Renamed from enum report_e.

Adjust to struniq renaming.

(strsuffix, stringappend): Remove; unused.

(xfopen, xfclose): Use get_errno instead of errno.
(strsuffix): Remove; unused.
(concat2): Renamed from stringappend.  Now static.

Include "gram.h".

(enum conflict_resolution): Renamed from enum conflict_resolution_e.

Don't include stdlib.h, string.h _LIBC stuff.

Adjust to type name changes.

(assoc): Renamed from assoc_t.

Adjust to type renamings.

(bison_SOURCES): Rename struniq.c and struniq.h to uniqstr.c and uniqstr.h.

(struct state_list): Renamed from struct state_list_s.
(state_list): Renamed from state_list_t.

(timevar_report): New decl.

(timevar_report): Renamed from time_report, for consistency with other
names.

(libbison_a_SOURCES): Add get-errno.h, get-errno.c.

(YYSTACK_GAP_MAXIMUM): Renamed from YYSTACK_GAP_MAX.

Rename struniq -> uniqstr to avoid collision with <string.h>.

Initial revision

Initial revision.

Rename struniq to uniqstr, to avoid collision with <string.h>.

* data/glr.c, data/lalr1.cc, data/yacc.c: Do not use __file__,
since this causes problems when __file__ contains character
sequences like "@" that are treated specially by src/scan-skel.l.
Instead, just use the file's basename.  This fixes the bug
reported by Martin Mokrejs in
<http://mail.gnu.org/pipermail/bug-bison/2002-December/001949.html>.

Add support for rules that do not have trailing semicolons, as
POSIX requires.  Improve the quality of locations in Bison
diagnostics.

Remove duplicate decl of current_file.
Fix minor promotion problem in templates.
Fix Debian Bug 102878.

(Invalid inputs): No longer `expecting ";"
or "|"' now that so many other tokens are allowed by the new grammar.

(Useless Rules, Reduced Automaton, Underivable Rules): Adjust reported
locations to match the more-precise results now expected.

(Invalid $n, Invalid @n, Type Clashes):
Adjust reported locations to match the more-precise
results now expected.

(S/R in initial, Defaulted Conflicted Reduction):
Adjust reported locations to match the more-precise
results now expected.

Include "files.h".
(YY_USER_INIT): Initialize scanner_cursor instead
of *loc.
(STEP): Remove.  No longer needed, now that adjust_location does
the work.  All uses removed.
(scanner_cursor): New var.
(adjust_location): Renamed from extend_location.  It now sets
*loc and adjusts the scanner cursor.  All uses changed.
Don't bother testing for CR.
(handle_syncline): Remove location arg; now updates scanner cursor.
All callers changed.
(unexpected_end_of_file): Now accepts start boundary of token or
comment, not location.  All callers changed.  Update scanner cursor,
not the location.
(SC_AFTER_IDENTIFIER): New state.
(context_state): Renamed from c_context.  All uses changed.
(id_loc, code_start, token_start): New local vars.
(<INITIAL,SC_AFTER_IDENTIFIER>): New initial context.  Move all
processing of Yacc white space and equivalents here.
(<INITIAL>{id}): Save id_loc.  Begin state SC_AFTER_IDENTIFIER
instead of returning ID immediately, since we need to search for
a subsequent colon.
(<INITIAL>"'", "\""): Save token_start.
(<INITIAL>"%{", "{", "%%"): Save code_start.
(<SC_AFTER_IDENTIFIER>): New state, looking for a colon.
(<SC_YACC_COMMENT>, <SC_COMMENT>, <SC_LINE_COMMENT>):
BEGIN context_state at end, not INITIAL.
(<SC_ESCAPED_STRING>"\"", <SC_ESCAPED_CHARACTER>"'",
<SC_BRACED_CODE>"}", <SC_PROLOGUE>"%}", <SC_EPILOGUE><<EOF>>):
Return correct token start.
(<SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE>): Save start boundary when
the start of a character, string or multiline comment is found.

(nonterminals_reduce): Use warn_at rather than rolling our own.

Include location.h.
(scanner_cursor): New decl.

Adjust to new location_t layout.

(lloc_default): New function, which handles
empty locations more accurately.
(YYLLOC_DEFAULT): Use it.
(%token COLON): Remove.
(%token ID_COLON): New token.
(rules): Use it.
(declarations, rules): Remove trailing semicolon.
(declaration, rules_or_grammar_declaration):
Allow empty (";") declaration.
(symbol_def): Remove empty actions; no longer needed.
(rules_or_grammar_declaration): Remove trailing semicolon.
(semi_colon.opt): Remove.

Do not include <quotearg.h>; no longer needed.
(boundary): New type.
(location_t): Use it.  This allows locations to span file boundaries.
All member uses changed: file -> start.file or end.file (as needed),
first_line -> start.line, first_column -> start.column,
last_line -> end.line, last_column -> end.column.
(equal_boundaries): New function.
(LOCATION_RESET, LOCATION_STEP): Remove.
(LOCATION_PRINT): Remove.  All callers changed to use location_print.
(empty_location): Now const.
(location_print): New decl.

* src/location.c: Include <quotearg.h>.
(empty_location): Now const.
(location_print): New function.  Follow the recommendation of the
GNU Coding Standards for locations that span file boundaries.

LOCATION_PRINT -> location_print

(current_file): Remove duplicate decl;
current_file is now owned by files.h.

Include files.h.  LOCATION_PRINT -> location_print.

(Table of Symbols): YYERROR_VERBOSE should
be #defined in the prologue, not in the Bison declarations.
This fixes Debian Bug 102878, reported by Shaul Karl.

(yy_reduce_print): Don't assume that yyrline[yyrule] promotes to int;
it might be unsigned int.

(yy_reduce_print): Don't assume that yyrline[yyrule]
promotes to int; it might be unsigned int.

Add strtoul.c
In src/scan-gram.l, move helper function definitions to epilogue.

(AC_REPLACE_FUNCS): Add strtoul.

New file, from gnulib.

(no_cr_read, extend_location): Move to epilogue,
and put only a forward declaration in the prologue.  This is for
consistency with the other scanner helper functions.

Type clashes now generate warnings, not errors, since it
appears that POSIX may allow some grammars with type clashes.

(grammar_current_rule_check): Warn about
type clashes instead of complaining.

(Type Clashes): Expect warnings, not complaints.

fund -> found

Add Yacc Library.
gram_error is static.
yylex and yyerror must be declared.

Describe Yacc library, lex-param and parse-param changes.

(gram_error): Now static.  Add static decl.
(print_token_value): Omit parameter names from forward decl,
for consistency.

(gram_error): Remove decl.

(Yacc Library): New node.  Regenerate top menu.

(lib_LIBRARIES, liby_a_SOURCES): New macros.

lib/main.c, lib/yyerror.c: New files.

Emphasize that yylex and yyerror must be declared before being used.
E.g., one should typically declare them in the prologue.  Use GNU
coding style in examples.  Put "const" consistently after the type it
modifies.  Mention that C99 supports "inline".  Mention that yyerror
traditionally returns "int".

[a-f] -> [abcdef], so that we don't assume the C locale.

%parse-param and %lex-param now take just one argument, the
declaration; the argument name is deduced from the declaration.

(Simple LALR Calculator): Adopt new convention for %parse-param
and %lex-param.

"," now elicits a warning, rather than being
a token; this is more compatible with byacc.

(add_param): New function.
(COMMA): Remove.
(declaration): Implement new rule for %parse-param and %lex-param.

(Parser Function, Pure Calling, Error Reporting, Table of Symbols):
%parse-param and %lex-param now take just one argument, the
declaration; the argument name is deduced from the declaration.

(reader): Remove gram_control; it wasn't being used.

Sync with gnulib.

Rename identifiers to avoid real and potential collisions.

(STEP): Renamed from YY_STEP.  All uses changed.
(STRING_GROW): Renamed from YY_OBS_GROW.  All uses changed.
(STRING_FINISH): Renamed from YY_OBS_FINISH.  All uses changed.
(STRING_FREE): Renamed from YY_OBS_FREE.  All uses changed.

(struct gram_control_s, gram_control_t): Remove.
(YY_DECL): Don't use "yy" at start of local variables.
All uses changed, e.g., yylloc -> loc.

(print_token_value): Renamed from yyprint.
All uses changed.
(YYPARSE_PARAM, YYLEX_PARAM, yycontrol): Remove.  All uses changed.
The name "yycontrol" violates the name space rules, and this stuff
wasn't being used anyway.
(input): Remove action; this stuff wasn't being used.
(gram_error): Rename local variable yylloc -> loc.

yyprint -> print_token_value

(_AT_CHECK_PRINTER_AND_DESTRUCTOR): yyout -> yyoutput,
to avoid collision with lex macro described by Bruce Lilly in
<http://mail.gnu.org/pipermail/bug-bison/2002-November/001929.html>.