* Changes in version 2.5 (2011-05-14):
** Grammar symbol names can now contain non-initial dashes:
Consistently with directives (such as %error-verbose) and with
%define variables (e.g. push-pull), grammar symbol names may contain
dashes in any position except the beginning. This is a GNU
extension over POSIX Yacc. Thus, use of this extension is reported
by -Wyacc and rejected in Yacc mode (--yacc).
** Named references:
Historically, Yacc and Bison have supported positional references
($n, $$) to allow access to symbol values from inside of semantic
actions code.
Starting from this version, Bison can also accept named references.
When no ambiguity is possible, original symbol names may be used
as named references:
if_stmt : "if" cond_expr "then" then_stmt ';'
{ $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
In the more common case, explicit names may be declared:
stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
{ $res = mk_if_stmt($cond, $then, $else); }
Location information is also accessible using @name syntax. When
accessing symbol names containing dots or dashes, explicit bracketing
($[sym.1]) must be used.
These features are experimental in this version. More user feedback
will help to stabilize them.
** IELR(1) and canonical LR(1):
IELR(1) is a minimal LR(1) parser table generation algorithm. That
is, given any context-free grammar, IELR(1) generates parser tables
with the full language-recognition power of canonical LR(1) but with
nearly the same number of parser states as LALR(1). This reduction
in parser states is often an order of magnitude. More importantly,
because canonical LR(1)'s extra parser states may contain duplicate
conflicts in the case of non-LR(1) grammars, the number of conflicts
for IELR(1) is often an order of magnitude less as well. This can
significantly reduce the complexity of developing of a grammar.
Bison can now generate IELR(1) and canonical LR(1) parser tables in
place of its traditional LALR(1) parser tables, which remain the
default. You can specify the type of parser tables in the grammar
file with these directives:
%define lr.type lalr
%define lr.type ielr
%define lr.type canonical-lr
The default-reduction optimization in the parser tables can also be
adjusted using `%define lr.default-reductions'. For details on both
of these features, see the new section `Tuning LR' in the Bison
manual.
These features are experimental. More user feedback will help to
stabilize them.
** LAC (Lookahead Correction) for syntax error handling:
Canonical LR, IELR, and LALR can suffer from a couple of problems
upon encountering a syntax error. First, the parser might perform
additional parser stack reductions before discovering the syntax
error. Such reductions can perform user semantic actions that are
unexpected because they are based on an invalid token, and they
cause error recovery to begin in a different syntactic context than
the one in which the invalid token was encountered. Second, when
verbose error messages are enabled (with %error-verbose or the
obsolete `#define YYERROR_VERBOSE'), the expected token list in the
syntax error message can both contain invalid tokens and omit valid
tokens.
The culprits for the above problems are %nonassoc, default
reductions in inconsistent states, and parser state merging. Thus,
IELR and LALR suffer the most. Canonical LR can suffer only if
%nonassoc is used or if default reductions are enabled for
inconsistent states.
LAC is a new mechanism within the parsing algorithm that solves
these problems for canonical LR, IELR, and LALR without sacrificing
%nonassoc, default reductions, or state merging. When LAC is in
use, canonical LR and IELR behave almost exactly the same for both
syntactically acceptable and syntactically unacceptable input.
While LALR still does not support the full language-recognition
power of canonical LR and IELR, LAC at least enables LALR's syntax
error handling to correctly reflect LALR's language-recognition
power.
Currently, LAC is only supported for deterministic parsers in C.
You can enable LAC with the following directive:
%define parse.lac full
See the new section `LAC' in the Bison manual for additional
details including a few caveats.
LAC is an experimental feature. More user feedback will help to
stabilize it.
** %define improvements:
*** Can now be invoked via the command line:
Each of these command-line options
-D NAME[=VALUE]
--define=NAME[=VALUE]
-F NAME[=VALUE]
--force-define=NAME[=VALUE]
is equivalent to this grammar file declaration
%define NAME ["VALUE"]
except that the manner in which Bison processes multiple definitions
for the same NAME differs. Most importantly, -F and --force-define
quietly override %define, but -D and --define do not. For further
details, see the section `Bison Options' in the Bison manual.
*** Variables renamed:
The following %define variables
api.push_pull
lr.keep_unreachable_states
have been renamed to
api.push-pull
lr.keep-unreachable-states
The old names are now deprecated but will be maintained indefinitely
for backward compatibility.
*** Values no longer need to be quoted in the grammar file:
If a %define value is an identifier, it no longer needs to be placed
within quotations marks. For example,
%define api.push-pull "push"
can be rewritten as
%define api.push-pull push
*** Unrecognized variables are now errors not warnings.
*** Multiple invocations for any variable is now an error not a warning.
** Unrecognized %code qualifiers are now errors not warnings.
** Character literals not of length one:
Previously, Bison quietly converted all character literals to length
one. For example, without warning, Bison interpreted the operators in
the following grammar to be the same token:
exp: exp '++'
| exp '+' exp
;
Bison now warns when a character literal is not of length one. In
some future release, Bison will start reporting an error instead.
** Destructor calls fixed for lookaheads altered in semantic actions:
Previously for deterministic parsers in C, if a user semantic action
altered yychar, the parser in some cases used the old yychar value to
determine which destructor to call for the lookahead upon a syntax
error or upon parser return. This bug has been fixed.
** C++ parsers use YYRHSLOC:
Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
macro and use it in the default YYLLOC_DEFAULT. You are encouraged
to use it. If, for instance, your location structure has `first'
and `last' members, instead of
# define YYLLOC_DEFAULT(Current, Rhs, N) \
do \
if (N) \
{ \
(Current).first = (Rhs)[1].location.first; \
(Current).last = (Rhs)[N].location.last; \
} \
else \
{ \
(Current).first = (Current).last = (Rhs)[0].location.last; \
} \
while (false)
use:
# define YYLLOC_DEFAULT(Current, Rhs, N) \
do \
if (N) \
{ \
(Current).first = YYRHSLOC (Rhs, 1).first; \
(Current).last = YYRHSLOC (Rhs, N).last; \
} \
else \
{ \
(Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
} \
while (false)
** YYLLOC_DEFAULT in C++:
The default implementation of YYLLOC_DEFAULT used to be issued in
the header file. It is now output in the implementation file, after
the user %code sections so that its #ifndef guard does not try to
override the user's YYLLOC_DEFAULT if provided.
** YYFAIL now produces warnings and Java parsers no longer implement it:
YYFAIL has existed for many years as an undocumented feature of
deterministic parsers in C generated by Bison. More recently, it was
a documented feature of Bison's experimental Java parsers. As
promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
semantic action now produces a deprecation warning, and Java parsers
no longer implement YYFAIL at all. For further details, including a
discussion of how to suppress C preprocessor warnings about YYFAIL
being unused, see the Bison 2.4.2 NEWS entry.
** Temporary hack for adding a semicolon to the user action:
Previously, Bison appended a semicolon to every user action for
reductions when the output language defaulted to C (specifically, when
neither %yacc, %language, %skeleton, or equivalent command-line
options were specified). This allowed actions such as
exp: exp "+" exp { $$ = $1 + $3 };
instead of
exp: exp "+" exp { $$ = $1 + $3; };
As a first step in removing this misfeature, Bison now issues a
warning when it appends a semicolon. Moreover, in cases where Bison
cannot easily determine whether a semicolon is needed (for example, an
action ending with a cpp directive or a braced compound initializer),
it no longer appends one. Thus, the C compiler might now complain
about a missing semicolon where it did not before. Future releases of
Bison will cease to append semicolons entirely.
** Verbose syntax error message fixes:
When %error-verbose or the obsolete `#define YYERROR_VERBOSE' is
specified, syntax error messages produced by the generated parser
include the unexpected token as well as a list of expected tokens.
The effect of %nonassoc on these verbose messages has been corrected
in two ways, but a more complete fix requires LAC, described above:
*** When %nonassoc is used, there can exist parser states that accept no
tokens, and so the parser does not always require a lookahead token
in order to detect a syntax error. Because no unexpected token or
expected tokens can then be reported, the verbose syntax error
message described above is suppressed, and the parser instead
reports the simpler message, `syntax error'. Previously, this
suppression was sometimes erroneously triggered by %nonassoc when a
lookahead was actually required. Now verbose messages are
suppressed only when all previous lookaheads have already been
shifted or discarded.
*** Previously, the list of expected tokens erroneously included tokens
that would actually induce a syntax error because conflicts for them
were resolved with %nonassoc in the current parser state. Such
tokens are now properly omitted from the list.
*** Expected token lists are still often wrong due to state merging
(from LALR or IELR) and default reductions, which can both add
invalid tokens and subtract valid tokens. Canonical LR almost
completely fixes this problem by eliminating state merging and
default reductions. However, there is one minor problem left even
when using canonical LR and even after the fixes above. That is,
if the resolution of a conflict with %nonassoc appears in a later
parser state than the one at which some syntax error is
discovered, the conflicted token is still erroneously included in
the expected token list. Bison's new LAC implementation,
described above, eliminates this problem and the need for
canonical LR. However, LAC is still experimental and is disabled
by default.
** Java skeleton fixes:
*** A location handling bug has been fixed.
*** The top element of each of the value stack and location stack is now
cleared when popped so that it can be garbage collected.
*** Parser traces now print the top element of the stack.
** -W/--warnings fixes:
*** Bison now properly recognizes the `no-' versions of categories:
For example, given the following command line, Bison now enables all
warnings except warnings for incompatibilities with POSIX Yacc:
bison -Wall,no-yacc gram.y
*** Bison now treats S/R and R/R conflicts like other warnings:
Previously, conflict reports were independent of Bison's normal
warning system. Now, Bison recognizes the warning categories
`conflicts-sr' and `conflicts-rr'. This change has important
consequences for the -W and --warnings command-line options. For
example:
bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
bison -Wnone gram.y # no conflicts are reported
bison -Werror gram.y # any conflict is an error
However, as before, if the %expect or %expect-rr directive is
specified, an unexpected number of conflicts is an error, and an
expected number of conflicts is not reported, so -W and --warning
then have no effect on the conflict report.
*** The `none' category no longer disables a preceding `error':
For example, for the following command line, Bison now reports
errors instead of warnings for incompatibilities with POSIX Yacc:
bison -Werror,none,yacc gram.y
*** The `none' category now disables all Bison warnings:
Previously, the `none' category disabled only Bison warnings for
which there existed a specific -W/--warning category. However,
given the following command line, Bison is now guaranteed to
suppress all warnings:
bison -Wnone gram.y
** Precedence directives can now assign token number 0:
Since Bison 2.3b, which restored the ability of precedence
directives to assign token numbers, doing so for token number 0 has
produced an assertion failure. For example:
%left END 0
This bug has been fixed.
* Changes in version 2.4.3 (2010-08-05):
** Bison now obeys -Werror and --warnings=error for warnings about
grammar rules that are useless in the parser due to conflicts.
** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
been fixed.
** Failures in the test suite for GCC 4.5 have been fixed.
** Failures in the test suite for some versions of Sun Studio C++ have
been fixed.
** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
warnings about undefined %prec identifiers will not be converted to
errors in Bison 2.5. They will remain warnings, which should be
sufficient for POSIX while avoiding backward compatibility issues.
** Minor documentation fixes.
* Changes in version 2.4.2 (2010-03-20):
** Some portability problems that resulted in failures and livelocks
in the test suite on some versions of at least Solaris, AIX, HP-UX,
RHEL4, and Tru64 have been addressed. As a result, fatal Bison
errors should no longer cause M4 to report a broken pipe on the
affected platforms.
** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately.
POSIX specifies that an error be reported for any identifier that does
not appear on the LHS of a grammar rule and that is not defined by
%token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
error report for the case when an identifier appears only after a
%prec directive. It is now restored. However, for backward
compatibility with recent Bison releases, it is only a warning for
now. In Bison 2.5 and later, it will return to being an error.
** Detection of GNU M4 1.4.6 or newer during configure is improved.
** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
avoided.
** %code is now a permanent feature.
A traditional Yacc prologue directive is written in the form:
%{CODE%}
To provide a more flexible alternative, Bison 2.3b introduced the
%code directive with the following forms for C/C++:
%code {CODE}
%code requires {CODE}
%code provides {CODE}
%code top {CODE}
These forms are now considered permanent features of Bison. See the
%code entries in the section "Bison Declaration Summary" in the Bison
manual for a summary of their functionality. See the section
"Prologue Alternatives" for a detailed discussion including the
advantages of %code over the traditional Yacc prologue directive.
Bison's Java feature as a whole including its current usage of %code
is still considered experimental.
** YYFAIL is deprecated and will eventually be removed.
YYFAIL has existed for many years as an undocumented feature of
deterministic parsers in C generated by Bison. Previously, it was
documented for Bison's experimental Java parsers. YYFAIL is no longer
documented for Java parsers and is formally deprecated in both cases.
Users are strongly encouraged to migrate to YYERROR, which is
specified by POSIX.
Like YYERROR, you can invoke YYFAIL from a semantic action in order to
induce a syntax error. The most obvious difference from YYERROR is
that YYFAIL will automatically invoke yyerror to report the syntax
error so that you don't have to. However, there are several other
subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
used. For a more detailed discussion, see:
http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
deterministic parsers in C will continue to implement it. However,
because YYFAIL is already flawed, it seems futile to try to make new
Bison features compatible with it. Thus, during parser generation,
Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
rule action. In a later release, YYFAIL will be disabled for
%error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
be removed altogether.
There exists at least one case where Bison 2.5's YYFAIL warning will
be a false positive. Some projects add phony uses of YYFAIL and other
Bison-defined macros for the sole purpose of suppressing C
preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
To avoid Bison's future warning, such YYFAIL uses can be moved to the
epilogue (that is, after the second `%%') in the Bison input file. In
this release (2.4.2), Bison already generates its own code to suppress
C preprocessor warnings for YYFAIL, so projects can remove their own
phony uses of YYFAIL if compatibility with Bison releases prior to
2.4.2 is not necessary.
** Internationalization.
Fix a regression introduced in Bison 2.4: Under some circumstances,
message translations were not installed although supported by the
host system.
The tools framework changed m4 on Darwin<9 from the native one to the
pkgsrc version last month. bison needs a PKGREVISION bump because it
embeds the path to m4 in its binary.
* Changes in version 2.4.1 (2008-12-11):
** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
declarations have been fixed.
** Temporary hack for adding a semicolon to the user action.
Bison used to prepend a trailing semicolon at the end of the user
action for reductions. This allowed actions such as
exp: exp "+" exp { $$ = $1 + $3 };
instead of
exp: exp "+" exp { $$ = $1 + $3; };
Some grammars still depend on this `feature'. Bison 2.4.1 restores
the previous behavior in the case of C output (specifically, when
neither %language or %skeleton or equivalent command-line options
are used) to leave more time for grammars depending on the old
behavior to be adjusted. Future releases of Bison will disable this
feature.
** A few minor improvements to the Bison manual.
a recent enough version of gettext(3) e.g. Mac OS X or Linux. Dynamically
adjust the package list depending on the configure result.
Bump the package revisions because the package list was incorrect on
various platforms.
Changes in version 2.4 (2008-11-02):
* %language is an experimental feature.
We first introduced this feature in test release 2.3b as a cleaner
alternative to %skeleton. Since then, we have discussed the possibility of
modifying its effect on Bison's output file names. Thus, in this release,
we consider %language to be an experimental feature that will likely evolve
in future releases.
* Forward compatibility with GNU M4 has been improved.
* Several bugs in the C++ skeleton and the experimental Java skeleton have been
fixed.
Changes in version 2.3b (2008-05-27):
* The quotes around NAME that used to be required in the following directive
are now deprecated:
%define NAME "VALUE"
* The directive `%pure-parser' is now deprecated in favor of:
%define api.pure
which has the same effect except that Bison is more careful to warn about
unreasonable usage in the latter case.
* Push Parsing
Bison can now generate an LALR(1) parser in C with a push interface. That
is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
push one token at a time to the parser using `yypush_parse', which will
return to the caller after processing each token. By default, the push
interface is disabled. Either of the following directives will enable it:
%define api.push_pull "push" // Just push; does not require yylex.
%define api.push_pull "both" // Push and pull; requires yylex.
See the new section `A Push Parser' in the Bison manual for details.
The current push parsing interface is experimental and may evolve. More user
feedback will help to stabilize it.
* The -g and --graph options now output graphs in Graphviz DOT format,
not VCG format. Like --graph, -g now also takes an optional FILE argument
and thus cannot be bundled with other short options.
* Java
Bison can now generate an LALR(1) parser in Java. The skeleton is
`data/lalr1.java'. Consider using the new %language directive instead of
%skeleton to select it.
See the new section `Java Parsers' in the Bison manual for details.
The current Java interface is experimental and may evolve. More user
feedback will help to stabilize it.
* %language
This new directive specifies the programming language of the generated
parser, which can be C (the default), C++, or Java. Besides the skeleton
that Bison uses, the directive affects the names of the generated files if
the grammar file's name ends in ".y".
* XML Automaton Report
Bison can now generate an XML report of the LALR(1) automaton using the new
`--xml' option. The current XML schema is experimental and may evolve. More
user feedback will help to stabilize it.
* The grammar file may now specify the name of the parser header file using
%defines. For example:
%defines "parser.h"
* When reporting useless rules, useless nonterminals, and unused terminals,
Bison now employs the terms "useless in grammar" instead of "useless",
"useless in parser" instead of "never reduced", and "unused in grammar"
instead of "unused".
* Unreachable State Removal
Previously, Bison sometimes generated parser tables containing unreachable
states. A state can become unreachable during conflict resolution if Bison
disables a shift action leading to it from a predecessor state. Bison now:
1. Removes unreachable states.
2. Does not report any conflicts that appeared in unreachable states.
WARNING: As a result, you may need to update %expect and %expect-rr
directives in existing grammar files.
3. For any rule used only in such states, Bison now reports the rule as
"useless in parser due to conflicts".
This feature can be disabled with the following directive:
%define lr.keep_unreachable_states
See the %define entry in the `Bison Declaration Summary' in the Bison manual
for further discussion.
* Lookahead Set Correction in the `.output' Report
When instructed to generate a `.output' file including lookahead sets
(using `--report=lookahead', for example), Bison now prints each reduction's
lookahead set only next to the associated state's one item that (1) is
associated with the same rule as the reduction and (2) has its dot at the end
of its RHS. Previously, Bison also erroneously printed the lookahead set
next to all of the state's other items associated with the same rule. This
bug affected only the `.output' file and not the generated parser source
code.
* --report-file=FILE is a new option to override the default `.output' file
name.
* The `=' that used to be required in the following directives is now
deprecated:
%file-prefix "parser"
%name-prefix "c_"
%output "parser.c"
* An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
Bison 2.3a provided a new set of directives as a more flexible alternative to
the traditional Yacc prologue blocks. Those have now been consolidated into
a single %code directive with an optional qualifier field, which identifies
the purpose of the code and thus the location(s) where Bison should generate
it:
1. `%code {CODE}' replaces `%after-header {CODE}'
2. `%code requires {CODE}' replaces `%start-header {CODE}'
3. `%code provides {CODE}' replaces `%end-header {CODE}'
4. `%code top {CODE}' replaces `%before-header {CODE}'
See the %code entries in section `Bison Declaration Summary' in the Bison
manual for a summary of the new functionality. See the new section `Prologue
Alternatives' for a detailed discussion including the advantages of %code
over the traditional Yacc prologues.
The prologue alternatives are experimental. More user feedback will help to
determine whether they should become permanent features.
* Revised warning: unset or unused mid-rule values
Since Bison 2.2, Bison has warned about mid-rule values that are set but not
used within any of the actions of the parent rule. For example, Bison warns
about unused $2 in:
exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
Now, Bison also warns about mid-rule values that are used but not set. For
example, Bison warns about unset $$ in the mid-rule action in:
exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
However, Bison now disables both of these warnings by default since they
sometimes prove to be false alarms in existing grammars employing the Yacc
constructs $0 or $-N (where N is some positive integer).
To enable these warnings, specify the option `--warnings=midrule-values' or
`-W', which is a synonym for `--warnings=all'.
* Default %destructor or %printer with `<*>' or `<>'
Bison now recognizes two separate kinds of default %destructor's and
%printer's:
1. Place `<*>' in a %destructor/%printer symbol list to define a default
%destructor/%printer for all grammar symbols for which you have formally
declared semantic type tags.
2. Place `<>' in a %destructor/%printer symbol list to define a default
%destructor/%printer for all grammar symbols without declared semantic
type tags.
Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
`<*>' and `<>' combined achieve the same effect with one exception: Bison no
longer applies any %destructor to a mid-rule value if that mid-rule value is
not actually ever referenced using either $$ or $n in a semantic action.
The default %destructor's and %printer's are experimental. More user
feedback will help to determine whether they should become permanent
features.
See the section `Freeing Discarded Symbols' in the Bison manual for further
details.
* %left, %right, and %nonassoc can now declare token numbers. This is required
by POSIX. However, see the end of section `Operator Precedence' in the Bison
manual for a caveat concerning the treatment of literal strings.
* The nonfunctional --no-parser, -n, and %no-parser options have been
completely removed from Bison.
Changes in version 2.3a, 2006-09-13:
* Instead of %union, you can define and use your own union type
YYSTYPE if your grammar contains at least one <type> tag.
Your YYSTYPE need not be a macro; it can be a typedef.
This change is for compatibility with other Yacc implementations,
and is required by POSIX.
* Locations columns and lines start at 1.
In accordance with the GNU Coding Standards and Emacs.
* You may now declare per-type and default %destructor's and %printer's:
For example:
%union { char *string; }
%token <string> STRING1
%token <string> STRING2
%type <string> string1
%type <string> string2
%union { char character; }
%token <character> CHR
%type <character> chr
%destructor { free ($$); } %symbol-default
%destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
%destructor { } <character>
guarantees that, when the parser discards any user-defined symbol that has a
semantic type tag other than `<character>', it passes its semantic value to
`free'. However, when the parser discards a `STRING1' or a `string1', it
also prints its line number to `stdout'. It performs only the second
`%destructor' in this case, so it invokes `free' only once.
[Although we failed to mention this here in the 2.3a release, the default
%destructor's and %printer's were experimental, and they were rewritten in
future versions.]
* Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
`--yacc', or `%yacc'), Bison no longer generates #define statements for
associating token numbers with token names. Removing the #define statements
helps to sanitize the global namespace during preprocessing, but POSIX Yacc
requires them. Bison still generates an enum for token names in all cases.
* Handling of traditional Yacc prologue blocks is now more consistent but
potentially incompatible with previous releases of Bison.
As before, you declare prologue blocks in your grammar file with the
`%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
prologue blocks that you've declared before the first %union. To generate
the post-prologue, Bison concatenates all prologue blocks that you've
declared after the first %union.
Previous releases of Bison inserted the pre-prologue into both the header
file and the code file in all cases except for LALR(1) parsers in C. In the
latter case, Bison inserted it only into the code file. For parsers in C++,
the point of insertion was before any token definitions (which associate
token numbers with names). For parsers in C, the point of insertion was
after the token definitions.
Now, Bison never inserts the pre-prologue into the header file. In the code
file, it always inserts it before the token definitions.
* Bison now provides a more flexible alternative to the traditional Yacc
prologue blocks: %before-header, %start-header, %end-header, and
%after-header.
For example, the following declaration order in the grammar file reflects the
order in which Bison will output these code blocks. However, you are free to
declare these code blocks in your grammar file in whatever order is most
convenient for you:
%before-header {
/* Bison treats this block like a pre-prologue block: it inserts it into
* the code file before the contents of the header file. It does *not*
* insert it into the header file. This is a good place to put
* #include's that you want at the top of your code file. A common
* example is `#include "system.h"'. */
}
%start-header {
/* Bison inserts this block into both the header file and the code file.
* In both files, the point of insertion is before any Bison-generated
* token, semantic type, location type, and class definitions. This is a
* good place to define %union dependencies, for example. */
}
%union {
/* Unlike the traditional Yacc prologue blocks, the output order for the
* new %*-header blocks is not affected by their declaration position
* relative to any %union in the grammar file. */
}
%end-header {
/* Bison inserts this block into both the header file and the code file.
* In both files, the point of insertion is after the Bison-generated
* definitions. This is a good place to declare or define public
* functions or data structures that depend on the Bison-generated
* definitions. */
}
%after-header {
/* Bison treats this block like a post-prologue block: it inserts it into
* the code file after the contents of the header file. It does *not*
* insert it into the header file. This is a good place to declare or
* define internal functions or data structures that depend on the
* Bison-generated definitions. */
}
If you have multiple occurrences of any one of the above declarations, Bison
will concatenate the contents in declaration order.
[Although we failed to mention this here in the 2.3a release, the prologue
alternatives were experimental, and they were rewritten in future versions.]
* The option `--report=look-ahead' has been changed to `--report=lookahead'.
The old spelling still works, but is not documented and may be removed
in a future release.
Changes in version 2.3, 2006-06-05:
* GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
for compatibility with LALR(1) grammars.
* It is now documented that any definition of YYSTYPE or YYLTYPE should
be to a type name that does not contain parentheses or brackets.
* The distribution terms for all Bison-generated parsers now permit
using the parsers in nonfree programs. Previously, this permission
was granted only for Bison-generated LALR(1) parsers in C.
* %name-prefix changes the namespace name in C++ outputs.
* The C++ parsers export their token_type.
* Bison now allows multiple %union declarations, and concatenates
their contents together.
* New warning: unused values
Right-hand side symbols whose values are not used are reported,
if the symbols have destructors. For instance:
exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
| exp "+" exp
;
will trigger a warning about $$ and $5 in the first rule, and $3 in
the second ($1 is copied to $$ by the default rule). This example
most likely contains three errors, and could be rewritten as:
exp: exp "?" exp ":" exp
{ $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
| exp "+" exp
{ $$ = $1 ? $1 : $3; if ($1) free ($3); }
;
However, if the original actions were really intended, memory leaks
and all, the warnings can be suppressed by letting Bison believe the
values are used, e.g.:
exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
| exp "+" exp { $$ = $1; (void) $3; }
;
If there are mid-rule actions, the warning is issued if no action
uses it. The following triggers no warning: $1 and $3 are used.
exp: exp { push ($1); } '+' exp { push ($3); sum (); };
The warning is intended to help catching lost values and memory leaks.
If a value is ignored, its associated memory typically is not reclaimed.
* %destructor vs. YYABORT, YYACCEPT, and YYERROR.
Destructors are now called when user code invokes YYABORT, YYACCEPT,
and YYERROR, for all objects on the stack, other than objects
corresponding to the right-hand side of the current rule.
* %expect, %expect-rr
Incorrect numbers of expected conflicts are now actual errors,
instead of warnings.
* GLR, YACC parsers.
The %parse-params are available in the destructors (and the
experimental printers) as per the documentation.
* Bison now warns if it finds a stray `$' or `@' in an action.
* %require "VERSION"
This specifies that the grammar file depends on features implemented
in Bison version VERSION or higher.
* lalr1.cc: The token and value types are now class members.
The tokens were defined as free form enums and cpp macros. YYSTYPE
was defined as a free form union. They are now class members:
tokens are enumerations of the `yy::parser::token' struct, and the
semantic values have the `yy::parser::semantic_type' type.
If you do not want or can update to this scheme, the directive
`%define "global_tokens_and_yystype" "1"' triggers the global
definition of tokens and YYSTYPE. This change is suitable both
for previous releases of Bison, and this one.
If you wish to update, then make sure older version of Bison will
fail using `%require "2.2"'.
* DJGPP support added.
Changes in version 2.1, 2005-09-16:
* Bison-generated parsers now support the translation of diagnostics like
"syntax error" into languages other than English. The default
language is still English. For details, please see the new
Internationalization section of the Bison manual. Software
distributors should also see the new PACKAGING file. Thanks to
Bruno Haible for this new feature.
* Wording in the Bison-generated parsers has been changed slightly to
simplify translation. In particular, the message "memory exhausted"
has replaced "parser stack overflow", as the old message was not
always accurate for modern Bison-generated parsers.
* Destructors are now called when the parser aborts, for all symbols left
behind on the stack. Also, the start symbol is now destroyed after a
successful parse. In both cases, the behavior was formerly inconsistent.
* When generating verbose diagnostics, Bison-generated parsers no longer
quote the literal strings associated with tokens. For example, for
a syntax error associated with '%token NUM "number"' they might
print 'syntax error, unexpected number' instead of 'syntax error,
unexpected "number"'.
run-time dependency (DEPENDS) on a tool is to append a ":run" modifier
to the tool name, e.g.,
USE_TOOLS+= perl:run
Tools without modifiers or with an explicit ":build" modifier will
cause build dependencies (BUILD_DEPENDS) on those tools to be added.
This makes the notation a bit more compact.
Changes in version 2.0, 2004-12-25:
* Possibly-incompatible changes
- Bison-generated parsers no longer default to using the alloca function
(when available) to extend the parser stack, due to widespread
problems in unchecked stack-overflow detection. You can "#define
YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
the manual to determine safe values for YYMAXDEPTH in that case.
- Error token location.
During error recovery, the location of the syntax error is updated
to cover the whole sequence covered by the error token: it includes
the shifted symbols thrown away during the first part of the error
recovery, and the lookahead rejected during the second part.
- Semicolon changes:
. Stray semicolons are no longer allowed at the start of a grammar.
. Semicolons are now required after in-grammar declarations.
- Unescaped newlines are no longer allowed in character constants or
string literals. They were never portable, and GCC 3.4.0 has
dropped support for them. Better diagnostics are now generated if
forget a closing quote.
- NUL bytes are no longer allowed in Bison string literals, unfortunately.
* New features
- GLR grammars now support locations.
- New directive: %initial-action.
This directive allows the user to run arbitrary code (including
initializing @$) from yyparse before parsing starts.
- A new directive "%expect-rr N" specifies the expected number of
reduce/reduce conflicts in GLR parsers.
- %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
This is a GNU extension.
- The option `--report=lookahead' was changed to `--report=look-ahead'.
The old spelling still works, but is not documented and will be
removed.
- Experimental %destructor support has been added to lalr1.cc.
- New configure option --disable-yacc, to disable installation of the
yacc command and -ly library introduced in 1.875 for POSIX conformance.
* Bug fixes
- For now, %expect-count violations are now just warnings, not errors.
This is for compatibility with Bison 1.75 and earlier (when there are
reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
are too many or too few shift/reduce conflicts). However, in future
versions of Bison we plan to improve the %expect machinery so that
these violations will become errors again.
- Within Bison itself, numbers (e.g., goto numbers) are no longer
arbitrarily limited to 16-bit counts.
- Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
quotearg_buffer_restyled(). Google suggests this function has issues
with other non-gcc compilers using optimisation as well.
Notable fixes include net/libIDL's parser.y
