-*- outline -*-
+* Short term
+** Use syntax_error from the scanner?
+This would provide a means to raise syntax error from function called
+from the scanner. Actually, there is no good solution to report a
+lexical error in general. Usually they are kept at the scanner level
+only, ignoring the guilty token. But that might not be the best bet,
+since we don't benefit from the syntactic error recovery.
+
+We still have the possibility to return an invalid token number, which
+does the trick. But then the error message from the parser is poor
+(something like "unexpected $undefined"). Since the scanner probably
+already reported the error, we should directly enter error-recovery,
+without reporting the error message (i.e., YYERROR's semantics).
+
+Back to lalr1.cc (whose name is now quite unfortunate, since it also
+covers lr and ielr), if we support exceptions from yylex, should we
+propose a lexical_error in addition to syntax_error? Should they have
+a common root, say parse_error? Should syntax_error be renamed
+syntactic_error for consistency with lexical_error?
+
+** Variable names.
+What should we name `variant' and `lex_symbol'?
+
+** Use b4_symbol in all the skeleton
+Then remove the older system, including the tables generated by
+output.c
+
+** Update the documentation on gnu.org
+
+** Get rid of fake #lines [Bison: ...]
+Possibly as simple as checking whether the column number is nonnegative.
+
+I have seen messages like the following from GCC.
+
+<built-in>:0: fatal error: opening dependency file .deps/libltdl/argz.Tpo: No such file or directory
+
+
+** Discuss about %printer/%destroy in the case of C++.
+It would be very nice to provide the symbol classes with an operator<<
+and a destructor. Unfortunately the syntax we have chosen for
+%destroy and %printer make them hard to reuse. For instance, the user
+is invited to write something like
+
+ %printer { debug_stream() << $$; } <my_type>;
+
+which is hard to reuse elsewhere since it wants to use
+"debug_stream()" to find the stream to use. The same applies to
+%destroy: we told the user she could use the members of the Parser
+class in the printers/destructors, which is not good for an operator<<
+since it is no longer bound to a particular parser, it's just a
+(standalone symbol).
+
+** Rename LR0.cc
+as lr0.cc, why upper case?
+
+** bench several bisons.
+Enhance bench.pl with %b to run different bisons.
+
+** Use b4_symbol everywhere.
+Move its definition in the more standard places and deploy it in other
+skeletons.
+
+* Various
+** YYPRINT
+glr.c inherits its symbol_print function from c.m4, which supports
+YYPRINT. But to use YYPRINT yytoknum is needed, which not defined by
+glr.c.
+
+Anyway, IMHO YYPRINT is obsolete and should be restricted to yacc.c.
+
+** YYERRCODE
+Defined to 256, but not used, not documented. Probably the token
+number for the error token, which POSIX wants to be 256, but which
+Bison might renumber if the user used number 256. Keep fix and doc?
+Throw away?
+
+Also, why don't we output the token name of the error token in the
+output? It is explicitly skipped:
+
+ /* Skip error token and tokens without identifier. */
+ if (sym != errtoken && id)
+
+Of course there are issues with name spaces, but if we disable we have
+something which seems to be more simpler and more consistent instead
+of the special case YYERRCODE.
+
+ enum yytokentype {
+ error = 256,
+ // ...
+ };
+
+
+We could (should?) also treat the case of the undef_token, which is
+numbered 257 for yylex, and 2 internal. Both appear for instance in
+toknum:
+
+ const unsigned short int
+ parser::yytoken_number_[] =
+ {
+ 0, 256, 257, 258, 259, 260, 261, 262, 263, 264,
+
+while here
+
+ enum yytokentype {
+ TOK_EOF = 0,
+ TOK_EQ = 258,
+
+so both 256 and 257 are "mysterious".
+
+ const char*
+ const parser::yytname_[] =
+ {
+ "\"end of command\"", "error", "$undefined", "\"=\"", "\"break\"",
+
+
+** YYFAIL
+It is seems to be *really* obsolete now, shall we remove it?
+
+** YYBACKUP
+There is no test about it, no examples in the doc, and I'm not sure
+what it should look like. For instance what follows crashes.
+
+ %error-verbose
+ %debug
+ %pure-parser
+ %code {
+ # include <stdio.h>
+ # include <stdlib.h>
+ # include <assert.h>
+
+ static void yyerror (const char *msg);
+ static int yylex (YYSTYPE *yylval);
+ }
+ %%
+ exp:
+ 'a' { printf ("a: %d\n", $1); }
+ | 'b' { YYBACKUP('a', 123); }
+ ;
+ %%
+ static int
+ yylex (YYSTYPE *yylval)
+ {
+ static char const input[] = "b";
+ static size_t toknum;
+ assert (toknum < sizeof input);
+ *yylval = (toknum + 1) * 10;
+ return input[toknum++];
+ }
+
+ static void
+ yyerror (const char *msg)
+ {
+ fprintf (stderr, "%s\n", msg);
+ }
+
+ int
+ main (void)
+ {
+ yydebug = !!getenv("YYDEBUG");
+ return yyparse ();
+ }
+
+** yychar == yyempty_
+The code in yyerrlab reads:
+
+ if (yychar <= YYEOF)
+ {
+ /* Return failure if at end of input. */
+ if (yychar == YYEOF)
+ YYABORT;
+ }
+
+There are only two yychar that can be <= YYEOF: YYEMPTY and YYEOF.
+But I can't produce the situation where yychar is YYEMPTY here, is it
+really possible? The test suite does not exercise this case.
+
+This shows that it would be interesting to manage to install skeleton
+coverage analysis to the test suite.
+
+** Table definitions
+It should be very easy to factor the definition of the various tables,
+including the separation bw declaration and definition. See for
+instance b4_table_define in lalr1.cc. This way, we could even factor
+C vs. C++ definitions.
+
+* From lalr1.cc to yacc.c
+** Single stack
+Merging the three stacks in lalr1.cc simplified the code, prompted for
+other improvements and also made it faster (probably because memory
+management is performed once instead of three times). I suggest that
+we do the same in yacc.c.
+
+** yysyntax_error
+The code bw glr.c and yacc.c is really alike, we can certainly factor
+some parts.
+
* Header guards
From Franc,ois: should we keep the directory part in the CPP guard?
Before releasing, make sure the documentation ("Understanding your
parser") refers to the current `output' format.
-* lalr1.cc
-** I18n
-Catch up with yacc.c.
-
* Report
+** Figures
+Some statistics about the grammar and the parser would be useful,
+especially when asking the user to send some information about the
+grammars she is working on. We should probably also include some
+information about the variables (I'm not sure for instance we even
+specify what LR variant was used).
+
** GLR
How would Paul like to display the conflicted actions? In particular,
what when two reductions are possible on a given lookahead token, but one is
We should find a means to provide an access to values deep in the
stack. For instance, instead of
- baz: qux { $$ = $<foo>-1 + $<bar>0 + $1; }
+ baz: qux { $$ = $<foo>-1 + $<bar>0 + $1; }
we should be able to have:
part of %if. Akim Demaille thinks it should be in the parser, so as
to avoid falling into another CPP mistake.
-** -D, --define-muscle NAME=VALUE
-To define muscles via cli. Or maybe support directly NAME=VALUE?
-
** XML Output
There are couple of available extensions of Bison targeting some XML
output. Some day we should consider including them. One issue is
* Unit rules
Maybe we could expand unit rules, i.e., transform
- exp: arith | bool;
- arith: exp '+' exp;
- bool: exp '&' exp;
+ exp: arith | bool;
+ arith: exp '+' exp;
+ bool: exp '&' exp;
into
- exp: exp '+' exp | exp '&' exp;
+ exp: exp '+' exp | exp '&' exp;
when there are no actions. This can significantly speed up some
grammars. I can't find the papers. In particular the book `LR
Some history of Bison and some bibliography would be most welcome.
Are there any Texinfo standards for bibliography?
-
+** %printer
+Wow, %printer is not documented. Clearly mark YYPRINT as obsolete.
* Java, Fortran, etc.
* Coding system independence
Paul notes:
- Currently Bison assumes 8-bit bytes (i.e. that UCHAR_MAX is
- 255). It also assumes that the 8-bit character encoding is
- the same for the invocation of 'bison' as it is for the
- invocation of 'cc', but this is not necessarily true when
- people run bison on an ASCII host and then use cc on an EBCDIC
- host. I don't think these topics are worth our time
- addressing (unless we find a gung-ho volunteer for EBCDIC or
- PDP-10 ports :-) but they should probably be documented
- somewhere.
+ Currently Bison assumes 8-bit bytes (i.e. that UCHAR_MAX is
+ 255). It also assumes that the 8-bit character encoding is
+ the same for the invocation of 'bison' as it is for the
+ invocation of 'cc', but this is not necessarily true when
+ people run bison on an ASCII host and then use cc on an EBCDIC
+ host. I don't think these topics are worth our time
+ addressing (unless we find a gung-ho volunteer for EBCDIC or
+ PDP-10 ports :-) but they should probably be documented
+ somewhere.
- More importantly, Bison does not currently allow NUL bytes in
- tokens, either via escapes (e.g., "x\0y") or via a NUL byte in
- the source code. This should get fixed.
+ More importantly, Bison does not currently allow NUL bytes in
+ tokens, either via escapes (e.g., "x\0y") or via a NUL byte in
+ the source code. This should get fixed.
* --graph
Show reductions.
makes it impossible to have modular precedence information. We should
move to partial orders (sounds like series/parallel orders to me).
-** Correlation b/w precedence and associativity
-Also, I fail to understand why we have to assign the same
-associativity to operators with the same precedence. For instance,
-why can't I decide that the precedence of * and / is the same, but the
-latter is nonassoc?
-
-If there is really no profound motivation, we should find a new syntax
-to allow specifying this.
-
** RR conflicts
See if we can use precedence between rules to solve RR conflicts. See
what POSIX says.
* Better graphics
Equip the parser with a means to create the (visual) parse tree.
+* Complaint submessage indentation.
+We already have an implementation that works fairly well for named
+reference messages, but it would be nice to use it consistently for all
+submessages from Bison. For example, the "previous definition"
+submessage or the list of correct values for a %define variable might
+look better with indentation.
+
+However, the current implementation makes the assumption that the
+location printed on the first line is not usually much shorter than the
+locations printed on the submessage lines that follow. That assumption
+may not hold true as often for some kinds of submessages especially if
+we ever support multiple grammar files.
+
+Here's a proposal for how a new implementation might look:
+
+ http://lists.gnu.org/archive/html/bison-patches/2009-09/msg00086.html
+
-----
-Copyright (C) 2001, 2002, 2003, 2004, 2006, 2008 Free Software Foundation,
-Inc.
+Copyright (C) 2001-2004, 2006, 2008-2012 Free Software Foundation, Inc.
This file is part of Bison, the GNU Compiler Compiler.