Undo my 2007-02-07 change, switching back to the c-strcase module

[bison.git] / ChangeLog

diff --git a/ChangeLog b/ChangeLog
index 3f9eda0be56b2a3a6544821b1c69a946f7a37f82..dde006f1cd3875c2c5321792b2eb870495ab69c6 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,76 @@
+2007-02-11  Paul Eggert  <eggert@cs.ucla.edu>
+
+       Undo my 2007-02-07 change, switching back to the c-strcase module
+       introduced in the 2007-02-03 change.  Bruno Haible reported that
+       the 2007-02-07 change would be dangerous in Turkish if we add a
+       language whose name contains "i", since "i" is not lowercase "I"
+       in Turkish.
+       * bootstrap.conf (gnulib_modules): Add c-strcase.  Remove strcase.
+       * lib/.cvsignore: Add c-ctype.c, c-ctype.h, c-strcase.h,
+       c-strcasecomp.c, c-strncasecmp.c.  Remove strcasecmp.c, strncasecmp.c.
+       * m4/.cvsignore: Remove strcase.m4.
+       * src/getargs.c: Revert 2007-02-07 change, as follows.
+       Include c-strcase.h.
+       (language_argmatch): Use c_strcasecmp rather than strcasecmp.
+
+2007-02-11  Bruno Haible  <bruno@clisp.org>
+
+       Enable the Java related testsuite tests when the only Java compiler
+       found is a gcj < 4.3.  Discussed at
+       <http://lists.gnu.org/archive/html/bug-bison/2007-02/msg00016.html>.
+       * configure.ac (gt_JAVACOMP): Don't specify a target_version.
+
+2007-02-11  Joel E. Denny  <jdenny@ces.clemson.edu>
+
+       * data/Makefile.am: Update copyright date.
+       * data/push.c (yypull_parse): Report memory exhaustion and return 2 if
+       yypstate_new returns NULL.
+       (yypstate_new): Return NULL if malloc does.
+       * src/reader.c (packgram): Move translation of rule actions from the
+       beginning of packgram to...
+       (check_and_convert_grammar): ... here right before packgram is invoked
+       so it's easier to write more complete comments, and remove redundant
+       code.
+
+2007-02-10  Joel E. Denny  <jdenny@ces.clemson.edu>
+
+       As in semantic actions, make @$ in %initial-action, %destructor, and
+       %printer imply %locations.
+       * src/scan-code.l (SC_SYMBOL_ACTION): Set locations_flag = true when
+       scanning @$.
+       * tests/actions.at (AT_CHECK_ACTION_LOCATIONS): New macro supporting...
+       (@$ in %initial-action implies %locations,
+       @$ in %destructor implies %locations,
+       @$ in %printer implies %locations): ... these new test cases.
+
+2007-02-07  Paul Eggert  <eggert@cs.ucla.edu>
+
+       Undo most of the 2007-02-03 change, switching to the strcase module
+       now that gnulib strcase has been fixed.
+       * bootstrap.conf (gnulib_modules): Remove c-strcase.  Add strcase.
+       * lib/.cvsignore: Remove c-ctype.c, c-ctype.h, c-strcase.h,
+       c-strcasecomp.c, c-strncasecmp.c.  Add strcasecmp.c, strncasecmp.c
+       * m4/.cvsignore: Add strcase.m4.
+       * src/getargs.c: Revert 2007-02-03 change, as follows.
+       Don't include c-strcase.h.
+       (language_argmatch): Use strcasecmp rather than c_strcasecmp.
+       strcasecmp has "unspecified behavior" outside the POSIX locale,
+       but it works fine in practice if at least one argument is ASCII,
+       as is the case in Bison.
+
+2007-02-07  Paolo Bonzini  <bonzini@gnu.org>
+
+       * tests/java.at: Skip tests if only one of javac/java is present.
+       Reported by Joel E. Denny.
+       * tests/atlocal.in: Adjust copyright years.
+
+2007-02-05  Paolo Bonzini  <bonzini@gnu.org>
+
+       * data/lalr1.java (Stack): Work around old verifiers that disallow
+       access to the private fields of an inner class, from the outer class.
+       We can make Stack's fields public because user code doesn't have access
+       to the instance of Stack used by parse().  Reported by Paul Eggert.
+
 2007-02-03  Paul Eggert  <eggert@cs.ucla.edu>
 
        * .cvsignore: Add javacomp.sh, javaexec.sh.  Is this really