ecos: host/libcdl/expr.cxx comparison

comparison host/libcdl/expr.cxx @ 76:435cced73e2f ecos-v1_3_1-release

eCos v1.3.1 merged from eCos master repository on 2000-03-27-23:22:51-BST

author	jlarmour
date	Tue, 28 Mar 2000 14:10:45 +0000
parents
children	6736c52df507

comparison

equal deleted inserted replaced

-:41bf073c0c32
+:435cced73e2f
+//{{{  Banner
+//============================================================================
+//
+//      expr.cxx
+//
+//      Implementation of the various CDL expression classes.
+//
+//============================================================================
+//####COPYRIGHTBEGIN####
+//
+// ----------------------------------------------------------------------------
+// Copyright (C) 1999, 2000 Red Hat, Inc.
+//
+// This file is part of the eCos host tools.
+//
+// This program is free software; you can redistribute it and/or modify it
+// under the terms of the GNU General Public License as published by the Free
+// Software Foundation; either version 2 of the License, or (at your option)
+// any later version.
+//
+// This program is distributed in the hope that it will be useful, but WITHOUT
+// ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+// FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+// more details.
+//
+// You should have received a copy of the GNU General Public License along with
+// this program; if not, write to the Free Software Foundation, Inc.,
+// 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
+//
+// ----------------------------------------------------------------------------
+//
+//####COPYRIGHTEND####
+//============================================================================
+//#####DESCRIPTIONBEGIN####
+//
+// Author(s):   bartv
+// Contact(s):  bartv
+// Date:        1999/02/02
+// Version:     0.02
+//
+//####DESCRIPTIONEND####
+//============================================================================
+//}}}
+//{{{  #include's
+// ----------------------------------------------------------------------------
+#include "cdlconfig.h"
+// Get the infrastructure types, assertions, tracing and similar
+// facilities.
+#include <cyg/infra/cyg_ass.h>
+#include <cyg/infra/cyg_trac.h>
+// <cdlcore.hxx> defines everything implemented in this module.
+// It implicitly supplies <string>, <vector> and <map> because
+// the class definitions rely on these headers.
+#include <cdlcore.hxx>
+//}}}
+//{{{  Statics
+// ----------------------------------------------------------------------------
+CYGDBG_DEFINE_MEMLEAK_COUNTER(CdlEvalContext);
+CYGDBG_DEFINE_MEMLEAK_COUNTER(CdlExpressionBody);
+CYGDBG_DEFINE_MEMLEAK_COUNTER(CdlListExpressionBody);
+CYGDBG_DEFINE_MEMLEAK_COUNTER(CdlGoalExpressionBody);
+//}}}
+//{{{  CdlEvalContext
+// ----------------------------------------------------------------------------
+// A utility class to keep track of the context in which expression
+// evaluation is happening.
+CdlEvalContext::CdlEvalContext(CdlTransaction transaction_arg, CdlNode node_arg, CdlProperty property_arg,
+CdlToplevel toplevel_arg)
+{
+CYG_REPORT_FUNCNAME("CdlEvalContext::constructor");
+CYG_REPORT_FUNCARG4XV(this, transaction_arg, node_arg, property_arg);
+transaction = transaction_arg;
+node        = node_arg;
+property    = property_arg;
+if (0 == toplevel_arg) {
+if (0 != transaction) {
+toplevel = transaction->get_toplevel();
+} else if (0 != node) {
+toplevel = node->get_toplevel();
+} else {
+toplevel = toplevel_arg;
+}
+} else {
+toplevel = toplevel_arg;
+}
+cdlevalcontext_cookie = CdlEvalContext_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+CdlEvalContext::~CdlEvalContext()
+{
+CYG_REPORT_FUNCNAME("CdlEvalContext::destructor");
+CYG_PRECONDITION_THISC();
+cdlevalcontext_cookie       = CdlEvalContext_Invalid;
+transaction = 0;
+node        = 0;
+property    = 0;
+toplevel    = 0;
+CYGDBG_MEMLEAK_DESTRUCTOR();
+CYG_REPORT_RETURN();
+}
+bool
+CdlEvalContext::check_this(cyg_assert_class_zeal zeal) const
+{
+if (CdlEvalContext_Magic != cdlevalcontext_cookie) {
+return false;
+}
+CYGDBG_MEMLEAK_CHECKTHIS();
+if ((0 != transaction) && !transaction->check_this(zeal)) {
+return false;
+}
+if ((0 != toplevel) && !toplevel->check_this(zeal)) {
+return false;
+}
+if ((0 != node) && !node->check_this(zeal)) {
+return false;
+}
+if ((0 != property) && !property->check_this(zeal)) {
+return false;
+}
+return true;
+}
+//}}}
+//{{{  Expression parsing
+//{{{  Description
+// ----------------------------------------------------------------------------
+// There are a number of different entry points related to expression parsing,
+// largely to support list and goal expressions. All of these eventually
+// end up calling the function
+//    continue_parse(expr, data, index, token, token_end)
+//
+// The expr argument holds an existing expression object that needs to be
+// updated. If token is Invalid then we are at the start of an expression
+// (but not necessarily at the start of the string).
+//
+// The data string holds all of the expression that should be parsed.
+// It is formed by concatenating all non-option arguments to the
+// appropriate property command, with spaces between them.
+//
+// index is an input/output variable. On input it indicates where in
+// the string parsing should continue. On output it indicates the
+// location within the string where the terminating token began.
+//
+// token is an input/output variable. On input it can have the values
+// Invalid or And. The former means that we are parsing a completely
+// new expression. The latter is used for goal expressions: it is
+// necessary to parse a new expression and then combine it with the
+// existing one.
+//
+// token_end is an output variable. It indicates the location within
+// the string where the terminating token ended. This is useful for
+// e.g. ranges in a list expression.
+//
+// A conventional recursive descent parser is used.
+//}}}
+//{{{  Tokenization
+// ----------------------------------------------------------------------------
+// Tokenization.
+//{{{  token enum
+// A separate token enum is necessary, rather than re-using the CdlExprOp
+// enum. Some tokens may correspond to several operators, and some tokens
+// such as close-bracket do not correspond directly to an operator at all.
+enum token {
+T_Invalid           = -2,
+T_EOD               = -1,
+T_Reference         =  1,   // CYGPKG_HAL
+T_String            =  2,   // "hello"
+T_Integer           =  3,   // 123
+T_Double            =  4,   // 3.1415
+T_Range             =  5,   // to
+T_OpenBracket       =  6,   // (
+T_CloseBracket      =  7,   // )
+T_Minus             =  8,   // -
+T_Plus              =  9,   // +
+T_Times             = 10,   // *
+T_Divide            = 11,   // /
+T_Exclamation       = 12,   // !
+T_Tilde             = 13,   // ~
+T_Questionmark      = 14,   // ?
+T_Remainder         = 15,   // %
+T_LeftShift         = 16,   // <<
+T_RightShift        = 17,   // >>
+T_LessThan          = 18,   // <
+T_LessEqual         = 19,   // <=
+T_GreaterThan       = 20,   // >
+T_GreaterEqual      = 21,   // >=
+T_Equal             = 22,   // ==
+T_NotEqual          = 23,   // !=
+T_BitAnd            = 24,   // &
+T_BitXor            = 25,   // ^
+T_BitOr             = 26,   // |
+T_And               = 27,   // &&
+T_Or                = 28,   // ||
+T_Colon             = 29    // : (in a conditional)
+};
+//}}}
+//{{{  Statics
+// Statics to keep track of the current state.
+static std::string      current_data            = "";
+static unsigned int     current_index           = 0;
+static unsigned int     token_start             = 0;
+static int              current_char            = EOF;
+static token            current_token           = T_Invalid;
+static std::string      current_string          = "";
+static std::string      current_reference       = "";
+static cdl_int          current_int             = 0;
+static double           current_double          = 0.0;
+static CdlValueFormat   current_format          = CdlValueFormat_Default;
+//}}}
+//{{{  Character access
+// ----------------------------------------------------------------------------
+// Individual character access.
+// Note that current_index is one character past current_char.
+// Return the next character in the string, or EOF
+static void
+next_char()
+{
+if (current_index >= current_data.size()) {
+current_char = EOF;
+} else {
+current_char = current_data[current_index++];
+}
+}
+// Go back a character. This is useful when parsing
+// strings. It is the responsibility of the calling code
+// to make sure that we are not at the start of the buffer.
+static void
+backup_char()
+{
+CYG_ASSERTC(((EOF == current_char) && (0 < current_index)) || (1 < current_index));
+if (EOF != current_char) {
+current_index--;
+}
+current_char = current_data[current_index - 1];
+}
+//}}}
+//{{{  get_error_location()
+// ----------------------------------------------------------------------------
+// Construct part of a diagnostic message, indicating the
+// area in the data where the error occurred. This string
+// is of the form {...data} ^char^ {data...}. Ideally
+// the ^ markers would be on a subsequent line, eliminating
+// the need for braces, but there is insufficient control
+// of how the message gets presented to the user.
+//
+// Care has to be taken with EOD.
+static std::string
+get_error_location()
+{
+CYG_REPORT_FUNCNAME("get_error_location");
+std::string result = "";
+// token_start is probably the best place for centering the error.
+// current_index is past the point where the error has occurred.
+if (token_start > 1) {
+if (token_start > 16) {
+result = "{..." + current_data.substr(token_start - 13, 13) + "} ";
+} else {
+result = "{" + current_data.substr(0, token_start) + "}";
+}
+}
+if (current_char == EOF) {
+result += " <end of data>";
+} else {
+result += " ^" + std::string(1, current_data[token_start]) + "^ ";
+}
+if (token_start < current_data.size()) {
+if ((token_start + 16) < current_data.size()) {
+result += "{" + current_data.substr(token_start + 1, current_data.size() - (token_start+1)) + "}";
+} else {
+result += "{" + current_data.substr(token_start, 13) + "...}";
+}
+}
+CYG_REPORT_RETURN();
+return result;
+}
+//}}}
+//{{{  Token translation
+// ----------------------------------------------------------------------------
+// Convert a token into a binary expression operator
+static CdlExprOp
+token_to_binary_expr_op()
+{
+CYG_REPORT_FUNCNAMETYPE("token_to_expr_op", "op %d");
+CdlExprOp result = CdlExprOp_Invalid;
+switch(current_token) {
+case T_Minus:             result = CdlExprOp_Subtract; break;
+case T_Plus:              result = CdlExprOp_Add; break;
+case T_Times:             result = CdlExprOp_Multiply; break;
+case T_Divide:            result = CdlExprOp_Divide; break;
+case T_Remainder:         result = CdlExprOp_Remainder; break;
+case T_LeftShift:         result = CdlExprOp_LeftShift; break;
+case T_RightShift:        result = CdlExprOp_RightShift; break;
+case T_LessThan:          result = CdlExprOp_LessThan; break;
+case T_LessEqual:         result = CdlExprOp_LessEqual; break;
+case T_GreaterThan:       result = CdlExprOp_GreaterThan; break;
+case T_GreaterEqual:      result = CdlExprOp_GreaterEqual; break;
+case T_Equal:             result = CdlExprOp_Equal; break;
+case T_NotEqual:          result = CdlExprOp_NotEqual; break;
+case T_BitAnd:            result = CdlExprOp_BitAnd; break;
+case T_BitXor:            result = CdlExprOp_BitXor; break;
+case T_BitOr:             result = CdlExprOp_BitOr; break;
+case T_And:               result = CdlExprOp_And; break;
+case T_Or:                result = CdlExprOp_Or; break;
+default: result = CdlExprOp_Invalid; break;
+}
+CYG_REPORT_RETVAL(result);
+return result;
+}
+// Convert a token into an ExprOp. This way the internal token enum does
+// not need to be exported in order to define the interface.
+//
+// In practice the higher level code will only look for a handful of
+// cases, mainly EOD and the range operator, but we might as well
+// do the job property.
+static CdlExprOp
+token_to_expr_op()
+{
+CYG_REPORT_FUNCNAMETYPE("token_to_expr_op", "expr op %d");
+CdlExprOp result;
+// Many of the tokens are already handled for binary operators.
+result = token_to_binary_expr_op();
+if (CdlExprOp_Invalid == result) {
+switch(current_token) {
+case T_EOD:             result = CdlExprOp_EOD; break;
+case T_Reference:       result = CdlExprOp_Reference; break;
+case T_String:          result = CdlExprOp_StringConstant; break;
+case T_Integer:         result = CdlExprOp_IntegerConstant; break;
+case T_Double:          result = CdlExprOp_DoubleConstant; break;
+case T_Range:           result = CdlExprOp_Range; break;
+case T_Exclamation:     result = CdlExprOp_LogicalNot; break;
+case T_Tilde:           result = CdlExprOp_BitNot; break;
+case T_Questionmark:
+case T_Colon:           result = CdlExprOp_Cond; break; // best guess
+case T_OpenBracket:
+case T_CloseBracket:
+case T_Invalid:
+default:                result = CdlExprOp_Invalid; break;
+}
+}
+CYG_REPORT_RETVAL(result);
+return result;
+}
+// A utility routine to turn the current token back into a string
+// This is used for diagnostics.
+static std::string
+token_to_string()
+{
+CYG_REPORT_FUNCNAME("token_to_string");
+std::string result = "";
+switch(current_token) {
+case T_EOD:               result = "<end of data>"; break;
+case T_Reference:         result = "reference to " + current_reference; break;
+case T_String:            result = "string \"" + current_string + "\""; break;
+case T_Integer:
+{
+std::string tmp;
+Cdl::integer_to_string(current_int, tmp, current_format);
+result = "integer constant " + tmp;
+break;
+}
+case T_Double:
+{
+std::string tmp;
+Cdl::double_to_string(current_double, tmp, current_format);
+result = "double constant " + tmp;
+break;
+}
+case T_Range:             result = "range operator \"to\""; break;
+case T_OpenBracket:       result = "open bracket ("; break;
+case T_CloseBracket:      result = "close bracket )"; break;
+case T_Minus:             result = "minus sign -"; break;
+case T_Plus:              result = "plus sign +"; break;
+case T_Times:             result = "multiply operator *"; break;
+case T_Divide:            result = "divide operator /"; break;
+case T_Exclamation:       result = "not operator !"; break;
+case T_Tilde:             result = "bitwise not operator ~"; break;
+case T_Questionmark:      result = "question mark ?"; break;
+case T_Remainder:         result = "remainder operator %"; break;
+case T_LeftShift:         result = "left shift operator <<"; break;
+case T_RightShift:        result = "right shift operator >>"; break;
+case T_LessThan:          result = "less-than operator <"; break;
+case T_LessEqual:         result = "less-or-equal operator <="; break;
+case T_GreaterThan:       result = "greater-than operator >"; break;
+case T_GreaterEqual:      result = "greater-or-equal operator >="; break;
+case T_Equal:             result = "equality operator =="; break;
+case T_NotEqual:          result = "not-equal operator !="; break;
+case T_BitAnd:            result = "bitwise and operator &"; break;
+case T_BitXor:            result = "bitwise xor operator ^"; break;
+case T_BitOr:             result = "bitwise or operator |"; break;
+case T_And:               result = "and operator &&"; break;
+case T_Or:                result = "or operator ||"; break;
+case T_Colon:             result = "colon"; break;
+case T_Invalid:
+default:                  result = "<invalid token>"; break;
+}
+CYG_REPORT_RETURN();
+return result;
+}
+//}}}
+//{{{  Literals
+// ----------------------------------------------------------------------------
+//{{{  process_string()
+// The start of a string has been detected. Work out the entire string,
+// allowing for backslash escapes.
+static void
+process_string()
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("process_string");
+CYG_ASSERTC('"' == current_char);
+CYG_ASSERTC("" == current_string);
+std::string result = "";
+// Move past the leading quote mark.
+next_char();
+while ('"' != current_char) {
+if (EOF == current_char) {
+throw CdlParseException("Premature end of data in string constant.\n" + get_error_location());
+} else if ('\\' == current_char) {
+// Allow \a, \b, \f, \n, \r, \t, \v, \ddd and \xhh.
+// Also copy with \newline space.
+// Any other character gets passed through unchanged.
+next_char();
+switch(current_char) {
+case EOF:
+throw CdlParseException("Premature end of data after backslash in string constant.\n" + get_error_location());
+case 'a':
+result += '\a';
+break;
+case 'b':
+result += '\b';
+break;
+case 'f':
+result += '\f';
+break;
+case 'n':
+result += '\n';
+break;
+case 'r':
+result += '\r';
+break;
+case 't':
+result += '\t';
+break;
+case 'v':
+result += '\v';
+break;
+case 'x':
+{
+cdl_int tmp = 0;
+next_char();
+if (!isxdigit(current_char)) {
+throw CdlParseException("Non-hexadecimal digit detected in string \\x escape sequence.\n" +
+get_error_location());
+}
+// NOTE: there is no overflow detection here.
+do {
+tmp *= 16;
+if (('0' <= current_char) && (current_char <= '9')) {
+tmp += (current_char - '0');
+} else if (('a' <= current_char) && (current_char <= 'f')) {
+tmp += 10 + (current_char - 'a');
+} else if (('A' <= current_char) && (current_char <= 'F')) {
+tmp += 10 + (current_char - 'A');
+} else {
+CYG_FAIL("C library error, isxdigit() succeeded on non-hexadecimal character");
+}
+next_char();
+} while(isxdigit(current_char));
+backup_char();
+result += (char) tmp;
+}
+case '\n':
+next_char();
+while ((EOF != current_char) && isspace(current_char)) {
+next_char();
+}
+// We have gone one too far, back up.
+backup_char();
+result += " ";
+break;
+default:
+if (('0' <= current_char) && (current_char <= '7')) {
+// A sequence of octal digits.
+cdl_int tmp = 0;
+do {
+tmp = (8 * tmp) + (current_char - '0');
+next_char();
+} while (('0' <= current_char) && (current_char <= '7'));
+backup_char();
+result += (char) tmp;
+} else {
+// For all other backslash sequences, just add the second character
+result += (char) current_char;
+}
+}
+} else {
+result += (char) current_char;
+}
+next_char();
+}
+// The closing quote has been reached, move past it.
+next_char();
+// And all done.
+current_token  = T_String;
+current_string = result;
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  process_number()
+// The start of a number has been detected. This number may be an
+// integer or a double. It is necessary to figure out where the number
+// ends and invoke the appropriate Cdl:: conversion utility.
+//
+// Care has to be taken with termination. Consider a token such as
+// 134_5. This is not a string because there are no quote marks, nor
+// is it a valid reference, and because it begins with a digit it
+// should be interpreted as a number. The 134 bit works fine, then
+// number processing stops leaving current_char as '_'. If we are
+// parsing a list expression then the following _5 will actually
+// be interpreted as a reference. To avoid this, here is a utility
+// which checks number completion and throws an exception if
+// necessary.
+static void check_number_termination()
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("check_number_termination");
+// End-of-data or any whitespace is ok.
+if ((EOF != current_char) && !isspace(current_char)) {
+// Any valid operator is ok as well, or brackets for that matter.
+if (('-' != current_char) && ('+' != current_char) && ('*' != current_char) &&
+('/' != current_char) && ('!' != current_char) && ('~' != current_char) &&
+('?' != current_char) && ('%' != current_char) && ('<' != current_char) &&
+('>' != current_char) && ('=' != current_char) && ('&' != current_char) &&
+('^' != current_char) && ('|' != current_char) && (':' != current_char) &&
+('(' != current_char) && (')' != current_char)) {
+std::string tmp;
+Cdl::integer_to_string(current_int, tmp);
+throw CdlParseException("Invalid character detected after number " + tmp + "\n" + get_error_location());
+}
+}
+CYG_REPORT_RETURN();
+}
+static void
+process_number()
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("process_number");
+std::string tmp      = "";
+bool        is_float = false;
+// Detect the special cases of 0x and octal numbers.
+if ('0' == current_char) {
+next_char();
+if (('x' == current_char) || ('X' == current_char)) {
+next_char();
+if (!isxdigit(current_char)) {
+throw CdlParseException("Invalid hexadecimal number, expected at least one hexadecimal digit after 0x.\n"
++ get_error_location());
+}
+current_int = 0;
+do {
+current_int *= 16;
+if (('0' <= current_char) && (current_char <= '9')) {
+current_int += (current_char - '0');
+} else if (('a' <= current_char) && (current_char <= 'f')) {
+current_int += 10 + (current_char - 'a');
+} else {
+current_int += 10 + (current_char - 'A');
+}
+next_char();
+} while(isxdigit(current_char));
+current_token  = T_Integer;
+current_format = CdlValueFormat_Hex;
+check_number_termination();
+CYG_REPORT_RETURN();
+return;
+} else if (('0' <= current_char) && (current_char <= '7')) {
+current_int = 0;
+do {
+current_int *= 8;
+current_int += (current_char - '0');
+next_char();
+} while (('0' <= current_char) && (current_char <= '7'));
+current_token  = T_Integer;
+current_format = CdlValueFormat_Octal;
+check_number_termination();
+CYG_REPORT_RETURN();
+return;
+} else if (('8' == current_char) || ('9' == current_char)) {
+throw CdlParseException("08... and 09... are not valid  octal numbers.\n" + get_error_location());
+} else {
+// This could be plain 0, or 0.123
+// Backup, and let the rest of the code take care of things
+backup_char();
+}
+}
+do {
+tmp += (char) current_char;
+next_char();
+} while(isdigit(current_char));
+// If we have found a . then we have a floating point number with a fraction.
+if ('.' == current_char) {
+tmp += '.';
+next_char();
+if (!isdigit(current_char)) {
+throw CdlParseException("Invalid floating point constant, expected a digit for the fractional part.\n" +
+get_error_location());
+}
+is_float = true;
+do {
+tmp += (char) current_char;
+next_char();
+} while(isdigit(current_char));
+}
+// If we have found e or E then we have a floating point number with an exponent
+if (('e' == current_char) || ('E' == current_char)) {
+tmp += 'E';
+next_char();
+if (('+' == current_char) || ('-' == current_char)) {
+tmp += current_char;
+next_char();
+}
+if (!isdigit(current_char)) {
+throw CdlParseException("Invalid floating point constant, expected a digit for the exponent.\n" +
+get_error_location());
+}
+is_float = true;
+do {
+tmp += (char) current_char;
+next_char();
+} while(isdigit(current_char));
+}
+if (is_float) {
+if (!Cdl::string_to_double(tmp, current_double)) {
+throw CdlParseException("Invalid floating point constant `" + tmp + "'.\n" + get_error_location());
+} else {
+current_token = T_Double;
+}
+} else {
+if (!Cdl::string_to_integer(tmp, current_int)) {
+throw CdlParseException("Invalid integer constant `" + tmp + "'.\n" + get_error_location());
+} else {
+current_token = T_Integer;
+}
+}
+check_number_termination();
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  process_reference()
+// The start of a reference has been detected.
+// A reference name must be a valid C preprocessor name, so the
+// only characters allowed are underscore, upper and lower case
+// characters, and digits. The first character cannot be a digit,
+// but that has been checked already.
+//
+// Some care has to be taken with locale's, the C library may decide
+// that a character is a letter even though the same character is not
+// valid as far as the preprocessor is concerned.
+static void
+process_reference()
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("process_reference");
+do {
+current_reference += (char) current_char;
+next_char();
+} while (('_' == current_char) || isdigit(current_char) ||
+(('a' <= current_char) && (current_char <= 'z')) ||
+(('A' <= current_char) && (current_char <= 'Z')));
+current_token = T_Reference;
+CYG_REPORT_RETURN();
+}
+//}}}
+//}}}
+//{{{  next_token()
+// ----------------------------------------------------------------------------
+// Work out what the next token is. This includes the handling of
+// strings, integers, doubles, and references.
+static void
+next_token()
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("next_token", "token %d");
+// Make sure there is no dross left lying around from the previous call.
+current_token       = T_Invalid;
+current_string      = "";
+current_reference   = "";
+current_int         = 0;
+current_double      = 0.0;
+current_format      = CdlValueFormat_Default;
+// Skip leading white space. This includes newlines, tabs, etc,
+// consider the case of:
+//    ...
+//    legal_values {
+//        1
+//        2
+//        4
+//        ..
+//    }
+//    ...
+// which is perfectly legitimate. White space inside strings
+// is handled by the string literal code, and does not get filtered
+// out here.
+//
+// Exactly which characters are white-space is implementation-defined,
+// so a special check for EOF is in order.
+while ((EOF != current_char) && isspace(current_char)) {
+next_char();
+}
+// Remember the token starting point. next_char() has actually moved
+// the index on by one.
+token_start = current_index - 1;
+// The simple cases can be handled inline, the more complicated cases
+// involve other functions
+switch(current_char) {
+case EOF:
+current_token = T_EOD;
+break;
+case '"':
+process_string();
+break;
+case '(':
+current_token = T_OpenBracket;
+next_char();
+break;
+case ')':
+current_token = T_CloseBracket;
+next_char();
+break;
+// At this level it is not possible to distinguish between
+// unary and binary operators, so no attempt is made to
+// turn - and + into part of a number.
+case '-':
+current_token = T_Minus;
+next_char();
+break;
+case '+':
+current_token = T_Plus;
+next_char();
+break;
+case '*':
+current_token = T_Times;
+next_char();
+break;
+case '/':
+current_token = T_Divide;
+next_char();
+break;
+case '!':
+next_char();
+if ('=' == current_char) {
+current_token = T_NotEqual;
+next_char();
+} else {
+current_token = T_Exclamation;
+}
+break;
+case '~':
+current_token = T_Tilde;
+next_char();
+break;
+case '?':
+current_token = T_Questionmark;
+next_char();
+break;
+case '%':
+current_token = T_Remainder;
+next_char();
+break;
+case '<':
+next_char();
+if ('<' == current_char) {
+current_token = T_LeftShift;
+next_char();
+} else if ('=' == current_char) {
+current_token = T_LessEqual;
+next_char();
+} else {
+current_token = T_LessThan;
+}
+break;
+case '>':
+next_char();
+if ('>' == current_char) {
+current_token = T_RightShift;
+next_char();
+} else if ('=' == current_char) {
+current_token = T_GreaterEqual;
+next_char();
+} else {
+current_token = T_GreaterThan;
+}
+break;
+case '=':
+next_char();
+if ('=' != current_char) {
+throw CdlParseException(std::string("Incomplete == operator in expression.\n") + get_error_location());
+} else {
+current_token = T_Equal;
+next_char();
+}
+break;
+case '&':
+next_char();
+if ('&' == current_char) {
+current_token = T_And;
+next_char();
+} else {
+current_token = T_BitAnd;
+}
+break;
+case '^':
+current_token = T_BitXor;
+next_char();
+break;
+case '|':
+next_char();
+if ('|' == current_char) {
+current_token = T_Or;
+next_char();
+} else {
+current_token = T_BitOr;
+}
+break;
+case ':':
+current_token = T_Colon;
+next_char();
+break;
+default:
+// String constants have been handled already. The only
+// valid tokens that are left are numbers, references and
+// the rang eoperator.
+//
+// Numbers should begin with a digit (plus and minus are
+// tokenized separately).
+//
+// References must be valid C preprocessor symbols, i.e.
+// they must begin with either a letter or an underscore.
+// The range operator is handled most conveniently as
+// a special case of a reference.
+if (isdigit(current_char)) {
+process_number();
+} else if (('_' == current_char) ||
+(('a' <= current_char) && (current_char <= 'z')) ||
+(('A' <= current_char) && (current_char <= 'Z'))) {
+process_reference();
+if ("to" == current_reference) {
+current_reference = "";
+current_token  = T_Range;
+} else {
+current_token = T_Reference;
+}
+} else {
+std::string msg = "Unexpected character '";
+msg += (char) current_char;
+msg += "' in expression.\n";
+msg += get_error_location();
+throw CdlParseException(msg);
+}
+break;
+}
+CYG_REPORT_RETVAL(current_token);
+}
+//}}}
+//{{{  initialise_tokenisation()
+// ----------------------------------------------------------------------------
+// This is called at the start of expression parsing. It
+// sets up the appropriate statics, and provides initial
+// values for current_char and current_token.
+static void
+initialise_tokenisation(std::string data, int index)
+{
+CYG_REPORT_FUNCNAME("initialise_tokenization");
+current_data        = data;
+current_index       = static_cast<unsigned int>(index);
+token_start         = current_index;
+next_char();
+next_token();
+CYG_REPORT_RETURN();
+}
+//}}}
+//}}}
+//{{{  Syntactic analysis
+// ----------------------------------------------------------------------------
+// Syntactic analysis.
+//
+// The BNF of CDL expressions is something like this:
+//
+//   <expression>   ::= <conditional>
+//   <conditional>  ::= <or> ? <conditional> : <conditional> | <or>
+//   <or>           ::= <and>    [<or op>     <and>]           ||
+//   <and>          ::= <bitor>  [<and op>    <bitor>]         ??
+//   <bitor>        ::= <bitxor> [<bitor op>  <bitxor>]        |
+//   <bitxor>       ::= <bitand> [<bitxor op> <bitand>]        ^
+//   <bitand>       ::= <eq>     [<bitand op> <eq>]            &
+//   <eq>           ::= <comp>   [<eq op>     <comp>]          == !=
+//   <comp>         ::= <shift>  [<comp op>   <shift>]         < <= > >=
+//   <shift>        ::= <add>    [<shift op>  <add>]           << >>
+//   <add>          ::= <mult>   [<add op>    <mult>]          + -
+//   <mult>         ::= <unary>  [<mult op>   <unary>]         * / %
+//   <unary>        ::= -<unary> | +<unary> | !<unary> | *<unary> | ?<unary> |
+//                      <string constant> | <integer constant> |
+//                      <double constant> | <reference> |
+//                      ( <expression> )
+//
+// There are separate functions for each of these terms.
+// A forward declaration, needed for bracketed subexpressions.
+static void parse_expression(CdlExpression) throw(CdlParseException, std::bad_alloc);
+// A utility to add a reference to the current expression, returning
+// the index.
+static int
+push_reference(CdlExpression expr, const std::string& reference)
+{
+CYG_REPORT_FUNCNAMETYPE("push_reference", "new index %d");
+CYG_PRECONDITION_CLASSC(expr);
+CdlReference ref(reference);
+expr->references.push_back(ref);
+int result = (int) expr->references.size() - 1;
+CYG_REPORT_RETVAL(result);
+return result;
+}
+// A utility to add a subexpression, returning its index.
+static void
+push_subexpression(CdlExpression expr, const CdlSubexpression& subexpr)
+{
+CYG_REPORT_FUNCNAME("push_subexpression");
+CYG_PRECONDITION_CLASSC(expr);
+expr->sub_expressions.push_back(subexpr);
+expr->first_subexpression = ((int) expr->sub_expressions.size()) - 1;
+CYG_REPORT_RETURN();
+}
+// Another utility to hold of the most recent subexpression
+static CdlSubexpression&
+current_subexpression(CdlExpression expr)
+{
+CYG_REPORT_FUNCNAME("current_subexpression");
+CdlSubexpression& result = expr->sub_expressions[expr->first_subexpression];
+CYG_REPORT_RETURN();
+return result;
+}
+static void
+parse_unary(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_operand");
+CYG_REPORT_FUNCARG1XV(expr);
+CYG_PRECONDITION_CLASSC(expr);
+CdlSubexpression subexpr;
+switch(current_token) {
+case T_EOD :
+{
+// This warrants a special case
+throw CdlParseException("End of expression reached when expecting an operand.\n" + get_error_location());
+}
+case T_Reference :
+{
+subexpr.op              = CdlExprOp_Reference;
+subexpr.reference_index = push_reference(expr, current_reference);
+push_subexpression(expr, subexpr);
+next_token();
+break;
+}
+case T_String :
+{
+subexpr.op              = CdlExprOp_StringConstant;
+subexpr.constants       = current_string;
+push_subexpression(expr, subexpr);
+next_token();
+break;
+}
+case T_Integer :
+{
+subexpr.op               = CdlExprOp_IntegerConstant;
+subexpr.constants.set_integer_value(current_int, current_format);
+push_subexpression(expr, subexpr);
+next_token();
+break;
+}
+case T_Double :
+{
+subexpr.op              = CdlExprOp_DoubleConstant;
+subexpr.constants.set_double_value(current_double, current_format);
+push_subexpression(expr, subexpr);
+next_token();
+break;
+}
+case T_OpenBracket :
+{
+next_token();
+parse_expression(expr);
+if (T_CloseBracket != current_token) {
+throw CdlParseException("Missing close bracket after subexpression.\n" + get_error_location());
+}
+next_token();
+break;
+}
+case T_Minus :
+{
+next_token();
+parse_unary(expr);
+CdlSubexpression& last_sub      = current_subexpression(expr);
+if (CdlExprOp_IntegerConstant == last_sub.op) {
+// Do the negating inline, no need for another subexpression.
+last_sub.constants = last_sub.constants.get_integer_value() * -1;
+} else if (CdlExprOp_DoubleConstant == last_sub.op) {
+last_sub.constants = last_sub.constants.get_double_value() * -1;
+} else {
+// We could detect certain cases such as string constants etc.
+// For now don't bother.
+subexpr.op          = CdlExprOp_Negate;
+subexpr.lhs_index   = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+break;
+}
+case T_Plus :
+{
+next_token();
+parse_unary(expr);
+CdlSubexpression& last_sub      = current_subexpression(expr);
+if ((CdlExprOp_IntegerConstant == last_sub.op) || (CdlExprOp_DoubleConstant == last_sub.op)) {
+// No need to do anything here.
+} else {
+subexpr.op          = CdlExprOp_Plus;
+subexpr.lhs_index   = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+break;
+}
+case T_Times :
+{
+next_token();
+parse_unary(expr);
+subexpr.op            = CdlExprOp_Indirect;
+subexpr.lhs_index     = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+break;
+}
+case T_Exclamation :
+{
+next_token();
+parse_unary(expr);
+subexpr.op            = CdlExprOp_LogicalNot;
+subexpr.lhs_index     = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+break;
+}
+case T_Tilde :
+{
+next_token();
+parse_unary(expr);
+subexpr.op            = CdlExprOp_BitNot;
+subexpr.lhs_index     = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+break;
+}
+case T_Questionmark:
+{
+// This is the `active' operator, it can only be applied directly to a reference.
+next_token();
+parse_unary(expr);
+CdlSubexpression& last_sub = current_subexpression(expr);
+if (CdlExprOp_Reference != last_sub.op) {
+throw CdlParseException("The active operator ? can only be applied directly to a reference.\n" +
+get_error_location());
+}
+// There is no point in creating a new subexpression object, just modify
+// the existing one. This has the useful side effect of avoiding
+// reference substitution in the eval code.
+last_sub.op           = CdlExprOp_Active;
+break;
+}
+default:
+{
+throw CdlParseException("Unexpected token `" + token_to_string() + "', expecting an operand.\n" +
+get_error_location());
+}
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_multiply(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_multiply");
+parse_unary(expr);
+while ((T_Times == current_token) || (T_Divide == current_token) || (T_Remainder == current_token)) {
+CdlSubexpression subexpr;
+subexpr.op      =
+(T_Times  == current_token) ? CdlExprOp_Multiply :
+(T_Divide == current_token) ? CdlExprOp_Divide : CdlExprOp_Remainder;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_unary(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_add(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_add");
+parse_multiply(expr);
+while ((T_Plus == current_token) || (T_Minus == current_token)) {
+CdlSubexpression subexpr;
+subexpr.op = (T_Plus == current_token) ? CdlExprOp_Add : CdlExprOp_Subtract;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_multiply(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_shift(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_shift");
+parse_add(expr);
+while ((T_LeftShift == current_token) || (T_RightShift == current_token)) {
+CdlSubexpression subexpr;
+subexpr.op = (T_LeftShift == current_token) ? CdlExprOp_LeftShift : CdlExprOp_RightShift;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_add(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_comparison(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_comparison");
+parse_shift(expr);
+while ((T_LessThan == current_token)    || (T_LessEqual    == current_token) ||
+(T_GreaterThan == current_token) || (T_GreaterEqual == current_token))  {
+CdlSubexpression subexpr;
+subexpr.op =
+(T_LessThan    == current_token) ? CdlExprOp_LessThan :
+(T_LessEqual   == current_token) ? CdlExprOp_LessEqual :
+(T_GreaterThan == current_token) ? CdlExprOp_GreaterThan : CdlExprOp_GreaterEqual;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_shift(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_equals(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_equals");
+parse_comparison(expr);
+while ((T_Equal == current_token) || (T_NotEqual == current_token)) {
+CdlSubexpression subexpr;
+subexpr.op = (T_Equal == current_token) ? CdlExprOp_Equal : CdlExprOp_NotEqual;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_comparison(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_bitand(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_bitand");
+parse_equals(expr);
+while (T_BitAnd == current_token) {
+CdlSubexpression subexpr;
+subexpr.op = CdlExprOp_BitAnd;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_equals(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_bitxor(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_bitxor");
+parse_bitand(expr);
+while (T_BitXor == current_token) {
+CdlSubexpression subexpr;
+subexpr.op = CdlExprOp_BitXor;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_bitand(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_bitor(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_bitor");
+parse_bitxor(expr);
+while (T_BitOr == current_token) {
+CdlSubexpression subexpr;
+subexpr.op = CdlExprOp_BitOr;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_bitxor(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_and(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_and");
+parse_bitor(expr);
+while (T_And == current_token) {
+CdlSubexpression subexpr;
+subexpr.op = CdlExprOp_And;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_bitor(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_or(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_or");
+parse_and(expr);
+while (T_Or == current_token) {
+CdlSubexpression subexpr;
+subexpr.op = CdlExprOp_Or;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_and(expr);
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_conditional(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_conditional");
+parse_or(expr);
+if (T_Questionmark == current_token) {
+CdlSubexpression subexpr;
+subexpr.op = CdlExprOp_Cond;
+subexpr.lhs_index = expr->first_subexpression;
+next_token();
+parse_conditional(expr);
+subexpr.rhs_index = expr->first_subexpression;
+if (T_Colon != current_token) {
+throw CdlParseException("Expected colon in conditional expression.\n" + get_error_location());
+}
+next_token();
+parse_conditional(expr);
+subexpr.rrhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+CYG_REPORT_RETURN();
+}
+static void
+parse_expression(CdlExpression expr)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("parse_expression");
+parse_conditional(expr);
+CYG_REPORT_RETURN();
+}
+// ----------------------------------------------------------------------------
+// The entry point.
+void
+CdlExpressionBody::continue_parse(CdlExpression expr, std::string data, int& index, CdlExprOp& token, int& token_end)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlExpression::continue_parse");
+CYG_REPORT_FUNCARG1XV(expr);
+CYG_PRECONDITION_CLASSC(expr);
+CYG_PRECONDITIONC((CdlExprOp_Invalid == token) || (CdlExprOp_And == token));
+int current_subexpr = expr->first_subexpression;
+initialise_tokenisation(data, index);
+parse_expression(expr);
+if (CdlExprOp_And == token) {
+CdlSubexpression subexpr;
+subexpr.op        = CdlExprOp_And;
+subexpr.lhs_index = current_subexpr;
+subexpr.rhs_index = expr->first_subexpression;
+push_subexpression(expr, subexpr);
+}
+token       = token_to_expr_op();
+index       = token_start;
+token_end   = current_index;
+CYG_REPORT_RETURN();
+}
+//}}}
+//}}}
+//{{{  Expression Evaluation
+// ----------------------------------------------------------------------------
+// Expression evaluation. This always happens in the context of a
+// particular toplevel. The parsed expression is held in what amounts
+// to a simple tree, so evaluation involves some recursion and a big
+// switch statement.
+static void
+evaluate_subexpr(CdlEvalContext& context, CdlExpression expr, int subexpr_index, CdlSimpleValue& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("evaluate_subexpr");
+CYG_REPORT_FUNCARG2XV(expr, subexpr_index);
+CYG_ASSERTC((subexpr_index >= 0) && ((unsigned int)subexpr_index < expr->sub_expressions.size()));
+const CdlSubexpression& subexpr = expr->sub_expressions[subexpr_index];
+switch(subexpr.op) {
+case CdlExprOp_StringConstant :
+case CdlExprOp_IntegerConstant :
+case CdlExprOp_DoubleConstant :
+{
+result = subexpr.constants;
+break;
+}
+case CdlExprOp_Reference :
+{
+// This expression may be happening in the context of a particular
+// property. If so then the destination may or may not be resolved,
+// and this is significant in the context of loading and unloading.
+// Alternatively this expression may be being evaluated inside
+// some Tcl code, with no particular context.
+CdlNode destination = 0;
+if (0 != context.property) {
+// There is a property, use the bound/unbound reference.
+destination = expr->references[subexpr.reference_index].get_destination();
+} else {
+// The destination name can be retrieved, but we still need some
+// way of resolving it.
+if (0 != context.toplevel) {
+std::string destination_name = expr->references[subexpr.reference_index].get_destination_name();
+destination = context.toplevel->lookup(destination_name);
+}
+}
+if (0 == destination) {
+// There are two ways of handling this.
+//   1) throw an eval exception, which will usually result
+//      in a new conflict object
+//   2) substitute a value of 0.
+// There should already be a conflict object for an
+// unresolved reference, and having two conflicts for
+// essentially the same error is not useful. Using a value
+// of 0 allows things to continue for a bit longer. It is
+// consistent with active vs. inactive values, gives
+// basically the right result for "requires" properties,
+// and so on.
+//
+// For now option (2) has it, but this decision may be
+// reversed in future.
+result = false;
+} else {
+CdlValuable valuable = dynamic_cast<CdlValuable>(destination);
+if (0 == valuable) {
+// This is a serious problem, an exception is warranted.
+throw CdlEvalException("The expression references `" + destination->get_class_name() + " " +
+destination->get_name() + "' which does not have a value.");
+} else {
+CdlSimpleValue::eval_valuable(context, valuable, result);
+}
+}
+break;
+}
+case CdlExprOp_Negate :
+{
+// Unary -. Evaluate the target. If it is numeric, fine. Otherwise
+// an error is warranted.
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if (result.has_integer_value()) {
+result.set_integer_value(-1 * result.get_integer_value());
+} else if (result.has_double_value()) {
+result.set_double_value(-1.0 * result.get_double_value());
+} else {
+throw CdlEvalException("Attempt to negate non-numeric value `" + result.get_value() + "'.");
+}
+break;
+}
+case CdlExprOp_Plus :
+{
+// Unary +. Essentially this just checks that the current value is numeric.
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if ((!result.has_integer_value()) && (!result.has_double_value())) {
+throw CdlEvalException("Attempt to apply unary + operator to non-numeric value `" + result.get_value() + "'.");
+}
+break;
+}
+case CdlExprOp_LogicalNot :
+{
+// !x
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if (result.get_bool_value()) {
+result = false;;
+} else {
+result = true;
+}
+result.set_value_format(CdlValueFormat_Default);
+break;
+}
+case CdlExprOp_BitNot :
+{
+// ~x. The operand must be an integer value.
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if (result.has_integer_value()) {
+cdl_int tmp = result.get_integer_value();
+result = ~tmp;
+} else {
+throw CdlEvalException("Attempt to apply unary ~ operator to non-integer value `" + result.get_value() + "'.");
+}
+break;
+}
+case CdlExprOp_Indirect :
+{
+// *x. The operand must evaluate to a string, and that string should be
+// the name of a CdlValuable object.
+CdlNode destination = 0;
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+std::string name = result.get_value();
+if (0 != context.toplevel) {
+destination = context.toplevel->lookup(name);
+} else {
+CYG_FAIL("This situation should probably never happen.");
+}
+if (0 == destination) {
+throw CdlEvalException("Attempt to apply unary indirection operator * to `" + name +
+"', which is not the name of a known CDL entity.");
+} else {
+CdlValuable valuable = dynamic_cast<CdlValuable>(destination);
+if (0 == valuable) {
+throw CdlEvalException("Attempt to apply unary indirection operator * to `" + name +
+"', which does not have a value.");
+} else {
+CdlSimpleValue::eval_valuable(context, valuable, result);
+}
+}
+break;
+}
+case CdlExprOp_Active :
+{
+// ?x. If x is currently unresolved then default to 0.
+// See the CdlExprOp_Reference code above for a similar case.
+CdlNode destination = 0;
+if (0 != context.property) {
+destination =  expr->references[subexpr.reference_index].get_destination();
+} else {
+if (0 != context.toplevel) {
+std::string destination_name = expr->references[subexpr.reference_index].get_destination_name();
+destination = context.toplevel->lookup(destination_name);
+}
+}
+bool active = false;
+if ((0 != destination) && context.transaction->is_active(destination)) {
+active = true;
+}
+if (active) {
+result = true;
+} else {
+result = false;
+}
+break;
+}
+case CdlExprOp_Multiply :
+{
+// x * y. For now this only makes sense for numerical data,
+// but it is possible to mix and match integer and double
+// precision data.
+//
+// Strictly speaking the rhs need only be evaluated if it
+// is known that the lhs is numeric.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((!(lhs.has_integer_value() || lhs.has_double_value())) ||
+(!(rhs.has_integer_value() || rhs.has_double_value()))) {
+throw CdlEvalException("Attempt to multiply non-numerical values: `" + lhs.get_value() + "' * `" +
+rhs.get_value() + "'.");
+}
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+result = lhs.get_integer_value() * rhs.get_integer_value();
+} else {
+result = lhs.get_double_value() * rhs.get_double_value();
+}
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_Divide :
+{
+// x / y. Basically the same as multiplication, apart from a check for
+// division by zero.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((!(lhs.has_integer_value() || lhs.has_double_value())) ||
+(!(rhs.has_integer_value() || rhs.has_double_value()))) {
+throw CdlEvalException("Attempt to divide non-numerical values: `" + lhs.get_value() + "' / `" +
+rhs.get_value() + "'.");
+}
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+cdl_int rhs_val = rhs.get_integer_value();
+if (0 == rhs_val) {
+throw CdlEvalException("Division by zero error: `" + lhs.get_value() + "' / `" + rhs.get_value() + "'.");
+} else {
+result = lhs.get_integer_value() / rhs_val;
+}
+} else {
+double rhs_val = rhs.get_double_value();
+if (0.0 == rhs_val) {
+throw CdlEvalException("Division by zero error: `" + lhs.get_value() + "' / `" + rhs.get_value() + "'.");
+}
+result = lhs.get_double_value() / rhs_val;
+}
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_Remainder :
+{
+// x % y. Both operands must be integral.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if (!(lhs.has_integer_value() && rhs.has_integer_value())) {
+throw CdlEvalException("Attempt to use the remainder operator on non integral data: `" +
+lhs.get_value() + "' % `" + rhs.get_value() + "'.");
+}
+cdl_int rhs_val = rhs.get_integer_value();
+if (0 == rhs_val) {
+throw CdlEvalException("Division by zero error: `" + lhs.get_value() + "' % `" + rhs.get_value() + "'.");
+}
+result = lhs.get_integer_value() % rhs_val;
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_Add :
+{
+// x + y. For now this only makes sense for numerical data,
+// but it is possible to mix and match integer and double
+// precision data. Arguably for string data this operator
+// should mean concatenation, but it would probably be
+// safer to have a separate operator for that.
+//
+// Strictly speaking the rhs need only be evaluated if it
+// is known that the lhs is numeric.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((!(lhs.has_integer_value() || lhs.has_double_value())) ||
+(!(rhs.has_integer_value() || rhs.has_double_value()))) {
+throw CdlEvalException("Attempt to add non-numerical values: `" + lhs.get_value() + "' + `" +
+rhs.get_value() + "'.");
+}
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+result = lhs.get_integer_value() + rhs.get_integer_value();
+} else {
+result = lhs.get_double_value() + rhs.get_double_value();
+}
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_Subtract :
+{
+// x - y. Again only numerical data is supported for now.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((!(lhs.has_integer_value() || lhs.has_double_value())) ||
+(!(rhs.has_integer_value() || rhs.has_double_value()))) {
+throw CdlEvalException("Attempt to subtract non-numerical values: `" + lhs.get_value() + "' - `" +
+rhs.get_value() + "'.");
+}
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+result = lhs.get_integer_value() - rhs.get_integer_value();
+} else {
+result = lhs.get_double_value() - rhs.get_double_value();
+}
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_LeftShift :
+{
+// x << y. Both operands must be integral. For now there is no
+// check on the value of y.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if (!(lhs.has_integer_value() && rhs.has_integer_value())) {
+throw CdlEvalException("Attempt to use the left-shift operator on non integral data: `" +
+lhs.get_value() + "' << `" + rhs.get_value() + "'.");
+}
+result = lhs.get_integer_value() << rhs.get_integer_value();
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_RightShift :
+{
+// x >> y. Both operands must be integral. For now there is no
+// check on the value of y.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if (!(lhs.has_integer_value() && rhs.has_integer_value())) {
+throw CdlEvalException("Attempt to use the right-shift operator on non integral data: `" +
+lhs.get_value() + "' >> `" + rhs.get_value() + "'.");
+}
+result = lhs.get_integer_value() >> rhs.get_integer_value();
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_LessThan :
+case CdlExprOp_LessEqual :
+case CdlExprOp_GreaterThan :
+case CdlExprOp_GreaterEqual :
+{
+// x < y, and similar comparison operators. These share
+// sufficient code to warrant a common implementation. Only
+// numerical data is supported for now. These operator could
+// be interpreted as e.g. substring operations, but arguably
+// separate operators would be better for that.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((!(lhs.has_integer_value() || lhs.has_double_value())) ||
+(!(rhs.has_integer_value() || rhs.has_double_value()))) {
+std::string op_str =
+(CdlExprOp_LessThan    == subexpr.op) ? "<" :
+(CdlExprOp_LessEqual   == subexpr.op) ? "<=" :
+(CdlExprOp_GreaterThan == subexpr.op) ? ">" : ">=";
+throw CdlEvalException("Attempt to compare non-numerical values: `" + lhs.get_value() +
+"' " + op_str + " `" + rhs.get_value() + "'.");
+}
+bool val = false;
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+cdl_int lhs_val = lhs.get_integer_value();
+cdl_int rhs_val = rhs.get_integer_value();
+val =
+(CdlExprOp_LessThan    == subexpr.op) ? (lhs_val <  rhs_val) :
+(CdlExprOp_LessEqual   == subexpr.op) ? (lhs_val <= rhs_val) :
+(CdlExprOp_GreaterThan == subexpr.op) ? (lhs_val >  rhs_val) : (lhs_val >= rhs_val);
+} else {
+double lhs_val = lhs.get_double_value();
+double rhs_val = rhs.get_double_value();
+val =
+(CdlExprOp_LessThan    == subexpr.op) ? (lhs_val <  rhs_val) :
+(CdlExprOp_LessEqual   == subexpr.op) ? (lhs_val <= rhs_val) :
+(CdlExprOp_GreaterThan == subexpr.op) ? (lhs_val >  rhs_val) : (lhs_val >= rhs_val);
+}
+result = val;
+break;
+}
+case CdlExprOp_Equal :
+{
+// x == y. For numerical data this should be a numerical comparison.
+// Otherwise a string comparison has to be used.
+bool val = false;
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((lhs.has_integer_value() || lhs.has_double_value()) &&
+(rhs.has_integer_value() || rhs.has_double_value())) {
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+if (lhs.get_integer_value() == rhs.get_integer_value()) {
+val = true;
+} else {
+val = false;
+}
+} else {
+if (lhs.get_double_value() == rhs.get_double_value()) {
+val = true;
+} else {
+val = false;
+}
+}
+} else {
+// At least one of the two sides is non-numerical. Do a string comparison.
+if (lhs.get_value() == rhs.get_value()) {
+val = true;
+} else {
+val = false;
+}
+}
+result = val;
+break;
+}
+case CdlExprOp_NotEqual :
+{
+// x != y. For numerical data this should be a numerical comparison.
+// Otherwise a string comparison has to be used.
+bool val = false;
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if ((lhs.has_integer_value() || lhs.has_double_value()) &&
+(rhs.has_integer_value() || rhs.has_double_value())) {
+if (lhs.has_integer_value() && rhs.has_integer_value()) {
+if (lhs.get_integer_value() != rhs.get_integer_value()) {
+val = true;
+} else {
+val = false;
+}
+} else {
+if (lhs.get_double_value() != rhs.get_double_value()) {
+val = true;
+} else {
+val = false;
+}
+}
+} else {
+// At least one of the two sides is non-numerical. Do a string comparison.
+if (lhs.get_value() != rhs.get_value()) {
+val = true;
+} else {
+val = false;
+}
+}
+result = val;
+break;
+}
+case CdlExprOp_BitAnd :
+{
+// x & y. Only integer data is supported.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if (!(lhs.has_integer_value() && rhs.has_integer_value())) {
+throw CdlEvalException("Attempt to use the bitwise and operator on non integral data: `" +
+lhs.get_value() + "' & `" + rhs.get_value() + "'.");
+}
+result = lhs.get_integer_value() & rhs.get_integer_value();
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_BitXor :
+{
+// x ^ y. Only integer data is supported.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if (!(lhs.has_integer_value() && rhs.has_integer_value())) {
+throw CdlEvalException("Attempt to use the bitwise xor operator on non integral data: `" +
+lhs.get_value() + "' ^ `" + rhs.get_value() + "'.");
+}
+result = lhs.get_integer_value() ^ rhs.get_integer_value();
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_BitOr :
+{
+// x | y. Only integer data is supported.
+CdlSimpleValue lhs;
+CdlSimpleValue rhs;
+evaluate_subexpr(context, expr, subexpr.lhs_index, lhs);
+evaluate_subexpr(context, expr, subexpr.rhs_index, rhs);
+if (!(lhs.has_integer_value() && rhs.has_integer_value())) {
+throw CdlEvalException("Attempt to use the bitwise or operator on non integral data: `" +
+lhs.get_value() + "' | `" + rhs.get_value() + "'.");
+}
+result = lhs.get_integer_value() | rhs.get_integer_value();
+result.set_value_format(lhs, rhs);
+break;
+}
+case CdlExprOp_And :
+{
+// x && y. Both sides should be interpreted as boolean values,
+// and "y" should only be evaluated if necessary.
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if (!result.get_bool_value()) {
+result = false;
+} else {
+evaluate_subexpr(context, expr, subexpr.rhs_index, result);
+if (result.get_bool_value()) {
+result = true;
+} else {
+result = false;
+}
+}
+break;
+}
+case CdlExprOp_Or :
+{
+// x || y. Both sides should be interpreted as boolean values,
+// and "y" should only be evaluated if necessary.
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if (result.get_bool_value()) {
+result = true;
+} else {
+evaluate_subexpr(context, expr, subexpr.rhs_index, result);
+if (result.get_bool_value()) {
+result = true;
+} else {
+result = false;
+}
+}
+break;
+}
+case CdlExprOp_Cond :
+{
+// x ? a : b.
+// First evaluate the condition. Then evaluate either the second
+// or third argument, as appropriate.
+evaluate_subexpr(context, expr, subexpr.lhs_index, result);
+if (result.get_bool_value()) {
+evaluate_subexpr(context, expr, subexpr.rhs_index, result);
+} else {
+evaluate_subexpr(context, expr, subexpr.rrhs_index, result);
+}
+break;
+}
+default:
+break;
+}
+CYG_REPORT_RETURN();
+}
+// ----------------------------------------------------------------------------
+void
+CdlExpressionBody::eval_internal(CdlEvalContext& context, CdlSimpleValue& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlExpression::eval_internal)");
+CYG_REPORT_FUNCARG3XV(this, &context, &result);
+CYG_INVARIANT_THISC(CdlExpressionBody);
+CYG_PRECONDITION_CLASSOC(context);
+evaluate_subexpr(context, this, first_subexpression, result);
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  CdlExpression
+//{{{  Construction
+// ----------------------------------------------------------------------------
+// Ordinary expressions.
+//
+// The default constructor is private and does very little. Expressions
+// are created primarily by means of the parse() member function. There
+// is an argument for having constructors that take the same arguments
+// as the parse() member functions and relying on exception handling,
+// but that gets tricky for goal expressions and continue_parse().
+//
+// The copy constructor is protected and is used when creating e.g.
+// a default_value property object, which inherits from the ordinary
+// expression class. Again it might be better to do the parsing in
+// the constructor itself.
+//
+// The assignment operator is private and illegal.
+CdlExpressionBody::CdlExpressionBody()
+{
+CYG_REPORT_FUNCNAME("CdlExpression:: default constructor");
+CYG_REPORT_FUNCARG1XV(this);
+expression_string           = "";
+first_subexpression         = -1;
+cdlexpressionbody_cookie    = CdlExpressionBody_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+CdlExpressionBody::CdlExpressionBody(const CdlExpressionBody& original)
+{
+CYG_REPORT_FUNCNAME("CdlExpression:: copy constructor");
+CYG_REPORT_FUNCARG2XV(this, &original);
+CYG_INVARIANT_CLASSOC(CdlExpressionBody, original);
+// Sub-expressions are simple structs, so this should result in a bit-wise
+// copy of each vector element
+sub_expressions     = original.sub_expressions;
+// Simple scalar
+first_subexpression = original.first_subexpression;
+// The CdlReference class has a valid copy constructor and assignment
+// operator, provided that the reference is not yet bound. This should
+// be true when this copy constructor gets invoked, after parsing
+// and during the construction of a derived property object.
+references          = original.references;
+expression_string   = original.expression_string;
+cdlexpressionbody_cookie    = CdlExpressionBody_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  check_this()
+// ----------------------------------------------------------------------------
+// check_this(). Expression objects can exist before any parsing has
+// happened, not to mention in the middle of parsing. The
+// first_subexpression field can be used to detect this.
+bool
+CdlExpressionBody::check_this(cyg_assert_class_zeal zeal) const
+{
+if (CdlExpressionBody_Magic != cdlexpressionbody_cookie) {
+return false;
+}
+CYGDBG_MEMLEAK_CHECKTHIS();
+if (-1 == first_subexpression) {
+return true;
+}
+switch(zeal) {
+case cyg_system_test    :
+case cyg_extreme        :
+case cyg_thorough       :
+{
+for (std::vector<CdlReference>::const_iterator i = references.begin(); i != references.end(); i++) {
+if (!i->check_this(cyg_quick)) {
+return false;
+}
+}
+}
+case cyg_quick          :
+if ((unsigned)first_subexpression >= sub_expressions.size()) {
+return false;
+}
+case cyg_trivial        :
+case cyg_none           :
+break;
+}
+return true;
+}
+//}}}
+//{{{  Destruction
+CdlExpressionBody::~CdlExpressionBody()
+{
+CYG_REPORT_FUNCNAME("CdlExpression::destructor");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+cdlexpressionbody_cookie    = CdlExpressionBody_Invalid;
+first_subexpression         = -1;
+sub_expressions.clear();
+expression_string           = "";
+// This assumes that all references have been unbound already by
+// higher-level destructors.
+references.clear();
+CYGDBG_MEMLEAK_DESTRUCTOR();
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  Parsing - exported interface
+// ----------------------------------------------------------------------------
+// parse(string) invokes parse(string, ...) and checks that the expression
+// has terminated with EOD. Parsing of list expressions etc. can terminate
+// with some other token.
+//
+// parse(string, ...) allocates the expression object and invokes
+// continue_parse().
+//
+// continue_parse() is supposed to do all the hard work.
+CdlExpression
+CdlExpressionBody::parse(std::string data)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlExpression::parse", "result %p");
+CdlExpression       result  = 0;
+int                 index   = 0;
+CdlExprOp           next_op = CdlExprOp_Invalid;
+int                 end_index;
+result = parse(data, index, next_op, end_index);
+// Either there has already been a parsing or out-of-memory
+// exception, or we should be at the end of the expression string.
+if (CdlExprOp_EOD != next_op) {
+delete result;
+throw CdlParseException("unexpected data at end of expression.\n" + get_error_location());
+}
+// Keep a copy of the original string for diagnostics purposes.
+result->expression_string = data;
+CYG_REPORT_RETVAL(result);
+return result;
+}
+CdlExpression
+CdlExpressionBody::parse(std::string data, int& index, CdlExprOp& next_token, int& token_end)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlExpression::parse", "result %d");
+CdlExpression result = new CdlExpressionBody;
+try {
+continue_parse(result, data, index, next_token, token_end);
+}
+catch (...) {
+delete result;
+throw;
+}
+CYG_REPORT_RETVAL(result);
+return result;
+}
+//}}}
+//{{{  update()
+// ----------------------------------------------------------------------------
+// There has been a change in the toplevel which involves entities being
+// created or destroyed, and reference resolution is required.
+bool
+CdlExpressionBody::update(CdlTransaction transaction, CdlNode source, CdlProperty source_prop, CdlNode dest, CdlUpdate change)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlExpression::update", "result %d");
+CYG_REPORT_FUNCARG6XV(this, transaction, source, source_prop, dest, change);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSC(source);
+CYG_PRECONDITION_CLASSC(source_prop);
+CdlToplevel toplevel        = source->get_toplevel();
+bool        result          = false;
+std::vector<CdlReference>::iterator ref_i;
+switch(change) {
+case CdlUpdate_Loaded:
+{
+// The source package has just been loaded. Try to resolve every
+// reference, creating CdlConflict objects where necessary.
+CYG_ASSERTC(0 == dest);
+for (ref_i = references.begin(); ref_i != references.end(); ref_i++) {
+dest = toplevel->lookup(ref_i->get_destination_name());
+if (0 == dest) {
+CdlConflict_UnresolvedBody::make(transaction, source, source_prop, ref_i->get_destination_name());
+} else {
+ref_i->bind(source, source_prop, dest);
+}
+}
+result = true;
+break;
+}
+case CdlUpdate_Unloading:
+{
+// The source package is being unloaded. Unbind all currently bound references.
+// Also destroy any unresolved conflicts.
+CYG_ASSERTC(0 == dest);
+for (ref_i = references.begin(); ref_i != references.end(); ref_i++) {
+dest = ref_i->get_destination();
+if (0 != dest) {
+ref_i->unbind(source, source_prop);
+}
+}
+result = true;
+break;
+}
+case CdlUpdate_Created :
+{
+// A previously unresolved reference can now be resolved.
+// It is necessary to search the vector for an unresolved
+// reference with the desired name, and do the binding.
+// This search may fail in the case of list expressions.
+CYG_ASSERT_CLASSC(dest);
+std::string dest_name = dest->get_name();
+for (ref_i = references.begin(); !result && (ref_i != references.end()); ref_i++) {
+if ((dest_name == ref_i->get_destination_name()) && (0 == ref_i->get_destination())) {
+ref_i->bind(source, source_prop, dest);
+result = true;
+std::vector<CdlConflict> conflicts;
+std::vector<CdlConflict>::iterator conf_i;
+transaction->get_structural_conflicts(source, source_prop, &CdlConflict_UnresolvedBody::test, conflicts);
+for (conf_i = conflicts.begin(); conf_i != conflicts.end(); conf_i++) {
+CdlConflict_Unresolved real_conf = dynamic_cast<CdlConflict_Unresolved>(*conf_i);
+CYG_ASSERTC(0 != real_conf);
+if (dest_name == real_conf->get_target_name()) {
+transaction->clear_conflict(real_conf);
+break;
+}
+}
+CYG_ASSERTC(conf_i != conflicts.end());
+}
+}
+break;
+}
+case CdlUpdate_Destroyed :
+{
+// A previously resolved reference is about to become illegal.
+// Search the vector for a resolved reference object matching
+// the destination, and unbind it. Also create a conflict
+// object. The search can fail in the case of list expressions
+CYG_ASSERT_CLASSC(dest);
+for (ref_i = references.begin(); !result && (ref_i != references.end()); ref_i++) {
+if (dest == ref_i->get_destination()) {
+ref_i->unbind(source, source_prop);
+CdlConflict_UnresolvedBody::make(transaction, source, source_prop, ref_i->get_destination_name());
+result = true;
+}
+}
+break;
+}
+default :
+CYG_FAIL("Illegal change type passed to CdlExpression::update");
+break;
+}
+CYG_REPORT_RETVAL(result);
+return result;
+}
+//}}}
+//{{{  Evaluation
+// ----------------------------------------------------------------------------
+// Expression evaluation. At the end of the day everything filters through
+// to eval_internal() which should all the hard work.
+//
+// The eval() member function handles EvalException conflicts. The
+// eval_internal() member function does not, and is used for list
+// and goal expressions as well.
+void
+CdlExpressionBody::eval(CdlEvalContext& context, CdlSimpleValue& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlExpression::eval");
+try {
+eval_internal(context, result);
+// Evaluation has succeeded, so if there was an EvalException
+// conflict get rid of it. This can only happen in the context
+// of a transaction.
+if ((0 != context.transaction) && (0 != context.node) && (0 != context.property)) {
+context.transaction->clear_conflicts(context.node, context.property, &CdlConflict_EvalExceptionBody::test);
+}
+} catch(CdlEvalException e) {
+if ((0 != context.transaction) && (0 != context.node) && (0 != context.property)) {
+CdlConflict conflict = context.transaction->get_conflict(context.node, context.property,
+&CdlConflict_EvalExceptionBody::test);
+if (0 == conflict) {
+CdlConflict_EvalExceptionBody::make(context.transaction, context.node, context.property, e.get_message());
+} else {
+CdlConflict_EvalException eval_conf = dynamic_cast<CdlConflict_EvalException>(conflict);
+CYG_ASSERTC(0 != eval_conf);
+if (eval_conf->get_explanation() != e.get_message()) {
+// Replace the conflict object. That way higher level code gets informed
+// there has been a change.
+context.transaction->clear_conflict(conflict);
+CdlConflict_EvalExceptionBody::make(context.transaction, context.node, context.property, e.get_message());
+}
+}
+}
+throw;
+}
+}
+//}}}
+//{{{  Misc
+// ----------------------------------------------------------------------------
+std::string
+CdlExpressionBody::get_original_string() const
+{
+CYG_REPORT_FUNCNAME("CdlExpression::get_original_string");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+CYG_REPORT_RETURN();
+return expression_string;
+}
+//}}}
+//}}}
+//{{{  CdlListExpression
+//{{{  Construction
+// ----------------------------------------------------------------------------
+// The normal sequence of events is:
+//
+// 1) higher level code calls CdlListExpressionbody::parse()
+// 2) this static member creates a new and empty list expression object.
+//    The constructor need not do very much.
+// 3) the parse() member then fills in the newly created object
+// 4) the object is returned to higher-level code
+// 5) usually the list expression will now become part of
+//    a property object by means of a copy constructor.
+//
+// The only complication is that a list expression contains a vector
+// of CdlExpression pointers which must be freed during the destructor.
+// The copy constructor does not make duplicates of the individual
+// expression objects, instead ownership is transferred.
+CdlListExpressionBody::CdlListExpressionBody()
+{
+CYG_REPORT_FUNCNAME("CdlListExpression:: default constructor");
+CYG_REPORT_FUNCARG1XV(this);
+expression_string           = "";
+cdllistexpressionbody_cookie = CdlListExpressionBody_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+CdlListExpressionBody::CdlListExpressionBody(const CdlListExpressionBody& original)
+{
+CYG_REPORT_FUNCNAME("CdlListExpression:: copy constructor");
+CYG_REPORT_FUNCARG2XV(this, &original);
+CYG_INVARIANT_CLASSOC(CdlListExpressionBody, original);
+expression_string           = original.expression_string;
+// These copy across the pointers
+data        = original.data;
+ranges      = original.ranges;
+// And this clears out the pointers, but leaves the expression objects lying around
+CdlListExpression tmp = const_cast<CdlListExpression>(&original);
+tmp->data.clear();
+tmp->ranges.clear();
+cdllistexpressionbody_cookie = CdlListExpressionBody_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  Destruction
+CdlListExpressionBody::~CdlListExpressionBody()
+{
+CYG_REPORT_FUNCNAME("CdlListExpression:: destructor");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+cdllistexpressionbody_cookie        = CdlListExpressionBody_Invalid;
+expression_string                   = "";
+for (std::vector<CdlExpression>::iterator i = data.begin(); i != data.end(); i++) {
+delete *i;
+*i = 0;
+}
+for (std::vector<std::pair<CdlExpression, CdlExpression> >::iterator j = ranges.begin(); j != ranges.end(); j++) {
+delete j->first;
+delete j->second;
+j->first = 0;
+j->second = 0;
+}
+data.clear();
+ranges.clear();
+CYGDBG_MEMLEAK_DESTRUCTOR();
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  check_this()
+// ----------------------------------------------------------------------------
+bool
+CdlListExpressionBody::check_this(cyg_assert_class_zeal zeal) const
+{
+if (CdlListExpressionBody_Magic != cdllistexpressionbody_cookie) {
+return false;
+}
+CYGDBG_MEMLEAK_CHECKTHIS();
+switch(zeal) {
+case cyg_system_test      :
+case cyg_extreme          :
+case cyg_thorough         :
+{
+for (std::vector<CdlExpression>::const_iterator i = data.begin(); i != data.end(); i++) {
+if (!(*i)->check_this(cyg_quick)) {
+return false;
+}
+}
+for (std::vector<std::pair<CdlExpression,CdlExpression> >::const_iterator j = ranges.begin();
+j != ranges.end();
+j++) {
+if (!(j->first->check_this(cyg_quick)) || !(j->second->check_this(cyg_quick))) {
+return false;
+}
+}
+}
+case cyg_quick            :
+case cyg_trivial          :
+case cyg_none             :
+default                   :
+break;
+}
+return true;
+}
+//}}}
+//{{{  Parsing
+// ----------------------------------------------------------------------------
+// Parsing a list expression involves repeated parsing of ordinary
+// expressions until an EOD token is reached.
+CdlListExpression
+CdlListExpressionBody::parse(std::string data)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlListExpression::parse", "result %p");
+// Allocate an expression object that can then be filled in.
+CdlListExpression result = new CdlListExpressionBody;
+// Do the parsing in a try/catch statement to make sure the
+// allocated expression gets freed on a parse error.
+try {
+int             index           = 0;
+int             end_index       = 0;
+CdlExprOp       op              = CdlExprOp_Invalid;
+CdlExpression   expr1           = 0;
+do {
+// Try to parse the next expression in the list
+op    = CdlExprOp_Invalid;
+expr1 = CdlExpressionBody::parse(data, index, op, end_index);
+// There should now be a valid expression, failure would have
+// resulted in an exception.
+CYG_ASSERT_CLASSC(expr1);
+// Allow for ranges.
+if (CdlExprOp_Range != op) {
+// A simple expression, just add it to the current data vector
+// "index" will contain the appropriate value.
+result->data.push_back(expr1);
+} else {
+// A range expression. Get the other end of the range.
+// This requires manipulating index a bit.
+CdlExpression expr2 = 0;
+index = end_index;
+op    = CdlExprOp_Invalid;
+try {
+expr2 = CdlExpressionBody::parse(data, index, op, end_index);
+}
+catch (...) {
+delete expr1;
+throw;
+}
+result->ranges.push_back(std::make_pair(expr1, expr2));
+}
+} while (CdlExprOp_EOD != op);
+}
+catch (...) {
+delete result;
+throw;
+}
+// Keep track of the original string for diagnostics purposes
+result->expression_string = data;
+CYG_REPORT_RETVAL(result);
+return result;
+}
+//}}}
+//{{{  update()
+// ----------------------------------------------------------------------------
+// This code is invoked when it is necessary to update the references
+// for the list expression. There are four situations in which this
+// can happen: the package has just been loaded; the package is being
+// unloaded; a referenced target is being created; a referenced target is
+// being destroyed.
+//
+// The first two cases simply involve processing every expression that
+// makes up the overall list expression. The last two cases involve
+// searching through the expressions until an applicable one is found.
+// Note that an expression may contain multiple references to another
+// object, resulting in multiple calls to this function.
+bool
+CdlListExpressionBody::update(CdlTransaction transact, CdlNode source, CdlProperty source_prop, CdlNode dest, CdlUpdate change)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlListExpression::update", "result %d");
+CYG_REPORT_FUNCARG6XV(this, transact, source, source_prop, dest, change);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSC(source);
+CYG_PRECONDITION_CLASSC(source_prop);
+bool result = false;
+if ((CdlUpdate_Loaded == change) || (CdlUpdate_Unloading == change)) {
+std::vector<CdlExpression>::const_iterator expr_i;
+std::vector<std::pair<CdlExpression, CdlExpression> >::const_iterator pair_i;
+for (expr_i = data.begin(); expr_i != data.end(); expr_i++) {
+bool handled = (*expr_i)->update(transact, source, source_prop, dest, change);
+CYG_ASSERTC(handled);
+CYG_UNUSED_PARAM(bool, handled);
+}
+for (pair_i = ranges.begin(); pair_i != ranges.end(); pair_i++) {
+bool handled = pair_i->first->update(transact, source, source_prop, dest, change);
+CYG_ASSERTC(handled);
+handled = pair_i->second->update(transact, source, source_prop, dest, change);
+CYG_ASSERTC(handled);
+}
+result = true;
+} else {
+CYG_ASSERTC((CdlUpdate_Created == change) || (CdlUpdate_Destroyed == change));
+std::vector<CdlExpression>::const_iterator expr_i;
+std::vector<std::pair<CdlExpression, CdlExpression> >::const_iterator pair_i;
+for (expr_i = data.begin(); !result && (expr_i != data.end()); expr_i++) {
+result = (*expr_i)->update(transact, source, source_prop, dest, change);
+}
+for (pair_i = ranges.begin(); !result && (pair_i != ranges.end()); pair_i++) {
+result = pair_i->first->update(transact, source, source_prop, dest, change);
+if (!result) {
+result = pair_i->second->update(transact, source, source_prop, dest, change);
+}
+}
+}
+CYG_REPORT_RETVAL(result);
+return result;
+}
+//}}}
+//{{{  Evaluation
+// ----------------------------------------------------------------------------
+// Evaluation. The hard work is actually done in eval_internal()
+void
+CdlListExpressionBody::eval(CdlEvalContext& context, CdlListValue& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlListExpression::eval");
+CYG_REPORT_FUNCARG3XV(this, &context, &result);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+this->eval_internal(context, result);
+CYG_REPORT_RETURN();
+}
+// ----------------------------------------------------------------------------
+// This requires evaluating each expression in the data and ranges
+// vectors and adding the result to the appropriate vector in result.
+// Various error conditions are possible.
+void
+CdlListExpressionBody::eval_internal(CdlEvalContext& context, CdlListValue& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlListExpression::eval_internal");
+CYG_REPORT_FUNCARG2XV(this, &context);
+result.table.clear();
+result.integer_ranges.clear();
+result.double_ranges.clear();
+CdlSimpleValue val1;
+CdlSimpleValue val2;
+try {
+for (std::vector<CdlExpression>::const_iterator i = data.begin(); i != data.end(); i++) {
+(*i)->eval_internal(context, val1);
+if ("" != val1.get_value()) {
+result.table.push_back(val1);
+}
+}
+for (std::vector<std::pair<CdlExpression,CdlExpression> >::const_iterator j = ranges.begin(); j != ranges.end(); j++) {
+j->first->eval_internal(context, val1);
+j->second->eval_internal(context, val2);
+if (val1.has_integer_value() && val2.has_integer_value()) {
+cdl_int x1 = val1.get_integer_value();
+cdl_int x2 = val2.get_integer_value();
+if (x1 > x2) {
+cdl_int tmp = x1;
+x1 = x2;
+x2 = tmp;
+}
+result.integer_ranges.push_back(std::make_pair(x1, x2));
+} else if (val1.has_double_value() && val2.has_double_value()) {
+double x1 = val1.get_double_value();
+double x2 = val2.get_double_value();
+if (x1 > x2) {
+double tmp = x1;
+x1 = x2;
+x2 = tmp;
+}
+result.double_ranges.push_back(std::make_pair(x1, x2));
+} else {
+throw CdlEvalException("range expression involves non-numerical limits");
+}
+}
+// Any problems would have resulted in an exception. If there
+// was a previous EvalExeption for this property, it is no
+// longer applicable
+if ((0 != context.transaction) && (0 != context.node) && (0 != context.property)) {
+context.transaction->clear_conflicts(context.node, context.property, &CdlConflict_EvalExceptionBody::test);
+}
+} catch(CdlEvalException e) {
+if ((0 != context.transaction) && (0 != context.node) && (0 != context.property)) {
+CdlConflict conflict = context.transaction->get_conflict(context.node, context.property,
+&CdlConflict_EvalExceptionBody::test);
+if (0 == conflict) {
+CdlConflict_EvalExceptionBody::make(context.transaction, context.node, context.property, e.get_message());
+} else {
+CdlConflict_EvalException eval_conf = dynamic_cast<CdlConflict_EvalException>(conflict);
+CYG_ASSERTC(0 != eval_conf);
+if (eval_conf->get_explanation() != e.get_message()) {
+// Replace the conflict object. Higher level will be informed about this.
+context.transaction->clear_conflict(conflict);
+CdlConflict_EvalExceptionBody::make(context.transaction, context.node, context.property, e.get_message());
+}
+}
+}
+throw;
+}
+CYG_REPORT_RETURN();
+}
+//}}}
+//{{{  is_member()
+// ----------------------------------------------------------------------------
+bool
+CdlListExpressionBody::is_member(CdlEvalContext& context, CdlSimpleValue& val)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlListExpression::is_member (value)", "result %d");
+CYG_REPORT_FUNCARG3XV(this, &context, &val);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+bool result = false;
+CdlListValue list_val;
+eval_internal(context, list_val);
+result = list_val.is_member(val);
+CYG_REPORT_RETVAL(result);
+return result;
+}
+bool
+CdlListExpressionBody::is_member(CdlEvalContext& context, std::string val)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlListExpression::is_member (string)", "result %d");
+CYG_REPORT_FUNCARG2XV(this, &context);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+bool result = false;
+CdlListValue list_val;
+eval_internal(context, list_val);
+result = list_val.is_member(val);
+CYG_REPORT_RETVAL(result);
+return result;
+}
+bool
+CdlListExpressionBody::is_member(CdlEvalContext& context, cdl_int val)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlListExpression::is_member (int)", "result %d");
+CYG_REPORT_FUNCARG3XV(this, &context, (int) val);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+bool result = false;
+CdlListValue list_val;
+eval_internal(context, list_val);
+result = list_val.is_member(val);
+CYG_REPORT_RETVAL(result);
+return result;
+}
+bool
+CdlListExpressionBody::is_member(CdlEvalContext& context, double val)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlListExpression::is_member (double)", "result %d");
+CYG_REPORT_FUNCARG2XV(this, &context);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+bool result = false;
+CdlListValue list_val;
+eval_internal(context, list_val);
+result = list_val.is_member(val);
+CYG_REPORT_RETVAL(result);
+return result;
+}
+//}}}
+//{{{  Misc
+// ----------------------------------------------------------------------------
+std::string
+CdlListExpressionBody::get_original_string() const
+{
+CYG_REPORT_FUNCNAME("CdlListExpression::get_original_string");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+CYG_REPORT_RETURN();
+return expression_string;
+}
+//}}}
+//}}}
+//{{{  CdlGoalExpression
+// ----------------------------------------------------------------------------
+// Constructors etc. are pretty much as per ordinary and list
+// expressions. Most of the work is done in the private base class.
+CdlGoalExpressionBody::CdlGoalExpressionBody()
+: CdlExpressionBody()
+{
+CYG_REPORT_FUNCNAME("CdlGoalExpression::default_constructor");
+CYG_REPORT_FUNCARG1XV(this);
+expression_string           = "";
+cdlgoalexpressionbody_cookie = CdlGoalExpressionBody_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+CdlGoalExpressionBody::CdlGoalExpressionBody(const CdlGoalExpressionBody& original)
+: CdlExpressionBody(original)
+{
+CYG_REPORT_FUNCNAME("CdlGoalExpression:: copy constructor");
+CYG_REPORT_FUNCARG2XV(this, &original);
+CYG_INVARIANT_CLASSOC(CdlGoalExpressionBody, original);
+expression_string           = original.expression_string;
+cdlgoalexpressionbody_cookie = CdlGoalExpressionBody_Magic;
+CYGDBG_MEMLEAK_CONSTRUCTOR();
+CYG_POSTCONDITION_THISC();
+CYG_REPORT_RETURN();
+}
+CdlGoalExpressionBody::~CdlGoalExpressionBody()
+{
+CYG_REPORT_FUNCNAME("CdlGoalExpression:: destructor");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+cdlgoalexpressionbody_cookie = CdlGoalExpressionBody_Invalid;
+expression_string            = "";
+CYGDBG_MEMLEAK_DESTRUCTOR();
+CYG_REPORT_RETURN();
+}
+// ----------------------------------------------------------------------------
+// Parsing. A goal expression acts a bit like a list expression with
+// implicit && operators between the various expressions. It could be
+// implemented as a vector of expressions (which might make diagnostics
+// easier) but it is almost as easy to derive a goal expression from
+// an ordinary one.
+CdlGoalExpression
+CdlGoalExpressionBody::parse(std::string data)
+throw(CdlParseException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlGoalExpression::parse", "result %p");
+CdlGoalExpression result = new CdlGoalExpressionBody;
+try {
+int       index         = 0;
+CdlExprOp op            = CdlExprOp_Invalid;
+int       end_index     = 0;
+// Parse the first expression in the data.
+CdlExpressionBody::continue_parse(result, data, index, op, end_index);
+// At this stage we have reached end-of-data or we should be
+// at the start of another expression - any binary or ternary
+// operands would have been subsumed in the previous expression.
+// We need to keep adding && operators and new expressions until
+// end-of-data.
+while (CdlExprOp_EOD != op) {
+op = CdlExprOp_And;
+CdlExpressionBody::continue_parse(result, data, index, op, end_index);
+}
+}
+catch(...) {
+delete result;
+throw;
+}
+// Keep track of the original expression string for diagnostics purposes
+result->expression_string = data;
+CYG_REPORT_RETVAL(result);
+return result;
+}
+// ----------------------------------------------------------------------------
+void
+CdlGoalExpressionBody::eval(CdlEvalContext& context, bool& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlGoalExpression::eval");
+CYG_REPORT_FUNCARG2XV(this, &context);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+eval_internal(context, result);
+CYG_REPORT_RETURN();
+}
+bool
+CdlGoalExpressionBody::eval(CdlEvalContext& context)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAMETYPE("CdlGoalExpression::eval", "result %d");
+CYG_REPORT_FUNCARG2XV(this, &context);
+CYG_PRECONDITION_THISC();
+CYG_PRECONDITION_CLASSOC(context);
+bool result;
+eval_internal(context, result);
+CYG_REPORT_RETVAL(result);
+return result;
+}
+// ----------------------------------------------------------------------------
+// Provide access to the underlying CdlExpression object. This allows the
+// inference engine etc. to work out why a goal expression is failing
+CdlExpression
+CdlGoalExpressionBody::get_expression()
+{
+CYG_REPORT_FUNCNAMETYPE("CdlGoalExpression::get_expression", "result %p");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+CdlExpression result = this;
+CYG_REPORT_RETVAL(result);
+return result;
+}
+// ----------------------------------------------------------------------------
+bool
+CdlGoalExpressionBody::check_this(cyg_assert_class_zeal zeal) const
+{
+if (CdlGoalExpressionBody_Magic != cdlgoalexpressionbody_cookie) {
+return false;
+}
+CYGDBG_MEMLEAK_CHECKTHIS();
+// There is no data specific to a goal expression, just let the
+// underlying check_this() member do its stuff.
+return inherited::check_this(zeal);
+}
+// ----------------------------------------------------------------------------
+std::string
+CdlGoalExpressionBody::get_original_string() const
+{
+CYG_REPORT_FUNCNAME("CdlGoalExpression::get_original_string");
+CYG_REPORT_FUNCARG1XV(this);
+CYG_PRECONDITION_THISC();
+CYG_REPORT_RETURN();
+return expression_string;
+}
+// ----------------------------------------------------------------------------
+void
+CdlGoalExpressionBody::eval_internal(CdlEvalContext& context, bool& result)
+throw(CdlEvalException, std::bad_alloc)
+{
+CYG_REPORT_FUNCNAME("CdlGoalExpression::eval_internal");
+CYG_REPORT_FUNCARG2XV(this, &context);
+// The assertions are all done in the calling code
+// Start by evaluating the underlying expression
+CdlSimpleValue      val;
+try {
+inherited::eval_internal(context, val);
+// The evaluation succeeded. Do we have an integer, a string, ...?
+if (val.has_integer_value()) {
+result = (0 != val.get_integer_value());
+} else if (val.has_double_value()) {
+result = (0.0 != val.get_double_value());
+} else {
+result = ("" != val.get_value());
+}
+// If there is an EvalException conflict for this property, it is no longer applicable
+if ((0 != context.transaction) && (0 != context.node) && (0 != context.property)) {
+context.transaction->clear_conflicts(context.node, context.property,
+&CdlConflict_EvalExceptionBody::test);
+}
+} catch(CdlEvalException e) {
+if ((0 != context.transaction) && (0 != context.node) && (0 != context.property)) {
+CdlConflict conflict = context.transaction->get_conflict(context.node, context.property,
+&CdlConflict_EvalExceptionBody::test);
+if (0 == conflict) {
+CdlConflict_EvalExceptionBody::make(context.transaction, context.node, context.property, e.get_message());
+} else {
+CdlConflict_EvalException eval_conf = dynamic_cast<CdlConflict_EvalException>(conflict);
+CYG_ASSERTC(0 != eval_conf);
+if (eval_conf->get_explanation() != e.get_message()) {
+// Replace the conflict object. Higher level can detect this.
+context.transaction->clear_conflict(conflict);
+CdlConflict_EvalExceptionBody::make(context.transaction, context.node, context.property, e.get_message());
+}
+}
+throw;
+}
+}
+CYG_REPORT_RETURN();
+}
+//}}}

Mercurial > ecos

comparison host/libcdl/expr.cxx @ 76:435cced73e2f ecos-v1_3_1-release