This is a part of a grammar which implemented with YACC.
%right NOT
%left TIME DIV REALDIV MOD
%left LT LE
%left GT GE
%left EQ NE
%left AND OR
%right ASSIGN
%%
expression: expression EQ expression {};
|expression PLUS expression {};
| expression TIME expression{};
| PLUS expression {};
| variable PLUS PLUS {};
| variable {};
It has 9 shift reduce conflict. How to fix conflicts without changing the language? (the string which is generated with this grammar)
As written, the grammar is ambiguous because there is no way to distinguish the operator ++ from two instances of +.
Normally, this problem is resolved in the lexical scanner using the "maximal munch" rule, so that the expression a+++b would broken into the lexical items ID PLUSPLUS PLUS ID, resulting in the parse (a++) + b. In that case, if the user really meant a + (+ (+b))) and didn't want to use parentheses, they could simply split the tokens up with whitespace, writing a+ + +b. (In your grammar, <= is apparently scanned as a single lexical token, rather than the tokens < and =, so the inputs a < = b and a <= b result in different parses -- one of which is invalid. That's more or less expected. So there's no obvious reason why + + should be a permitted spelling of the increment operator ++.)
Making that change will change the language slightly. In your grammar, a++b can be interpreted as a+(+b); if ++ were lexically scanned as a single token, the resulting expression would be a syntax error. If that's a problem it can be resolved but I don't think the resulting complexity is worthwhile.
