Exercise 4.

Some languages such as PL1 permit a list of names to be given a list of attributes. The grammar G below abstracts the problem.

D	$$\longmapsto$$	namelist  attriblist$$\bf ;$$
namelist	$$\longmapsto$$	$$\bf id$$, namelist
namelist	$$\longmapsto$$	$$\bf id$$
attriblist	$$\longmapsto$$	A  attriblist
attriblist	$$\longmapsto$$	A
A	$$\longmapsto$$	$$\bf long$$
A	$$\longmapsto$$	$$\bf double$$
A	$$\longmapsto$$	$$\bf single$$
A	$$\longmapsto$$	$$\bf integer$$
A	$$\longmapsto$$	$$\bf float$$

where D, namelist, attriblist and A are nonterminals and where id, long, double, single, integer, float and ; are terminals. As usual id is the identifier token. For instance each of the following lines is a declaration generated by G.

i, j single integer;
u, v, w double float;

Give a syntax-directed definition for declarations generated by G. Here are some hints.

• You may simply write N and T instead of namelist and attriblist respectively.
• The code in the actions does not need to be C code but may be any pseudo-code that is convenient to you.
• You may consider that id has an attribute $\bf id$.name (as usual) giving the lexical value of id.
• Moreover you may assume that every variable in the symbol table may be associated with a list of attributes by an instruction like enter( $\bf id$.name, T.list)
• The instruction T.list := makelist(A.attrib) will create the list consisting of the single item A.attrib.
• The instruction T.list := cons(A.attrib, T.list) will add the element A.attrib at the beginning of the list T.list.

Answer 4