A syntax directed definition whose rules only print or emit output has
Answer: A
Such a definition is entirely about the order in which output appears, so it must be expressed as a translation scheme with the actions placed explicitly. A bare definition would leave the sequence of output undetermined.
Please log in or register to report this question.
State whether the following is true or false.
The parse tree is always constructed explicitly before attributes are evaluated.
Answer: B
Correct answer: False
For an S-attributed or L-attributed definition the attributes can be computed during the parse, with no tree ever built - the values live on a stack or in procedure parameters. The annotated tree is a way of describing what the definition means rather than an object the compiler must construct.
Please log in or register to report this question.
Attribute grammars as a formalism are more general than what compilers use because
Answer: C
The general formalism allows dependencies in any direction. Compilers restrict themselves to the acyclic S and L classes so that evaluation is guaranteed possible and cheap. The general formalism remains useful as a theoretical framework.
Please log in or register to report this question.
The chief practical benefit of expressing a translation as a syntax directed definition is
Answer: D
Reading the rules alongside the productions makes it plain what each construct means. The translation can be reviewed against the language definition production by production. That reviewability is worth a great deal on a compiler of any size.
Please log in or register to report this question.
A semantic rule that depends on the target machine belongs
Answer: A
Keeping machine dependence out of the front end is what allows one front end to serve several targets. Widths and alignments are the usual borderline cases and are normally parameterised. A clean separation is what allows one front end to serve several targets.
Please log in or register to report this question.
Field widths and alignments appearing in a front end translation are usually
Answer: B
A table of target properties keeps the rules themselves independent of any one machine. Hard coding the values would tie the front end to a particular architecture. The parameterisation is usually a small table consulted by the semantic rules.
Please log in or register to report this question.
A syntax directed translation for a control construct such as a while loop typically uses
Answer: C
The labels for the loop head and exit must be passed into the body while the generated code flows upward. The mixture is what makes such translations L-attributed rather than S-attributed. Almost every real control construct needs both directions of flow.
Please log in or register to report this question.
Generating a fresh label during translation is done by
Answer: D
A counter guarantees uniqueness. Labels are consumed by the jump instructions the translation emits and are resolved either immediately or by backpatching. The counter is reset for each procedure in some implementations.
Please log in or register to report this question.
The difference between emitting code as a side effect and accumulating it in an attribute is that the attribute approach
Answer: A
Holding code in attributes lets a rule place its children's fragments in any order. Emitting directly is cheaper but fixes the output order to the order of the actions. Which approach suits depends on whether the code must be rearranged later.
Please log in or register to report this question.
A syntax directed definition can be used to check a static semantic rule such as declaration before use because
Answer: B
The grammar cannot express the rule, but a semantic rule with access to the symbol table can check it. This is the standard division of labour between syntax and static semantics. The grammar handles form and the semantic rules handle everything else.
Please log in or register to report this question.
