Uwe Kastens. Implementation of visit-oriented attribute evaluators. In H. Alblas and B. Melichar, editors, Attribute Grammars, Applications and Systems, volume 545 of Lecture Notes in Computer Science, pages 114--139. Springer-Verlag, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....beyond the storage of the attribute values themselves, as the remainder of its state can be maintained in a few global variables. In our evaluator, we use an alternative implementation strategy in which visits are mapped onto ordinary subroutine calls in the target language of our implementation [18, 49]. An evaluation plan can be decomposed into a series of segments, each terminated by a SUSPEND instruction, and representing the actions to be performed upon the i th visit. We construct a set of visit procedures for each AST operator, one for each segment of the evaluation plan. For visit i, the ....

Uwe Kastens. Implementation of visit-oriented attribute evaluators. In H. Alblas and B. Melichar, editors, Attribute Grammars, Applications and Systems, volume 545 of Lecture Notes in Computer Science, pages 114--139. Springer-Verlag, 1991.

....onto the concrete syntax of the corresponding production; this table is called concreteSyntax. The second data structure is a table of all the auxiliary and predicate functions. Lastly the main table produced by the parser is called bnfNames and contains information about nodes, visit sequences [15, 16], attributes, dependencies and many other details extracted from the AGDL specification. Using this information, a parser and attributed tree generator may be constructed for the specified language. 5.2 Verification The second phase of the generation process is verification. In this stage the ....

Kastens, U. "Implementation of VisitOriented Attribute Evaluators", Attribute Grammars, Applications and Systems, vol 545, Prague, Czechoslovakia, pp.114-139, 1991.

....references to data types representing the abstract syntax trees. Attribute instances needed in dioeerent traversals of the evaluator are passed between traversals as results arguments of partial parameterized visit functions. No additional data structures are required to handle them, like trees [Kas91b, PSV92, SKS97] or stacks and queues [AS91] The visit functions nd all the values they need in their arguments. Furthermore, the resulting evaluators are higher order attribute evaluators. The arguments of the evaluators visit functions are other visitfunctions. Finally, the evaluators have an eOEcient memory ....

Uwe Kastens. Implementation of Visit-Oriented Attribute Evaluators. In H. Alblas and B. Melichar, editors, International Summer School on Attribute Grammars, Applications and Systems, volume 545 of LNCS, pages 114139. Springer-Verlag, 1991.

.... has been refuted by the work of Augusteijn, who has built an attribute grammar evaluator based on lazy evaluation [1] Its performance is on a par with other systems that do a sophisticated analysis of dependencies, and produce a schedule for the attribute computations based on that analysis [18, 21]. Augusteijn s system, named Elegant, has been widely used within Philips for implementing domain specific languages. The second objection remains valid, however, and indeed the Elegant system suffers from this problem. Essentially, all attribute definitions have to be grouped by production. It is ....

U. Kastens. Implementation of visit-oriented attribute evaluators. In Proceedings of the International Summer School on Attribute Grammars, Applications and Systems, volume 545 of Lecture Notes in Computer Science, pages 114--139, 1991.

....projection of y onto the attributes of the left hand side nonterminal of production q is identical to w. That we can always find such an evaluation order is supported by the theorem discussed later in this section. Note that an evaluation order for a production corresponds to an evaluation plan [6, 7]. The D function in the ComputePlan algorithm is used to avoid repeated processing of a tuple. In Figure 5, the notation [D D(q , w) def ined ] denotes a new function obtained from D that is identical to the original D except that the new D(q , w) becomes def ined . The ComputePlan algorithm ....

....for a nonterminal node in the syntax tree, it will be used throughout the evaluation process. Hence, a top down traversal over the un evaluated syntax tree will select appropriate plans for the nonterminal nodes in the tree. Further evaluation may proceed as traditional visitoriented evaluators [6, 7]. Figure 6 is the evaluation algorithm for MAG . The evaluation algorithm first calls the traverse procedure to select a plan for each nonterminal node. The plan for the root node is G[q , p 1 , p 2 , p k ] where q is the topmost production instance (that is, the production instance ....

U. Kastens, Implementation of visit-oriented attribute evaluators, Proceedings of the International Summer School SAGA, (Prague, Czechoslovakia, June 1991), Lecture Notes in Computer Science 545 pp. 114-139 Springer-Verlag, (1991).

....0 :s : L 1 :s. Fig. 2 illustrates the dependencies and operation of the list reversal AG on a 2 element list; node names (u i and d j ) will be used in subsequent examples. To complete this introduction, we give in Fig. 3 an attribute evaluator for the list reversal AG, based on visit sequences [11]. Informally, the visit function means that the evaluator is recursively called on an other node. The visit name is indexed by the corresponding non terminal and the production name; when called on some node, the evaluator fetches which production is applied at it to select the corresponding visit ....

Kastens, U. Implementation of Visit-Oriented Attribute Evaluators. In Attribute Grammars, Applications and Systems, H. Alblas and B. Melichar, Eds. Lect. Notes in Comp. Sci., vol. 545, Springer-Verlag, New York--Heidelberg-- Berlin, June 1991, 114--139. INRIA Static and Dynamic Coupling Attribute Evaluators 13

....(since the number of attributes is usually tremendous) and (2) provide an incremental re evaluation after a tree transformation. As a consequence, most modern attribute systems (see Synthesizer Generator, FNC 2, Mercury, Mj lner Orm) include powerful memory management and incremental evaluation [17, 25, 29, 32, 33, 35, 39] and AGs have been repeatedly and successfully used in real applications such as construction of editors, translators, and compilers. On the other hand, the framework of Natural Semantics lacks an industrial strength implementation despite a large theoretical background (Natural Deduction [18] and ....

....very fine static analysis of the lifetime of each attribute instance, which in turn allows to determine the most efficient way to store it. This is possible because of the necessity to have a statically determinable total evaluation order to produce visit sequence based evaluators (as described in [29]) In [25] we give exact conditions to decide whether a temporary attribute can be stored in a global variable rather than in a stack (all temporary attributes can be, at worst, stored in a stack) and whether a non temporary attribute can be stored in a global variable. Finally, the evaluator ....

Kastens U. "Implementation of Visit-Oriented Attribute Evaluators", Proc. of "The International Summer School on Attribute Grammars, Applications & Systems", LNCS 545, 1991.

....the evaluator generator is the most intricate and most important part of an Attribute Grammar system. However, as of today, the only evaluation method for which we have studied in detail the feasibility and correctness of the iplane shiftj technique described above is the visit sequence paradigm [Kas80, Eng84, Kas91] applicable to l ordered AGs and, with a very slight extension, to strongly non circular AGs [Par88] Other evaluation methods will be studied in a forthcoming paper. The visit sequence is our preferred method because: these evaluators reach the best compromise between the time and ....

....generator is the most intricate and most important part of an Attribute Grammar system. However, as of today, the only evaluation method for which we have studied in detail the feasibility and correctness of the iplane shiftj technique described above is the visit sequence paradigm [Kas80, Eng84, Kas91] applicable to l ordered AGs and, with a very slight extension, to strongly non circular AGs [Par88] Other evaluation methods will be studied in a forthcoming paper. The visit sequence is our preferred method because: these evaluators reach the best compromise between the time and space ....

[Article contains additional citation context not shown here]

Uwe Kastens. Implementation of visit-oriented attribute evaluators. In Alblas and Melichar [AM91], pages 114139.

....code generation for each execution of a backward jump. Hence, we need a closer look at its costs. First of all we want to make sure, that no unnecessary computations are repeated during iteration. The evaluator generated by the LIGA system is a tree walking algorithm driven by visit sequences [4,5]. The visit sequences are computed such that they obey the specified dependencies between computations. The computations of the static semantics usually contribute only to the precondition of the iteration, and thus are not repeated in the cycle without further consideration in the specification. ....

....the overhead resulting from the tree walking interpretation. The time for executing source machine function bodies is productive interpretation costs. Overhead is caused by the calls of these functions and by the tree walk operations. Subtree visits are implemented by calls of recursive procedures [5]. The evaluation chain usually needs one visit per node. Hence, for the four nodes and three machine instructions of an assignment a = 1 within a statement sequence we pay the time of seven function calls overhead. That is about 1.4 microseconds out of 28 microseconds interpretation time for ....

Kastens, U., "Implementation of Visit-Oriented Attribute Evaluators," in Proceedings of the International Summer School on Attribute Grammars, Application and Systems, Lecture Notes in Computer Science #545, Springer Verlag, New York--Heidelberg--Berlin, 1991, 114--139.

No context found.

Kastens, U., "Implementation of Visit-Oriented Attribute Evaluators," in Proceedings of the International Summer School on Attribute Grammars, Application and Systems, Lect. Notes in Comp. Sci. #545, Springer Verlag, New York--Heidelberg--Berlin, 1991, 114--139.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC