W. Vanhoof and B. Martens, "To parse or not to parse," in Logic Program Synthesis and Transformation., ed. N. Fuchs, LNCS 1463 (Springer-Verlag, 1998) 314--333.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... using various methods of program specialization (see [9, 31, 15] Partial deduction, a form of program specialization, has been used to reduce the interpretive overhead of metainterpreters, e.g. 48] Considerable success has been achieved and the refinement of these methods is an ongoing e#ort [11, 37, 55]. O#ine partial evaluation is wellknown for its power to substantially reduce interpretive overhead, e.g. 10, 12, 33, 34] and has been used successfully to optimize two levels of interpretation, e.g. 35] Recent work [13] applies o#ine partial evaluation to interpreters of domain specific ....

W. Vanhoof and B. Martens, "To parse or not to parse," in Logic Program Synthesis and Transformation., ed. N. Fuchs, LNCS 1463 (Springer-Verlag, 1998) 314--333.

....in the next section. However, it has been felt by several researchers that well founded orders are sometimes too rigid or (conceptually) too complex in an online setting. Recently, well quasi orders have therefore gained popularity to ensure online termination of program manipulation techniques [5, 59, 61, 38, 39, 14, 27, 2, 29, 1, 65]. Unfortunately, this move to well quasi orders has never been formally justified nor has the relation to well founded approaches been investigated. We strive to do so in the first part of this paper and will actually prove that a rather simple well quasi approach the homeomorphic embedding ....

....which enabled us to ignore arguments to predicates in e.g. Proposition 2. 12 of an expression can be beneficial for practical programs. But, as we have seen earlier, Theta alone is already very flexible for metainterpreters, even more so when combined with characteristic trees [39] see also [65]) We will return to the issue of metaprogramming in Section 7. Of course, for any wfo (monotonic or not) one can devise a wbr (cf. Lemma 1) which has the same admissible sequences. Still there are some feats that are easily attained, even by using Theta, but which cannot be achieved by a wfo ....

[Article contains additional citation context not shown here]

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs, editor, Proceedings of the International Workshop on Logic Program Synthesis and Transformation (LOPSTR'97), LNCS 1463, Leuven, Belgium, July 1997. 27

....rule, program points as abstraction to guide the local control of our specialisers, and tuple distributivity of the upper bound of RUL constraints. All these features combined together prevent specialisation of interpreters for languages with parameters and block declarations to parse away [VM97] the interpretation layer thus producing small programs useful for analysis and specialisation as we have shown. The problems can be summarised in one example. Assume we have an imperative program with procedures. A quick look at the list of visible variables during execution of this program with ....

Wim Vanhoof and Bern Martens. To parse or not to parse. In N. Fuchs, editor, Logic Program Synthesis and Transformation (LOPSTR'97), pages 322-342. Springer-Verlag, LNCS 1463, 1997.

....encoding of terms. Similar problems occur with other operations popular with online transformers. Existing transformers are only geared toward the direct transformation of programs. They often fail to perform well on an extra metasystem level, so that several refinements have been suggested, e.g. [29]. For a discussion of the invariance problem see also [17] Higher order terms provide a suitable data structure for representing programs in a higher order abstract syntax. Higher order logic programming utilizes higher order logic as the basis for computation, e.g. 20] The primary advantage of ....

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs (ed.), Logic Program Synthesis and Transformation (LOPSTR'97), Lecture Notes in Computer Science, Vol. 1463, pages 314--333, Springer-Verlag 1998.

....the third argument corresponds to code produced at the next level, i.e. at the level of the specialised program. 3. 5 An Example We now show that our cogen approach is actually powerful enough to satisfactorily specialise the vanilla metainterpreter (a task which has attracted a lot of attention [14, 54, 71] and is far from trivial) Example 4. The following is the well known vanilla metainterpreter for the nonground representation: demo(true) demo( P Q) demo(P) demo(Q) demo(A) dclause(A,Body) demo(Body) dclause(append( L,L) true) dclause(append( H X] Y, H Z] append(X,Y,Z) ....

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs, editor, Proceedings of the International Workshop on Logic Program Synthesis and Transformation (LOPSTR'97), LNCS 1463, pages 322--342, Leuven, Belgium, July 1997.

....Embedding It turns out that, for the online approach, well founded orders as used in Sections 6.1 and 6.2 are sometimes too rigid or (conceptually) too complex. Recently, well quasi orders have therefore gained popularity to ensure online termination of program manipulation techniques ([5, 42, 43, 21, 24, 31, 1, 26, 45, 32]) Definition 1 (well quasi order) An ordered set S( is called well quasi ordered iff for any infinite sequence e 1 ; e 2 ; of elements of S, there exist elements e i and e j with i j such that e i e j . 2 The additional power of well quasi orders stems from the fact that ....

W. Vanhoof and B. Martens. To parse or not to parse. pages 322--342. Also as Technical Report CW 251, K.U.Leuven. Benchmark Original sonic + ecce ecce mixtus

....in the next section. However, it has been felt by several researchers that well founded orders are sometimes too rigid or (conceptually) too complex in an online setting. Recently, well quasi orders have therefore gained popularity to ensure online termination of program manipulation techniques [6, 69, 72, 46, 47, 21, 35, 2, 37, 1, 76, 1, 13]. Unfortunately, this move to well quasi orders has never been formally justified nor has the relation to well founded approaches been investigated. We strive to do so in the first part of this paper and will actually prove that a rather simple well quasi approach the homeomorphic embedding ....

....wqo based approaches and whether the ability to completely ignore certain parts of an expression can be beneficial for practical programs. But, as we have seen earlier, Theta alone is already very flexible for metainterpreters, even more so when combined with characteristic trees [47] see also [76]) We will return to the issue of metaprogramming in Section 7. Of course, for any wfo (monotonic or not) one can devise a wbr (cf. Lemma 3.1) which has the same admissible sequences. Still there are some feats that are easily attained, even by using Theta, but which cannot be achieved by a wfo ....

[Article contains additional citation context not shown here]

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs, editor, Proceedings of the International Workshop on Logic Program Synthesis and Transformation (LOPSTR'97), LNCS 1463, pages 322--342, Leuven, Belgium, July 1997.

....built. It is this SLD tree that decides what the new predicate will look like, and, more importantly, what structure it will handle. Although both issues are quite related, it has not yet been fully sorted out how to exploit their relationship during partial deduction. ffl As was pointed out in [16], in the context of specialising meta interpreters (and presumably other programs handling internal structure) the decision where U should stop, should be influenced by the consequences this decision has for the generated code. 16] makes a first attempt to achieve this by using information from ....

....during partial deduction. ffl As was pointed out in [16] in the context of specialising meta interpreters (and presumably other programs handling internal structure) the decision where U should stop, should be influenced by the consequences this decision has for the generated code. [16] makes a first attempt to achieve this by using information from the global level: when unfolding stops at an atom that causes a generalisation, 16] will unfold this atom carefully further to see whether this unfolding results in a set of atoms for which generalisation is no longer necessary or ....

[Article contains additional citation context not shown here]

W. Vanhoof and B. Martens. To parse or not to parse. Technical Report CW251, Departement Computerwetenschappen, K.U.Leuven, Belgium, June 1997. Appeared in the pre-proceedings of LOPSTR'97.

....with the ancestor atoms from which the current atom descends (via resolution) However, in an on line setting, well founded orders are sometimes too rigid or too complex. Recently, well quasi orders have therefore gained popularity to ensure on line termination of program manipulation techniques [3, 41, 43, 26, 27, 13, 18, 1, 20, 46]. Indeed, as we will see below, well quasi orders are often much more flexible than well founded orders in an on line context. We start examining them in the next section. 2 Well quasi orders and homeomorphic embedding From now on, we suppose familiarity with basic notions in logic programming ....

....have to be taken into account (see e.g. the experiments in [24, 27, 18] For some applications, Theta as well as Theta and Theta remain too restrictive. In particular, they do not always deal satisfactorily with fluctuating structure (arising e.g. for certain metainterpretation tasks) [46]. The use of characteristic trees [24, 27] remedies this problem to some extent, but not totally. A further step towards a solution is presented in [46] In that light, it might be of interest to study whether the extensions of the homeomorphic embedding relation proposed in [39] and [22] in the ....

[Article contains additional citation context not shown here]

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs, editor, Pre-Proceedings of the International Workshop on Logic Program Synthesis and Transformation (LOPSTR'97), Leuven, Belgium, July 1997. Also as Technical Report CW 251, K.U.Leuven.

....in a general and completely automatic way, however, is far from trivial since control information, needed by the specialiser to decide whether or not to continue the specialisation, might also flow bottom up. In particular, Vanilla like) meta programs typically present difficulties of this kind [20]. Consider the following example, taken from [21] The predicate make list(T; I; R) can be used to create a list of a fixed length (type T ) with each element initialised with I. The result is returned in R. Example 1. f ill list(L; T; I; L) type(T; L) type(list1; X] f ill list(L; T; I; ....

.... is capable of specialising the Vanilla meta interpreter satisfactorily. Equally good results can be obtained by a top down control strategy alone, but often at the cost of not completely general and or automatic techniques (e.g. 12] or a non trivial and complex control mechanism. In [20], a sophisticated approach is described based on extending a local unfolding rule by global information. Although the resulting technique is completely general and automatic, in spite of its complexity, it turns out to be not sufficiently effective when more involved meta interpreters are at hand. ....

W. Vanhoof and B. Martens. To parse or not to parse. In N. E. Fuchs, editor, Proceedings of the Seventh International Workshop on Logic Program Synthesis and Transformation LOPSTR '97, volume 1463 of Lecture Notes in Computer Science, pages 314 -- 333, Leuven, Belgium, 1997. Springer-Verlag.

....it in a general and completely automatic way, however, is far from trivial, since control information, needed by the specialiser to decide whether or not to continue the specialisation, might also ow bottom up. In particular, Vanilla like) metaprograms typically present diculties of this kind [VM97] Consider the following example: The predicate make list(T; I; R) can be used to create a list of a xed length (type T ) with each element initialised with I. The result is returned in R. Example 2 We have f ill list(L; T; I; L) type(T; L) type(list1; X] f ill list(L; T; I; R) f ill ....

....excellent specialisation in the case of the Vanilla metainterpreter can be achieved. Equally good results can be obtained by a top down control strategy alone, but often at the cost of not completely general or automatic techniques (e.g. LS90] or a nontrivial and complex control mechanism. In [VM97] a sophisticated approach is described based on extending a local unfolding rule by global information. Although the resulting technique is completely general and automatic, in spite of its complexity, it turns out to be not suciently e ective 28 when more involved metainterpreters are at ....

W. Vanhoof and B. Martens. To parse or not to parse. In N. E. Fuchs, editor, Proceedings of the Seventh International Workshop on Logic Program Synthesis and Transformation LOPSTR '97, volume 1463 of Lecture Notes in Computer Science, pages 314-333, Leuven, Belgium, 1997. Springer-Verlag.

....in a general and completely automatic way, however, is far from trivial since control information, needed by the specialiser to decide whether or not to continue the specialisation, might also flow bottom up. In particular, Vanilla like) meta programs typically present difficulties of this kind [11]. Consider the following example, taken from [12] The predicate make list(T; I; R) can be used to create a list of a fixed length (type T ) with each element initialised with I. The result is returned in R. Example 1. f ill list(L; T; I; L) type(T; L) type(list1; X] f ill list(L; T; I; ....

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs, editor, Proceedings of LOPSTR'97, Leuven, Belgium, July 1997. To appear in Lecuture Notes in Computer Science series (Springer-Verlag).

....in this query reflects the fact that the necessary information for specialisation is supplied elsewhere. Due to the simplicity of the example, most top down specialisers will not have any difficulties in obtaining the required specialisation. For Example 2.2 this is no longer true. As shown in [22], fully general and automatic top down specialisation of the vanilla meta interpreter requires a sophisticated control mechanism with interaction between the local and global control. Apart from the fact that using an uninstantiated query to achieve specialisation with a system tuned to exploit ....

....results in conceptually cleaner, less complicated techniques with which at least the same results can be obtained as with one overall approach, in general requiring a much more complicated control. A concrete example is the specialisation of Vanilla like meta interpreters. As was demonstrated in [22], specialisation of non ground meta interpreters using completely general and fully automatic top down techniques still is a non trivial task. In that paper, a sophisticated approach is described based on extending a local unfolding rule by global information. Although the resulting technique is ....

W. Vanhoof and B. Martens. To parse or not to parse. In N. Fuchs, editor, Proceedings of LOPSTR'97, Leuven, Belgium, July 1997. To appear in Lecuture Notes in Computer Science series (Springer-Verlag).

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