Thomas J. Pennello, `Very fast LR parsing', Proceedings of the SIGPLAN '86 Symposium on Compiler Construction, SIGPLAN Notices, 21 (7), 145--151 (1986).  Home/Search   Document Not in Database   Summary   Related Articles   Check

....multiple dialects of a language for use in software reengineering [7] The primary objection to general parsing algorithms, then, is not one of functionality but of speed. For LR parsers, dramatic speed improvements have been obtained by producing hard coded, or directly executable parsers [3, 5, 14, 23, 24]. These directly executable LR parsers implement the parser for a given grammar as a specialized program, rather than using a typical table driven approach. In this paper we extend directly executable techniques for use in Earley s general parsing algorithm, to produce and evaluate what we believe ....

T. J. Pennello. Very Fast LR Parsing. Proceedings of the SIGPLAN '86 Symposium on Compiler Construction, SIGPLAN 21, 7 (1986), pp. 145-151.

....LF) characters to appear only at the end of a token and disallows the buffering of partial lines 5 . 3.3.2 Buffering RE2C generated scanners check if the buffer needs filling simply by comparing YYCURSOR and YYLIMIT. A method inspired by the mechanism used to guard against stack overflow in [17] 6 is used to reduce the amount of checking. Checks are only inserted in certain key states. These checks simply ensure that there is enough input in the buffer for the scan to proceed until the next key state. For example, in the DFA of Figure 3 it is sufficient to check that there are at ....

....from any state in the subset to any other. The set of key states consists of all of the states in non trivial SCCs, together with the start state. Note that for each SCC S, we actually only have to include a subset of states of S such that when the subset is removed, S becomes acyclic. Indeed, [17] describes a simple heuristic for choosing such a subset. However, since in practice most of the (non trivial) SCCs encountered will consist of a single state the current version of RE2C simply includes all states in non trivial SCCs 7 . An algorithm given in [3] was used to compute the SCCs. ....

Pennello, T. J. Very fast LR parsing. In Proceedings of the ACM SIGPLAN'86 Symposium on Compiler Construction (July 1986), ACM.

....tool such as yacc [Joh79] then generates an LR parse table for the grammar, assuming the grammar presents no conflicts. As discussed in Section 3, we optimize the tables produced by yacc, in a manner quite different from optimizations typically seen in the context of parsing programming languages [Pen86, HW90] The tables are in fact optimized for a modified LR parsing engine, which has the ability to skip tokens. It is this engine (driven by the optimized LR table) that acts as a packet filter, accepting packets conforming to the grammar specification. Filtering Mechanism Optimizer Network ....

....with warm cache. 5. Related Work In this section we describe related work in the areas of packet filters and LR parsing . We discuss four major packet filter schemes: CSPF [MRA87] BPF [MJ93] MPF [YBMM94] PathFinder [BGS 94] We also discuss related work in optimizing LR parsing [Pen86, HW90] 0 0.5 1 1.5 2 2.5 0 20 40 60 80 100 Filtering time (msecs) Number of endpoints BPF (warm cache) Grammar based filter (warm cache) BPF (cold cache) Grammar based filter (cold cache) Figure 12. Comparison between BPF and our Grammar based filter. Packet size (bytes) Time (msecs) 500 ....

[Article contains additional citation context not shown here]

T. J. Pennello. Very fast LR parsing. ACM SIGPLAN Notices, 21(7), 1986.

....produces ANSI C, not assembler. ffl mule accepts yacc bison input. ffl mule does yacc style error recovery. ffl mule manages a semantic stack for user code access (e.g. 1, 2, etc. 2. 1 Pennello Pennello created a parser generator that produced hard coded parsers in assembly language [Pen86]. His hard coded parsers showed a 6 to 10 fold improvement in speed over his table driven system. His system was not yacc compatible. Size increased by a factor of 2 to 4 enabling error recovery was responsible for the larger increases. Because the hard coded parsers directly manipulated jump ....

Thomas J. Pennello. Very fast LR parsing. In Proceedings of the SIGPLAN '86 Symposium on Compiler Construction, pages 145--151, 1986.

....and various other software tools that can be viewed as implementing the translation of a context free language. We are concentrating on the semantic phase of the translation process, rather than on scanning and parsing, since most modern compilers (should) spend relatively little time parsing [14]. In order to take advantage of the potential parallelism, we express the language translation process as an attribute grammar evaluation problem (see Section 2) We see three primary benefits to using attribute grammars: First, since attribute grammars provide a functional specification of the ....

....57 seconds including assembly. For proper comparison, parsing time must be added to the running time of our evaluators. Our parser takes about 3 seconds for the above program. A more efficient implementation could reduce parsing time significantly, for instance by using the techniques described in [14]. It must also be taken into consideration that our compiler only implements a subset of Pascal, although only two currently unimplemented language constructs appear to contribute to compilation time, namely operator overloading and proper treatment of write, writeln, and the like, which are ....

T. J. Pennello, "Very Fast LR Parsing", Proceedings of the SIGPLAN '86 Symposium on Compiler Construction, pp. 145-151 (1986).

....manipulation are not written with speed in mind. Using appropriate techniques [5] would result in further speed up. Therefore, even the time to generate the parsers is sufficient for practical application. 7 Related Work Of course, the pioneering work on functional LR parsing is due to Pennello [13]. He gives a low level implementation of a direct style functional parser. An overview of functional parsing can be found in the Leermakers book [11] A summary along with a description of the CPS based approach is given by the first author [15] He uses the CPS based parsing algorithm to ....

PENNELLO, T. Very fast LR parsing. SIGPLAN Notices 21, 7 (1986), 145--151.

No context found.

Thomas J. Pennello, `Very fast LR parsing', Proceedings of the SIGPLAN '86 Symposium on Compiler Construction, SIGPLAN Notices, 21 (7), 145--151 (1986).

No context found.

T. J. Pennello. Very Fast LR Parsing. In Proceedings SIGPLAN '86 Symposium on Compiler Construction,volume 21#7# of ACM SIGPLAN Notices, pages 145#151, 1986.

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