A. D. Flatau. A veri ed implementation of an applicative language with dynamic storage allocation. Phd thesis, University of Texas at Austin, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... in this article we use a very small source language, and our target machine is unrealistically abstract, our work has a lot in common with former work on machine formalization and compiler veri cation using ACL2 and its predecessor Nqthm for Gypsy and Piton on the FM8502 and FM9001 processors ([8, 1, 9]) There is a subtle di erence in the notion of correct compilation, though, and we would like to point this out before we start. Our notion of correctness states that if execution of m, a compilation of program p, computes a non erroneous result, then execution of p will also compute the same ....

A. D. Flatau. A veri ed implementation of an applicative language with dynamic storage allocation. PhD thesis, University of Texas at Austin, 1992.

....state transformations resp. predicate transformers. Mechanical proofs are often based on interpreter semantics, a further variant of the operational style [67] and sometimes include high level compiler implementation veri cation (step 2) with e.g. HL = Stanford Pascal [62] or Boyer Moore Lisp [48, 14, 49] or Standard ML [11, 12] One should keep in mind, however, that the running theorem prover implementations are, strictly speaking, not completely veri ed. Their correctnesses again depend on existing correct initial host compilers, which are not available up to now. Recalling section 1, hackers ....

A. D. Flatau. A veri ed implementation of an applicative language with dynamic storage allocation. PhD thesis, University of Texas at Austin, 1992.

....resp. predicate transformers. Mechanical proofs are often based on interpreter semantics, a further variant of the operational style [Sto77] and sometimes include high level compiler implementation veri cation (step 2) with e.g. HL = Stanford Pascal [Pol81] or Boyer Moore Lisp [Moo88, Fla92, Moo96] or Standard ML [Cur93, Cur94] M. Broy [Bro92] uses the Larch prover [GG91] One should keep in mind, however, that the running theorem prover implementations are, strictly speaking, not completely veri ed. Their correctnesses again depend on existing correct initial host compilers, ....

A. D. Flatau. A veri ed implementation of an applicative language with dynamic storage allocation. PhD thesis, University of Texas at Austin, 1992.

....source program correctness is sucient, which means that we require a correct compiler to preserve partial program correctness [12, 30] There is a subtle di erence to speci cation re nement (like in VDM [22] or to former work on compiler veri cation using ACL2 resp. its predecessor Nqthm ([27, 8, 28]) J Moore proves in [28] that every non erroneous result of (the Piton machine on) will also be computed by m (on the FM9001) that m is more de ned than . This allows for optimizations, but trusted execution of m requires total correctness of . 4 Compiler Correctness for Transformational ....

A. D. Flatau. A veri ed implementation of an applicative language with dynamic storage allocation. PhD thesis, University of Texas at Austin, 1992.

.... MSL ( out . Since MSL ( j P Q exactly is partial correctness of w.r.t. P and Q, it also shows the idea of the proof of = in theorem 3 below Compared to speci cation re nement in VDM [Jon90] or to former work on compiler veri cation using ACL2 resp. its predecessor Nqthm ( Moo88, Fla92, Moo96] there is a subtle di erence to our notion of correct compilation: In [Moo96] for instance J Moore proves, that every non erroneous result of (the Piton machine on) will also be computed by m (on the FM9001) that m is more de ned than . This allows for optimizations, but trusted ....

A. D. Flatau. A veried implementation of an applicative language with dynamic storage allocation. PhD thesis, University of Texas at Austin, 1992.

....in objects and illustrate method resolution. The de nitive paper for describing how to model machines in the Boyer Moore tradition is [4] where Boyer and Moore describe their so called small machine model, which formed the basis of much early work on veri ed microprocessors and the CLI Stack [1, 12, 10, 22, 16]. M4 omits many features of the JVM. Among the more glaring omissions are accurate support for the JVM primitive data types like ints, doubles, arrays, oats, etc. support for syntactic typing both in the naming convention in the instruction set (e.g. IADD versus DADD) and eld and method ....

A. D. Flatau. A veried implementation of an applicative language with dynamic storage allocation. Phd thesis, University of Texas at Austin, 1992.

....insofar as possible. The techniques for modeling microprocessors and programming languages in such a logic have been developed over a long period of time in the Boyer Moore community. A tour de force of the method is presented in the so called CLI Stack (produced by Computational Logic, Inc. [1, 8, 18] which is a hierarchy of veri ed components including a microprocessor, loader, linker, assembler, two compilers, an operating system and some applications programs, all quite simple but also actually fabricated and practical, and all of which have been formally speci ed and mechanically proved ....

A. D. Flatau. A veried implementation of an applicative language with dynamic storage allocation. Phd thesis, University of Texas at Austin, 1992.

No context found.

A. D. Flatau. A veri ed implementation of an applicative language with dynamic storage allocation. Phd thesis, University of Texas at Austin, 1992.

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