| Andrew Tolmach and Dino P. Oliva. From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming, 8(4):367-412, July 1998. |
....not the CPS transform itself. Instead, we are interested in lifting the CPS transform from the level of expressions to the level of type derivations, preparing for building a type preserving compiler for Dependent ML. People are interested in type preserving compilation for various reasons [15, 14, 16, 8], including facilitating compiler debugging, optimizing data representation, leveraging language interoperability, producing certi ed low level executable code, etc. We are currently interested in using dependent types in DML to capture important program properties (e.g. memory safety) and then ....
Andrew Tolmach and Dino P. Oliva. From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming, 8(4):367-412, July 1998.
.... safety include various ones compiling Java into Java virtual machine language (JVML) Touchstone compiling Safe C into a form of proof carrying code (which we call TPCC) 9] TIL [11] and its successor TILT and FLINT ML [10] compiling SML [5] into a typed intermediate language [11] and ROML [12] compiling a restricted set of ML into a portion of C that is type safe. DTAL is an extension of TAL with dependent types, and it can be readily transformed into a TAL like language if one erases all syntax related to type index expressions. In this respect, DTAL generalizes TAL. In DTAL, ....
A. Tolmach and D. P. Oliva. From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming, 8(4):367-412, July 1998.
.... Safe C into a form of proof carrying code (which we call TPCC) Necula and Lee 1998] TIL and its successor TILT compiling Standard ML [Milner et al. 1997] into a typed intermediate language [Tarditi et al. 1996] and ROML compiling a restricted set of ML into a portion of C that is type safe [Tolmach and Oliva 1998]. Certi ed binary has a number of bene ts, including facilitating safe exchange of code in an untrusted environment and improving the robustness of a compiler (by thinking of each transformation phase as a separate certifying compiler whose subsequent stages may check compliance with safety ....
Tolmach, A. and D. P. Oliva (1998, July). From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming 8 (4), 367-412.
.... call TPCC) Necula and Lee 1998) TIL and its successor TILT compiling Standard ML (Milner, Tofte, Harper, and MacQueen 1997) into a typed intermediate language (Tarditi, Morrisett, Cheng, Stone, Harper, and Lee 1996) and ROML compiling a restricted set of ML into a portion of C that is type safe (Tolmach and Oliva 1998). Certifying compilers have a number of bene ts, including facilitating safe exchange of code in an untrusted environment and improving the robustness of a compiler (by thinking of each transformation phase as a separate certifying compiler whose subsequent stages may check compliance with some ....
Tolmach, A. and D. P. Oliva (1998, July). From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming 8 (4), 367-412.
.... call TPCC) Necula and Lee 1998) TIL and its successor TILT compiling Standard ML (Milner, Tofte, Harper, and MacQueen 1997) into a typed intermediate language (Tarditi, Morrisett, Cheng, Stone, Harper, and Lee 1996) and ROML compiling a restricted set of ML into a portion of C that is type safe (Tolmach and Oliva 1998). Certifying compilers have a number of benefits, including facilitating safe exchange of code in an untrusted environment and improving the robustness of a compiler (by thinking of each transformation phase as a separate certifying compiler whose subsequent stages may check compliance with some ....
Tolmach, A. and D. P. Oliva (1998, July). From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming 8 (4), 367--412.
.... call TPCC) Necula and Lee 1998) TIL and its successor TILT compiling Standard ML (Milner, Tofte, Harper, and MacQueen 1997) into a typed intermediate language (Tarditi, Morrisett, Cheng, Stone, Harper, and Lee 1996) and ROML compiling a restricted set of ML into a portion of C that is type safe (Tolmach and Oliva 1998). Certifying compilers have a number of benefits, including facilitating safe exchange of code in an untrusted environment and improving the robustness of a compiler (by thinking of each transformation phase as a separate certifying compiler whose subsequent stages may check compliance with some ....
Tolmach, A. and D. P. Oliva (1998, July). From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming 8 (4), 367--412.
....limit the utility of this approach is that we do not yet know the complexity of logical characterizations of the properties of interest. ffl Extend the prototype transformation system Promote to accommodate the core ML language. We shall do this by extending Promote to work on expressions in RML [TO97] which is a typed, intermediate language for de sugared ML similar to FLINT [Sha97, SA95] and TIL [TMC 96] ML programs can be parsed into RML, transformed, and rewritten as ML text or translated into C or Ada by template instantiation, which is a particular capability of the RML compiler. ....
Andrew Tolmach and Dino P. Oliva. From ML to Ada(!?!): Strongly-typed language interoperability via source translation. Journal of Functional Programming, (to appear), 1997.
....78153 Le Chesnay, France. E mail: Xavier.Leroy inria.fr. If only core ML polymorphism is considered, a simpler alternative to these unboxing strategies is monomorphisation (duplicating polymorphic functions once for each instantiation type to obtain a monomorphic program) Experimental evidence [4, 11] suggests that monomorphisation does not result in major increase in code size, even though it remains delicate to implement efficiently in a separate compilation context. But the real challenge is with the SML module system, especially functors and type abstraction in structures, which results in ....
D. P. Oliva and A. Tolmach. From ML to Ada(!?!): strongly-typed language interoperability via source translation. Draft, available electronically, Nov. 1996.
....the abstract machine became stable, this implementation was replaced by an implementation in a hybrid of the ML like target language and Ada. This reimplementation exploits specific mechanisms for developing such hybrid implementations in the Program Instantiator (also know as the RML compiler) [10]. 6.5 Subsequent hackery After some experience with the system, the naive way in which alternatives were considered was revisited. A frequent special case was a large number of alternatives being mapped onto a simple enumeration type. A safe transformation that improved performance in this case ....
Dino P. Oliva and Andrew Tolmach. From ML to Ada(!?!): Strongly-typed language interoperability via source translation, June 1997. Submited for publication.
.... analysis and transformation (but see Section 5) The approach shares much with the work of Bell, Bellegarde and Hook [4] who present 1 Our work in this area was originally motivated by the need to translate into strongly typed high level languages; the overall compiler context is described in [11]. 2 We assume a flat closure representation in this paper, but more elaborate representations can also be handled by our framework. Original code: fun outer (y:int,q:int list) let fun g1 (x:int) int = x y fun g2 (x:int) int = x 2 fun map (f:int int,l:int list) int list = case l ....
D. Oliva and A. Tolmach. From ML to Ada(!?!): Strongly-typed language interoperability via source translation. in preparation.
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