Rascal: A DSL for SCAM

Transcription

1 Rascal: A DSL for SCAM Jurgen Vinju Tijs van der Storm Paul Klint Amsterdam, The Netherlands Hall of fame Bob Fuhrer Emilie Balland Arnold Lankamp Bas Basten

2 The complexity of bridging an analysis tool to a transformation tool shadows the complexity of the analyses and transformations themselves [Vinju & Cordy Dagstuhl 2005]

3 Every SCAM tool requires both analysis and transformation features

4 Refactoring Every SCAM tool requires both analysis and transformation features

5 Slicing Refactoring Every SCAM tool requires both analysis and transformation features

6 Slicing Refactoring Every SCAM tool requires both analysis and transformation features Reverse Engineering

7 Slicing Refactoring Every SCAM tool requires both analysis and transformation features Reverse Engineering Compilation

8 Restructuring Slicing Refactoring Every SCAM tool requires both analysis and transformation features Reverse Engineering Compilation

9 Restructuring Slicing Refactoring Every SCAM tool requires both analysis and transformation features Reverse Engineering Compilation Static checking

10 Most language parametric SCAM tools are geared towards either analysis or transformation

11 Most language parametric SCAM tools are geared towards either analysis or transformation TXL

12 Most language parametric SCAM tools are geared towards either analysis or transformation StrategoXT TXL

13 Crocopat Most language parametric SCAM tools are geared towards either analysis or transformation StrategoXT TXL

14 Crocopat GROK Most language parametric SCAM tools are geared towards either analysis or transformation StrategoXT TXL

15 Crocopat GROK Most language parametric SCAM tools are geared towards either analysis or transformation ASF+SDF StrategoXT TXL

16 Crocopat GROK RScript Most language parametric SCAM tools are geared towards either analysis or transformation ASF+SDF StrategoXT TXL

17 Many great SCAM tools are implemented in a GPL

18 Many great SCAM tools are implemented in a GPL A GPL allows fine-grained integration between analysis and transformation

19 Many great SCAM tools are implemented in a GPL A GPL allows fine-grained integration between analysis and transformation

20 A DSL for SCAM? That covers SCAM That scales down and scales up That is easier

21 The SCAM domain Transformation Source Code Extraction Generation Analysis Models Visualization Formalization Pictures Conversion

22 The SCAM domain Transformation Source Code Uses common... Data-structures Extraction Generation Algorithms Analysis Models Nothing new! But still.. Synthesize into a DSL Visualization Formalization Conceptually Pictures Conversion Syntactically Semantically

23 The SCAM domain Transformation Source Code Uses common... Data-structures Extraction Generation Algorithms Analysis Models Nothing new! But still.. Pattern Synthesize into a DSL Matching Visualization Formalization Conceptually Pictures Conversion Syntactically Semantically

24 Rascal has to scale down to make simple tasks short and easy to experiment with

25 Plain old REPL, no static type errors Rascal has to scale down to make simple tasks short and easy to experiment with

26 First regexps, Plain old REPL, then CFG s no static type errors Rascal has to scale down to make simple tasks short and easy to experiment with

27 First regexps, Plain old REPL, then CFG s no static type errors Rascal has to scale down to make simple tasks short and easy to experiment with Complete & domain specific, expression language: visit, pattern matching, relational calculus

28 Plain old REPL, no static type errors First regexps, then CFG s URI literals: files, projects, SVN,... Rascal has to scale down to make simple tasks short and easy to experiment with Complete & domain specific, expression language: visit, pattern matching, relational calculus

29 Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability

30 Modules are statically checked Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability

31 Modules are statically Immutable data checked Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability

32 Modules are statically checked Immutable data Co-variant sub-types Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability

33 Modules are statically checked Immutable data Co-variant sub-types h.o. parametric polymorphism, rank-1 and 2 Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability

34 Modules are statically checked Immutable data Co-variant sub-types h.o. parametric polymorphism, rank-1 and 2 Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability Efficient and fully typed (de)serialization

35 Modules are statically checked Immutable data Co-variant sub-types h.o. parametric polymorphism, rank-1 and 2 Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability Efficient and fully typed Functionlocal type inference (de)serialization

36 Modules are statically checked Immutable data Co-variant sub-types h.o. parametric polymorphism, rank-1 and 2 Rascal has to scale up to complex analysis and transformation algorithms that employ reuse of (library) functionality and allow design for maintainability Efficient and fully typed (de)serialization Functionlocal type inference lexically scoped backtracking

37 Modules are statically checked Immutable data Co-variant IDE sub-types h.o. parametric polymorphism, rank-1 and 2 bla Rascal has to scale up to complex analysis and bla support bla Concrete bla transformation algorithms that employ reuse of bla (library) bla functionality and allow design for syntax maintainability Visualizatio n support Efficient and fully typed (de)serialization Functionlocal type inference lexically scoped backtracking

38 Modules are statically checked Immutable data Co-variant IDE sub-types h.o. parametric polymorphism, rank-1 and 2 bla Rascal has to scale up to complex analysis and bla support bla Concrete bla transformation algorithms that employ reuse of bla (library) bla functionality and allow design for syntax maintainability Visualizatio n support Efficient and fully typed (de)serialization Functionlocal type inference lexically scoped backtracking

39 We have implemented Infer generic type arguments [Fuhrer et al. ECOOP2005] on Generic Featherweight Java [Igarashi et al. TOPLAS2002] The size of the Rascal code is equal to the size of the formal definitions in the papers

40 Rascal Analysis and manipulation are happily married. We should think of them as one.

41 Rascal Analysis and manipulation are happily married. We should think of them as one.