- template-haskell - Haskell.org

TH type not Typeable?
by Mike Gunter 10 Jun '05

10 Jun '05

Is the lack of Typeable instances for the TH types founded in some difficulty in providing them? Or something else? mike

3 3

TH with implicit parameters?
by WANG Meng 07 Jun '05

07 Jun '05

Hi All, Does TH currently support implicit parameters? Is it possible to write a TH funcition which can be evaluated into a Haskell expression which use implicit parameters? -W-M- @ @ | \_/

1 0

Call for Participation: GPCE'05 - Generative Programming and Component Engineering
by Eelco Visser 31 May '05

31 May '05

PRELIMINARY CALL for PARTICIPANTS ANNOUNCEMENT of WORKSHOPS and TUTORIALS ---------------------------------------------------------------------- 4th International Conference on Generative Programming and Component Engineering (GPCE'05) Sep 29 - Oct 1, 2005, Tallinn (Estonia) http://www.gpce.org/05 Sponsored by ACM SIGPLAN, in cooperation with ACM SIGSOFT co-located with ICFP'05 and TFP'05 ---------------------------------------------------------------------- Consult http://www.gpce.org/05 for UP-TO-DATE and DETAILED information especially the calls for workshop contributions FORTHCOMING: on-line registration should become active in mid June ---------------------------------------------------------------------- IMPORTANT DATES * Jun 13, 2005: suggested deadline for workshop contributions (please check on workshop web pages for details) * Jul 29, 2005: EARLY REGISTRATION DEADLINE * Sep 27-28, 2005: GPCE workshops and tutorials * Sep 29 - Oct 1, 2005: GPCE papers and demos ---------------------------------------------------------------------- SCOPE. Generative and component approaches have the potential to revolutionize software development in a similar way as automation and components revolutionized manufacturing. Generative Programming (developing programs that synthesize other programs), Component Engineering (raising the level of modularization and analysis in application design), and Domain-Specific Languages (elevating program specifications to compact domain-specific notations that are easier to write and maintain) are key technologies for automating program development. GPCE arose as a joint conference, merging the conference on Generative and Component-Based Software Engineering (GCSE) and the workshop on Semantics, Applications, and Implementation of Program Generation (SAIG). The goal of GPCE is to provide a meeting place for researchers and practitioners interested in cutting edge approaches to software development. We aim to foster further cross-fertilization between the software engineering research community and the programming languages community, in addition to supporting the original research goals of both the GCSE and the SAIG communities. ---------------------------------------------------------------------- 3 INVITED SPEAKERS: * Oscar Nierstrasz: Object-oriented reengineering patterns * Oege de Moor: TBA * Bernd Fischer: Certifiable generative programming 25 TECHNICAL PAPERS (see conference web page for the list of accepted papers) 2 DEMONSTRATIONS: * Developing Dynamic and Adaptable Applications with CAM/DAOP: a Virtual Office Application * Metamodeling made easy - MetaEdit+ 3 TUTORIALS T1: Multi-stage Programming in MetaOCaml (Presenters: W.Taha, C.Calcagno) T2: Implementing Domain-Specific Modelling Languages and Generators (Presenters: R.Pohjonen and J-P.Tolvanen) T3: Challenges and Best Practices of Generative Software Engineering in the Context of Large Complex Business Applications (Presenter: M.M.Davydov) 3 WORKSHOPS W1: Seventh Young Researchers Workshop (Organizers: D.R.Dechow, D.Foetsch, S.Kiebusch, S.Perugini, M.J.Rutherford, D.Shestakov) W2: Second MetaOCaml Workshop (Organizers: K.Swadi, W.Taha) W3: Graph and Model Transformations Workshop (Organizers: G.Karsai, G.Taentzer) ---------------------------------------------------------------------- PROGRAM COMMITTEE Chairs: * Robert Glück (U. of Copenhagen) * Michael Lowry (NASA) Members: * Don Batory (U. of Texas, USA) * Ira Baxter (Semantic Designs) * Cristiano Calcagno (Imperial College) * Prem Devanbu (U. of California at Davis) * Ulrich Eisenecker (U. of Leipzig) * Tom Ellman (Vassar College) * Robert Filman (NASA) * Zhenjiang Hu (U. of Tokyo) * Patricia Johann (Rutgers U.) * John Launchbury (Galois) * Anne-Françoise Le Meur (U. of Sci. and Tech. Lille) * Hong Mei (Peking U.) * Nicolas Rouquette (NASA) * William Scherlis (CMU) * Yannis Smaragdakis (Georgia Inst. of Tech.) * Walid Taha (Rice U.) * Todd Veldhuizen (Chalmers U. of Tech.) ORGANIZING COMMITTEE General Chair * Eugenio Moggi (Genova U.) Publicity Chair: * Eelco Visser (Utrecht U.) Workshops and Tutorials Chairs * Jeff Gray (U. of Alabama at Birmingham) * Andrew Malton (Waterloo U.) Local Arrangements Chair * Tarmo Uustalu (Inst. of Cybernetics, Tallinn) ----------------------------------------------------------------------

1 0

Types as first-class values
by Vivian McPhail 24 May '05

24 May '05

Hi, I have run into problems trying to implement a program, and think that I might be traipsing along the edge of ghc design. I have a NL system which uses a lexicon of words annotated with types and semantics. The grammer uses the type information to correctly structure the semantic nodes. The grammer uses simple combinators to manipulate the semantics, thus I have implemented my system as an untyped lambda calculus extended with a single data type. So I have > data Term a = Var String | Lam Var (Term a) | App (Term a) (App Term a) | Sem a -- Dynamic a A correct sentence will yield a term with only App and Sem. Because I have used the (user-supplied) type information, I know that application of Sem to Sem will be correct (given correct type information). Thus I could have a term > App (Sem f) (Sem g) where f is of type: s / np and g is of type: np I want to harness the power of haskell in my semantic terms, so currently I am implementing the member of (Sem a) in haskell and loading dynamically with hs-plugins. I have a type > data SemF a = ZSem (SST a) > | USem (SST a -> SST a) etc... This works well, except for the fact that in english some words take up to five arguments, and there are any number of possible combinations of argument types, so a particular word such as probably probably = (s\np)/(s\np) has a type (* -> *) -> (* -> *) so I need a separate data constructor > | U22Sem ((SST a -> SST a) -> (SST a -> SST a)) There are a huge number of possible types, and doing it this way requires instantiating a data constructor for each type. Modification 1: Someone suggested I use GADTs, so I tried implementing with: > data SemG where > ZSemG :: SST SNode -> SemG > USemG :: (Typeable b) => (b -> SST SNode) -> SemG (so USem represents functions of type * -> *, (* -> *) -> *, ... but when I go to use the function I can't, because trying to cast gives an ambiguous type variable (I think the technical term is monomorphic restriction): applySem :: SemG -> SemG -> SemG applySem (USemG a) (ZSemG b) = ZSemG $ ((unJust $ cast a) :: SST SNode -> SST SNode) ((unJust $ cast b) :: SST SNode) applySem (BSemG a) (ZSemG b) = USemG $ ((unJust $ cast a) :: Typeable a => SST SNode -> a -> SST SNode) -- this a here is problematic ((unJust $ cast b) :: SST SNode) This code (applySem) is compiled and linked, whereas the functions which are the data in this case are compiled and loaded at run-time. This is frustrating, I want to harness ghcs power without reimplementing. (The reason I want to use haskell in the semantic terms is that I need access to IORefs and other features which would be non-trivial to implement by hand in a lambda interpreter). Why reinvent the wheel? First, am I missing something? If not... Suggestion. The problem occurs at the location of the cast, I am applying two terms, I know they are correct, given that the user supplied syntactic type is correct, but to give each a type annotation by hand in the applySem code would require, again, > 5^4 possibilities. The type information of the semantic term is available at compile time (which in this case is while the main program is running). What I propose is this: two additions, one to template haskell, the other to the language as a primitive. 1) The ability in Template Haskell to provide the type of an expression as a first class value. This would require a hook into the compiler and datatypes to express the type. 2) a castWithType function which, given a type expression and a (Typeable a) expression will attempt to perform the cast to (Maybe a). This would solve my problem, because then when I compile the semantic expressions (while the main program is running) I can include their types as part of the expression. Then, when I go to use the semantic term, I can give the castWithType function the type information and the function to cast. I realise that this is in essence what the typeable class is supposed to provide. My problem is that I need a 'general' casting code, (part of the main run-time) which is capable of applying semantic functions with many possible different types. In summary, in order to avoid hand coding thousands of possible function types, I want to use dynamic casts. I have tried GADTs (which I may not have understood correctly :-). It appears that what I want to do is not possible at the moment. I think that this modification is worthwhile. The reflection abilities it provides (along with Template Haskell) would allow extension of Haskell as opposed to reimplementation. Cheers, Vivian P.S. If I have missed something, please could you let me know, I'm stumped. In which case I apologise for the rant.

3 2

RE: [Template-haskell] TH bug finally
by Simon Peyton-Jones 03 May '05

03 May '05

| ghc.EXE: panic! (the `impossible' happened, GHC version 6.4): | plusProv SpliceCG SpliceCG Good point. I've fixed this, in the HEAD and 6.4.1. Thanks for pointing it out. Simon | -----Original Message----- | From: template-haskell-bounces(a)haskell.org [mailto:template-haskell-bounces@haskell.org] On | Behalf Of kyra | Sent: 23 April 2005 03:03 | To: template-haskell(a)haskell.org | Subject: [Template-haskell] TH bug finally | | I realize that my previous TH programs were incorrect. | | But my compiant is about GHC not reporting something like "Multiple | declarations..." but panics or plainly crashes. | | Sorry. | | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

1 0

Profiling with Template Haskell, again
by Andre Pang 25 Apr '05

25 Apr '05

Hi all, I've run into the same problem as Sean Seefried, namely that I'd like to profile executables built with Template Haskell: http://www.haskell.org/pipermail/template-haskell/2003-July/000135.html (Specifically, profiling the final executable, not the splicing.) I guess this is just a simple message to say that there's someone else asking for profiling support with TH now, in the hopes that it'll bump up its priority on the ever-growing to-do list :). Thanks, -- % Andre Pang : trust.in.love.to.save <http://www.algorithm.com.au/>

1 0

TH bug finally
by kyra 23 Apr '05

23 Apr '05

I realize that my previous TH programs were incorrect. But my compiant is about GHC not reporting something like "Multiple declarations..." but panics or plainly crashes. Sorry.

1 0

TH bug again (GHC 6.4)
by kyra 23 Apr '05

23 Apr '05

If I simply try the following: <genCG.hs> {-# OPTIONS -fglasgow-exts #-} module GenCG where gdecls pbname pbniname = [d| valGetter pbname = "Get" ++ pbniname valSetter pbname = "Set" ++ pbniname fldGetter pbname = "Get" ++ pbniname ++ "Field" fldSetter pbname = "Set" ++ pbniname ++ "Field" arrGetter pbname = "Get" ++ pbniname ++ "ArrayItem" arrSetter pbname = "Set" ++ pbniname ++ "ArrayItem" gvarGetter pbname = "Get" ++ pbniname ++ "GlobalVar" gvarSetter pbname = "Set" ++ pbniname ++ "GlobalVar" svarGetter pbname = "Get" ++ pbniname ++ "SharedVar" svarSetter pbname = "Set" ++ pbniname ++ "SharedVar" |] </genCG.hs> and <spliceCG.hs> {-# OPTIONS -fglasgow-exts #-} module SpliceCG where import GenCG $(gdecls "integer" "Int") $(gdecls "unsignedinteger" "Uint") </spliceCG.hs> GHC 6.4 simply crashes with the memory access error - GHC (windows) reports: "The instruction at "0x00b9ad27" referenced memory at "0xff047687". The memory could not be "read". And again: <spliceCG.hs> {-# OPTIONS -fglasgow-exts #-} module SpliceCG where import GenCG $(gdecls "integer" [| "Int" |]) </spliceCG.hs> compiles just fine.

1 0

TH bug reported (GHC 6.4)
by kyra 23 Apr '05

23 Apr '05

I have the following Haskell modules: <genCG.hs> {-# OPTIONS -fglasgow-exts #-} module GenCG where gdecls pbname pbniname = [d| valGetter pbname = "Get" ++ $pbniname valSetter pbname = "Set" ++ $pbniname fldGetter pbname = "Get" ++ $pbniname ++ "Field" fldSetter pbname = "Set" ++ $pbniname ++ "Field" arrGetter pbname = "Get" ++ $pbniname ++ "ArrayItem" arrSetter pbname = "Set" ++ $pbniname ++ "ArrayItem" gvarGetter pbname = "Get" ++ $pbniname ++ "GlobalVar" gvarSetter pbname = "Set" ++ $pbniname ++ "GlobalVar" svarGetter pbname = "Get" ++ $pbniname ++ "SharedVar" svarSetter pbname = "Set" ++ $pbniname ++ "SharedVar" |] </genCG.hs> and <spliceCG.hs> {-# OPTIONS -fglasgow-exts #-} module SpliceCG where import GenCG $(gdecls "integer" [| "Int" |]) $(gdecls "unsignedinteger" [| "Uint" |]) </spliceCG.hs> Trying to compile 'spliceCG.hs' I'm recieving the following: Loading package base-1.0 ... linking ... done. Loading package haskell98-1.0 ... linking ... done. Loading package template-haskell-1.0 ... linking ... done. ghc.EXE: panic! (the `impossible' happened, GHC version 6.4): plusProv SpliceCG SpliceCG Please report it as a compiler bug to glasgow-haskell-bugs(a)haskell.org, or http://sourceforge.net/projects/ghc/. If I splice in a single declaration, for example: <spliceCG.hs> {-# OPTIONS -fglasgow-exts #-} module SpliceCG where import GenCG $(gdecls "integer" [| "Int" |]) </spliceCG.hs> it is compiled just fine.

1 0

RE: [Template-haskell] not an instance of Lift?
by Simon Peyton-Jones 05 Apr '05

05 Apr '05

Yes you can make new instances of Lift. The class itself is defined in Language.Haskell.TH.Syntax. Simon | -----Original Message----- | From: template-haskell-bounces(a)haskell.org [mailto:template-haskell-bounces@haskell.org] On | Behalf Of WANG Meng | Sent: 05 April 2005 05:20 | To: template-haskell(a)haskell.org | Subject: [Template-haskell] not an instance of Lift? | | Hi All, | | When I am trying to define the following program: | | test x = [|x|] | y = test (Just True) | | The compiler will complain that Maybe Bool is not an instance of Lift. | | Is there any way to resolve this? Can I add instances to the class Lift? | | | -W-M- | @ @ | | | \_/ | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

1 0