February 2004 - Glasgow-haskell-users

RE: Generics...
by Simon Peyton-Jones 06 Feb '04

06 Feb '04

I'm sorry, but I still can't figure out your specification. Given | type SqlColumn a = String | type TypedExpr a = String then the identity function will convert one to the other. And if they are different types, then my original observation still applies: I assume you aren't expecting just to replace all Ints by Floats, because that can't be well typed. To put it another way, you want to write a *generic* function; that is, one that works on arguments of many types. So what is the type that you want your function to have? For generic programming you want one type, like this gsize :: Data a => a -> Int gsize takes a data structure a and returns its size. You give the type of 'project' by giving several examples. So maybe you want a type class, rather than generic functions? I'm confused anyway. SImon | -----Original Message----- | From: glasgow-haskell-users-bounces(a)haskell.org [mailto:glasgow-haskell-users- | bounces(a)haskell.org] On Behalf Of MR K P SCHUPKE | Sent: 05 February 2004 16:01 | To: glasgow-haskell-users(a)haskell.org; k.schupke(a)imperial.ac.uk; Simon Peyton-Jones | Subject: RE: Generics... | | > Presumably not "replace all Ints by Floats" | | I've thought about it some more, and I don't think its | possible to do what I want using generics... Ideally I would like | to achieve the following type transformation: | | type SqlColumn a = String | type TypedExpr a = Int | | the function would map SqlColumns to TypedExpr... The actual code | generates an integer from a sequence to "tag" each column, but for | the sake of simplicity, the following types would be sufficient: | | type SqlColumn a = String | type TypedExpr a = String | | So we take a tuple of any length and return exactly the same | tuple, but with the types changed: | | project :: (SqlColumn a,SqlColumn b) -> (TypedExpr a,TypedExpr b) | | project :: (SqlColumn a,SqlColumn b,SqlColumn c) -> (TypedExpr a,TypedExpr b,TypedExpr c) | | | What do you think? | | Regards, | Keean Schupke. | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users

1 0

RE: Generics...
by MR K P SCHUPKE 05 Feb '04

05 Feb '04

> Presumably not "replace all Ints by Floats" I've thought about it some more, and I don't think its possible to do what I want using generics... Ideally I would like to achieve the following type transformation: type SqlColumn a = String type TypedExpr a = Int the function would map SqlColumns to TypedExpr... The actual code generates an integer from a sequence to "tag" each column, but for the sake of simplicity, the following types would be sufficient: type SqlColumn a = String type TypedExpr a = String So we take a tuple of any length and return exactly the same tuple, but with the types changed: project :: (SqlColumn a,SqlColumn b) -> (TypedExpr a,TypedExpr b) project :: (SqlColumn a,SqlColumn b,SqlColumn c) -> (TypedExpr a,TypedExpr b,TypedExpr c) What do you think? Regards, Keean Schupke.

1 0

RE: Generics...
by Simon Peyton-Jones 05 Feb '04

05 Feb '04

I'm not sure what you want to do. Presumably not "replace all Ints by Floats" because that wouldn't make sense if you applied it to the type data T = MkT Int T | Nil So, can you say precisely what you want your generic function to do? What type would you like it to have? Simon | -----Original Message----- | From: glasgow-haskell-users-bounces(a)haskell.org [mailto:glasgow-haskell-users- | bounces(a)haskell.org] On Behalf Of MR K P SCHUPKE | Sent: 04 February 2004 18:43 | To: glasgow-haskell-users(a)haskell.org | Subject: Re: Generics... | | | I now have a vague clue how to do what I wanted. I wanted to substitute | types in place... a silly example would be to convert Ints to Floats and | Floats to Ints... | | (3 :: Int, 4.0 :: Float) -> (3.0 :: Float, 4 :: Int) | | I already have a function that will convert the input to: | | [("3",IntType),("4.0",FloatType)] | | So I can then manipulate the list in this form to: | | [("3.0",FloatType),("4",IntType)] | | All I now need to do is reconstruct back into the data structure... | | I have however been unable to find any docs on "gunfold" which I | think is the way to do this... | | If somebody point me to some documentation for gunfold, or | help in any way I would be very grateful... | | Regards, | Keean Schupke. | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users

1 0

Re: Generics...
by MR K P SCHUPKE 04 Feb '04

04 Feb '04

I now have a vague clue how to do what I wanted. I wanted to substitute types in place... a silly example would be to convert Ints to Floats and Floats to Ints... (3 :: Int, 4.0 :: Float) -> (3.0 :: Float, 4 :: Int) I already have a function that will convert the input to: [("3",IntType),("4.0",FloatType)] So I can then manipulate the list in this form to: [("3.0",FloatType),("4",IntType)] All I now need to do is reconstruct back into the data structure... I have however been unable to find any docs on "gunfold" which I think is the way to do this... If somebody point me to some documentation for gunfold, or help in any way I would be very grateful... Regards, Keean Schupke.

1 0

Generics and Substitution...
by MR K P SCHUPKE 04 Feb '04

04 Feb '04

I have used generics in a couple of places (from the scrap your boilerplate paper) and want to know if it is possible to write a function that will (for example) change the types 'in place' in a type. For example replace all the elements in a tuple with a string (using Show): (3 :: Int,3.5 :: Float,True :: Bool) -> ("3" :: String,"3.5" :: String,"True" :: String) Looking at the paper it seems a version of MkT with a different type is required: MkTT :: (Typeable a,Typeable b,Typeable c,Typeable d) => (c -> d) -> a -> b You could then use: subInt :: Int -> String subInt i = showInt i "" again everywhere has the wrong type so something like: everywhere' :: (Term a,Term b) => (forall c.Term c,forall d.Term d => c -> d) -> a -> b everywhere' f x = f (gmapTT (everywhere' f) x) but can this "gmapTT" be defined in terms of "gfoldl"? I tried to think about this... but I haven't quite got a handle on gfoldl's type yet... Is it possible? Would the function look like the definition of "gmapT" with a different type? Regards, Keean Schupke.

1 0

RE: Non-exhaustive patterns in basicTypes/Var.lhs
by Simon Peyton-Jones 03 Feb '04

03 Feb '04

We can't reproduce this at all. What version of GHC are you using? Can you show a complete script? Simon | -----Original Message----- | From: glasgow-haskell-users-bounces(a)haskell.org [mailto:glasgow-haskell-users- | bounces(a)haskell.org] On Behalf Of Peter Simons | Sent: 18 January 2004 19:54 | To: glasgow-haskell-users(a)haskell.org | Subject: Non-exhaustive patterns in basicTypes/Var.lhs | | QuickCheck's 'generate' function works fine in GHCi, but | only for the _first_ time I call it. After that, I get an | error: | | | Ok, modules loaded: Main. | | *Main> generate 3 (mkStdGen 28) (return 'x') | | Loading package QuickCheck ... linking ... done. | | 'x' | | | | *Main> generate 3 (mkStdGen 28) (return 'x') | | *** Exception: basicTypes/Var.lhs:226:32-58: Non-exhaustive | | patterns in record update | | Peter | | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgow-haskell-users

1 0

RE: GHC kit for Windows
by Graham Klyne 03 Feb '04

03 Feb '04

At 15:34 03/02/04 +0000, Simon Peyton-Jones wrote: >Graham wrote > >| > I note that the GHC kit for Windows includes some of the (I presume) >| > MinGW >| > binutil components, but does not include 'ar'. I find that 'ar' is >| > needed >| > for building HaXml. May I suggest that 'ar' also be included in >this >| > kit? > >Sigbjorn says > >| Adding 'ar' is clearly trivial, but the tools that GHC installers >| include aren't meant to be used by the public, only by GHC >| to avoid having to rely on external tools when compiling& >| linking (e.g., 'gcc' and 'ld' don't work as-is, but need the right >| -B setting). I'm not strongly against it, but adding other dev >| tools does have potential support costs > >I'm inclined to agree with Sigbjorn here. Where would we stop? GHC >does include some Mingw stuff "in its belly", but you are not supposed >to know or care how GHC does its work. You certainly aren't expected to >invoke those tools directly. If you want anything more than GHC, I >think the best thing is to get Mingw or Cygwin or SFU. Fair point. My counter argument would be that the 'ar' utility is (I think) needed to create a GHC package (from just Haskell source files). Am I right about this? Maybe a useful touchstone might be: would be it appropriate to be able create a GHC package using *just* the GHC command? OTOH, if the GHC package structure is to be displaced by the new library infrastructure project, then I'd fully agree with you. Here's a possible simple compromise: in the documentation of GHC packages, include a link to the MinGW kit [1] that is needed to build a package on windows? (It took me some time to track down what I needed.) #g -- [1] http://www.mingw.org/ http://www.mingw.org/download.shtml http://prdownloads.sf.net/mingw/MinGW-3.1.0-1.exe?download ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact

1 0

RE: GHC kit for Windows
by Simon Peyton-Jones 03 Feb '04

03 Feb '04

Graham wrote | > I note that the GHC kit for Windows includes some of the (I presume) | > MinGW | > binutil components, but does not include 'ar'. I find that 'ar' is | > needed | > for building HaXml. May I suggest that 'ar' also be included in this | > kit? Sigbjorn says | Adding 'ar' is clearly trivial, but the tools that GHC installers | include aren't meant to be used by the public, only by GHC | to avoid having to rely on external tools when compiling& | linking (e.g., 'gcc' and 'ld' don't work as-is, but need the right | -B setting). I'm not strongly against it, but adding other dev | tools does have potential support costs I'm inclined to agree with Sigbjorn here. Where would we stop? GHC does include some Mingw stuff "in its belly", but you are not supposed to know or care how GHC does its work. You certainly aren't expected to invoke those tools directly. If you want anything more than GHC, I think the best thing is to get Mingw or Cygwin or SFU. Simon

1 0

RE: ghc6.2 on Solaris
by Simon Marlow 02 Feb '04

02 Feb '04

> I have produced, with Simon M's help, an unofficial binary > release of ghc 6.2 on > Solaris. This may be downloaded from > > http://www.informatik.uni-bremen.de/~ger/ghc/ Thanks to George for putting this together. We should be clear that this build is 6.2+patches (a snapshot along the 6.2.x branch). If you encouter bugs using this distribution, please state clearly that it was this distribution and not a plain 6.2 build to avoid confusion. Cheers, Simon

1 0

ghc6.2 on Solaris
by George Russell 02 Feb '04

02 Feb '04

I have produced, with Simon M's help, an unofficial binary release of ghc 6.2 on Solaris. This may be downloaded from http://www.informatik.uni-bremen.de/~ger/ghc/

1 0