December 2004 - template-haskell

Template Haskell in 6.4
by Simon Peyton-Jones 23 Dec '04

23 Dec '04

Dear Template Haskell folk I think I've reached the end of my urgent to-do list for Template Haskell in GHC 6.4. Can any of you build the HEAD and try it out? In particular, I think most of the dust has settled on the new Name data type. The worst feature at the moment is that if the programmer constructs an utterly bogus piece of TH syntax, and then splices it in, there's an abysmal error message (emitted by hsSyn/Convert.lhs). The point is that TH.Syntax allows many syntactically ill-formed programs (e.g. putting a forall type in a context) and that simply crashes the conversion from TH.Syntax -> HsSyn What is needed is for someone to take hsSyn/Convert in hand, and give it proper error messages. The conversion process is not unlike the renamer -- it may fail for many programmer reasons, and then it should produce a decent error message. Any volunteers? It's not difficult... Simon

1 0

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 23 Dec '04

23 Dec '04

Fair point. I'll make them print identically for now, and we can see if anyone has difficulty. Simon | -----Original Message----- | From: template-haskell-bounces(a)haskell.org [mailto:template-haskell-bounces@haskell.org] On | Behalf Of Duncan Coutts | Sent: 22 December 2004 22:35 | To: Simon Peyton-Jones | Cc: template-haskell(a)haskell.org | Subject: RE: [Template-haskell] Re: template-haskell names in 6.3 | | On Tue, 2004-12-21 at 17:05 +0000, Simon Peyton-Jones wrote: | > There is a difference between the result of calling mkName, and a quoted | > name e.g 'map. The former is dynamically bound and the latter is | > lexically bound. | > | > Because there's a difference, I've made them print differently in the | > pretty printer | > | > Foo.baz for the dynamically bound one | > Foo:baz for the lexically scoped one | > | > This might break someone's code. Maybe they should both print as | > "Foo.baz"? | | Since lexically scoped is the normal situation perhaps it should get the | ordinary Foo.baz syntax and the dynamically bound variables should get | the other syntax. | | Actually, for non-qualified variables there is still no syntactic | difference between dynamically and statically scoped ones so it begs the | question of why a syntactic difference is needed for the qualified case | (or perhaps if a syntactic difference in the non-qualified would be a | good idea). | | Duncan | | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

1 0

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 22 Dec '04

22 Dec '04

Dear Template Haskell folk I've now arranged that nameBase (mkName "Foo.baz") == "baz" nameModule (MkName "Foo.baz") == "Foo" as you might expect. There is a difference between the result of calling mkName, and a quoted name e.g 'map. The former is dynamically bound and the latter is lexically bound. Because there's a difference, I've made them print differently in the pretty printer Foo.baz for the dynamically bound one Foo:baz for the lexically scoped one This might break someone's code. Maybe they should both print as "Foo.baz"? There is no Haddock documentation to speak of in the TH libraries. Would someone (Ian?) like to add some? Simon | -----Original Message----- | From: Keean Schupke [mailto:k.schupke@imperial.ac.uk] | Sent: 21 December 2004 15:31 | To: Simon Peyton-Jones | Cc: template-haskell(a)haskell.org | Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | | It won't make a difference right now, so you can leave it out for now, | but I think it should work... | | Keean. | | Simon Peyton-Jones wrote: | | >aha, that's a good question. | > | >Hmm. It probably ought to work (though it won't today). Would it make | >a difference? | > | >S | > | >| -----Original Message----- | >| From: Keean Schupke [mailto:k.schupke@imperial.ac.uk] | >| Sent: 21 December 2004 14:03 | >| To: Simon Peyton-Jones | >| Cc: template-haskell(a)haskell.org | >| Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | >| | >| So does this work? | >| | >| nameModule (mkName "Foo.Bar") | >| | >| Keean. | > | > | >

2 1

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 21 Dec '04

21 Dec '04

aha, that's a good question. Hmm. It probably ought to work (though it won't today). Would it make a difference? S | -----Original Message----- | From: Keean Schupke [mailto:k.schupke@imperial.ac.uk] | Sent: 21 December 2004 14:03 | To: Simon Peyton-Jones | Cc: template-haskell(a)haskell.org | Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | | So does this work? | | nameModule (mkName "Foo.Bar") | | Keean.

2 1

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 21 Dec '04

21 Dec '04

| As to exactly what I am trying to do, it is to take a string say "Test" | and prefix | it with qCurrentModule to give the global name, so that: So you can just say do { m <- qCurrentModule ; return (mkName (m ++ ".Test")) } Let's leave at that for now, if that does what you need. I'm very cautious about adding equalities on Names. You can always ask if m <- qCurrentModule if (m == nameModule ''Test) ... Simon | module Main | data X | | main = do | m <- qCurrentModule | if mkQualName m "X" == ''X -- IE I want to test if m.X is really | the same X as ''X | then putStrLn "Okay" | else putStrLn "Fail" | | | | Keean. | | Simon Peyton-Jones wrote: | | >Because they the two can behave differently in different contexts. It's | >like saying why doesn't 1==2 because in the context (\x. x>10) both give | >False! | > | >I feel I am missing the point you are trying to make. I thought I had | >it when I suggested mkQualName, and you said "that's exactly what I | >want". But you can define mkQualName easily: | > mkQualName m s = mkName (m ++ "." ++ s). | > | >Why don't you give a complete (but small) example of what you are trying | >to do? | > | >Simon | > | >| -----Original Message----- | >| From: Keean Schupke [mailto:k.schupke@imperial.ac.uk] | >| Sent: 21 December 2004 10:48 | >| To: Simon Peyton-Jones | >| Cc: template-haskell(a)haskell.org | >| Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | >| | >| Okay but: | >| | >| >module Foo where | >| > data Test | >| > | >| >module Bar | >| > import Foo | >| > | >| > test = (''Test == mkName "Foo.Test") | >| | >| Test (in this scope) is Foo.Test, so why shouldn't it equal | >| mkName "Foo.Test" ? | >| | >| Keean. | >| | >| | >| Simon Peyton-Jones wrote: | >| | >| >No, certainly not. | >| > | >| > ''Test :: Name | >| > is a lexically-scoped Name, indissolubly connected to whatever | >| >Test means | >| > in the original lexical scope it appeared in | >| > | >| > mkName "Foo.Test" :: Name | >| > is a dynamically-scoped Name, captured by whatever context it's | >| >spliced into. | >| > | >| >They aren't the same, and shouldn't be! | >| > | >| >Simon | >| > | >| >| -----Original Message----- | >| >| From: template-haskell-bounces(a)haskell.org | >| >[mailto:template-haskell-bounces@haskell.org] On | >| >| Behalf Of Keean Schupke | >| >| Sent: 21 December 2004 10:06 | >| >| To: Simon Peyton-Jones | >| >| Cc: template-haskell(a)haskell.org | >| >| Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | >| >| | >| >| I think I am already using that... I wasn't sure if the | >'module-part' | >| >of the | >| >| name was the same as in a global name generated from ''... is: | >| >| | >| >| module Foo | >| >| data Test | >| >| | >| >| mkName "Foo.Test" == ''Test -- is this true? | >| >| | >| >| Keean | >| >| | >| >| Simon Peyton-Jones wrote: | >| >| | >| >| >Keean | >| >| > | >| >| >I was half way through implementing this when I realised it's | >there | >| >| >already. If you say | >| >| > | >| >| > mkName "Foo.baz" | >| >| > | >| >| >you'll get a dynamically scoped qualified name, just as you want. | >| >| > | >| >| >Is that OK? | >| >| > | >| >| >Simon | >| >| > | >| >| >| -----Original Message----- | >| >| >| From: template-haskell-bounces(a)haskell.org | >| >| >[mailto:template-haskell-bounces@haskell.org] On | >| >| >| Behalf Of Keean Schupke | >| >| >| Sent: 03 December 2004 14:38 | >| >| >| To: Simon Peyton-Jones | >| >| >| Cc: template-haskell(a)haskell.org | >| >| >| Subject: Re: [Template-haskell] Re: template-haskell names in | >6.3 | >| >| >| | >| >| >| Simon Peyton-Jones wrote: | >| >| >| | >| >| >| >do you mean | >| >| >| > n <- newName s | >| >| >| >or | >| >| >| > let n = mkName s | >| >| >| >? | >| >| >| > | >| >| >| > | >| >| >| I mean the latter... or perhaps: "n <- return $ mkName s"... | >| >| >| | >| >| >| >If the latter, you will get | >| >| >| > data Foo = Foo Foo | >| >| >| > | >| >| >| >... and then I see what you want. You want a version of mkName | >| >that | >| >| >is | >| >| >| >like giving a qualified name in Haskell. It's dynamically | >scoped | >| >| >just | >| >| >| >like mkName, but it obeys the usual rules for qualified names | >in | >| >| >| >Haskell. | >| >| >| > | >| >| >| >To be totally explicit, suppose I have | >| >| >| > | >| >| >| > foo = [| mkQualName "Foo" "baz" |] | >| >| >| > | >| >| >| >then if I call foo thus | >| >| >| > | >| >| >| > wibble = \baz -> $foo | >| >| >| > | >| >| >| >the $foo splice will expand to "Foo.baz", and that won't see | >the | >| >| >\baz; | >| >| >| >it'll see whatever Foo.baz is in scope. | >| >| >| > | >| >| >| > | >| >| >| This is exactly what I am after... | >| >| >| | >| >| >| >Is that what you seek? I can see it's reasonable. I'd need to | >| >add | >| >| >| > | >| >| >| > mkQualName :: String -> String -> Name | >| >| >| > | >| >| >| >That'd mean an extra form of Name. Currently, if you say | >| >| >| > 'Foo.baz | >| >| >| >you'll get a Name whose nameModule isn't necessarily Foo... | >it'll | >| >be | >| >| >the | >| >| >| >module that baz was actually defined in. | >| >| >| > | >| >| >| > | >| >| >| I'll leave that up to you, I don't mind if this translation | >| >happens, | >| >| >as | >| >| >| long as it refers to the Foo.baz in dynamic scope the actuall | >name | >| >| >| can change. | >| >| >| | >| >| >| Keean | >| >| >| _______________________________________________ | >| >| >| template-haskell mailing list | >| >| >| template-haskell(a)haskell.org | >| >| >| http://www.haskell.org/mailman/listinfo/template-haskell | >| >| > | >| >| > | >| >| | >| >| _______________________________________________ | >| >| template-haskell mailing list | >| >| template-haskell(a)haskell.org | >| >| http://www.haskell.org/mailman/listinfo/template-haskell | >| > | >| > | > | > | > | | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

2 1

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 21 Dec '04

21 Dec '04

Because they the two can behave differently in different contexts. It's like saying why doesn't 1==2 because in the context (\x. x>10) both give False! I feel I am missing the point you are trying to make. I thought I had it when I suggested mkQualName, and you said "that's exactly what I want". But you can define mkQualName easily: mkQualName m s = mkName (m ++ "." ++ s). Why don't you give a complete (but small) example of what you are trying to do? Simon | -----Original Message----- | From: Keean Schupke [mailto:k.schupke@imperial.ac.uk] | Sent: 21 December 2004 10:48 | To: Simon Peyton-Jones | Cc: template-haskell(a)haskell.org | Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | | Okay but: | | >module Foo where | > data Test | > | >module Bar | > import Foo | > | > test = (''Test == mkName "Foo.Test") | | Test (in this scope) is Foo.Test, so why shouldn't it equal | mkName "Foo.Test" ? | | Keean. | | | Simon Peyton-Jones wrote: | | >No, certainly not. | > | > ''Test :: Name | > is a lexically-scoped Name, indissolubly connected to whatever | >Test means | > in the original lexical scope it appeared in | > | > mkName "Foo.Test" :: Name | > is a dynamically-scoped Name, captured by whatever context it's | >spliced into. | > | >They aren't the same, and shouldn't be! | > | >Simon | > | >| -----Original Message----- | >| From: template-haskell-bounces(a)haskell.org | >[mailto:template-haskell-bounces@haskell.org] On | >| Behalf Of Keean Schupke | >| Sent: 21 December 2004 10:06 | >| To: Simon Peyton-Jones | >| Cc: template-haskell(a)haskell.org | >| Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | >| | >| I think I am already using that... I wasn't sure if the 'module-part' | >of the | >| name was the same as in a global name generated from ''... is: | >| | >| module Foo | >| data Test | >| | >| mkName "Foo.Test" == ''Test -- is this true? | >| | >| Keean | >| | >| Simon Peyton-Jones wrote: | >| | >| >Keean | >| > | >| >I was half way through implementing this when I realised it's there | >| >already. If you say | >| > | >| > mkName "Foo.baz" | >| > | >| >you'll get a dynamically scoped qualified name, just as you want. | >| > | >| >Is that OK? | >| > | >| >Simon | >| > | >| >| -----Original Message----- | >| >| From: template-haskell-bounces(a)haskell.org | >| >[mailto:template-haskell-bounces@haskell.org] On | >| >| Behalf Of Keean Schupke | >| >| Sent: 03 December 2004 14:38 | >| >| To: Simon Peyton-Jones | >| >| Cc: template-haskell(a)haskell.org | >| >| Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | >| >| | >| >| Simon Peyton-Jones wrote: | >| >| | >| >| >do you mean | >| >| > n <- newName s | >| >| >or | >| >| > let n = mkName s | >| >| >? | >| >| > | >| >| > | >| >| I mean the latter... or perhaps: "n <- return $ mkName s"... | >| >| | >| >| >If the latter, you will get | >| >| > data Foo = Foo Foo | >| >| > | >| >| >... and then I see what you want. You want a version of mkName | >that | >| >is | >| >| >like giving a qualified name in Haskell. It's dynamically scoped | >| >just | >| >| >like mkName, but it obeys the usual rules for qualified names in | >| >| >Haskell. | >| >| > | >| >| >To be totally explicit, suppose I have | >| >| > | >| >| > foo = [| mkQualName "Foo" "baz" |] | >| >| > | >| >| >then if I call foo thus | >| >| > | >| >| > wibble = \baz -> $foo | >| >| > | >| >| >the $foo splice will expand to "Foo.baz", and that won't see the | >| >\baz; | >| >| >it'll see whatever Foo.baz is in scope. | >| >| > | >| >| > | >| >| This is exactly what I am after... | >| >| | >| >| >Is that what you seek? I can see it's reasonable. I'd need to | >add | >| >| > | >| >| > mkQualName :: String -> String -> Name | >| >| > | >| >| >That'd mean an extra form of Name. Currently, if you say | >| >| > 'Foo.baz | >| >| >you'll get a Name whose nameModule isn't necessarily Foo... it'll | >be | >| >the | >| >| >module that baz was actually defined in. | >| >| > | >| >| > | >| >| I'll leave that up to you, I don't mind if this translation | >happens, | >| >as | >| >| long as it refers to the Foo.baz in dynamic scope the actuall name | >| >| can change. | >| >| | >| >| Keean | >| >| _______________________________________________ | >| >| template-haskell mailing list | >| >| template-haskell(a)haskell.org | >| >| http://www.haskell.org/mailman/listinfo/template-haskell | >| > | >| > | >| | >| _______________________________________________ | >| template-haskell mailing list | >| template-haskell(a)haskell.org | >| http://www.haskell.org/mailman/listinfo/template-haskell | > | >

2 1

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 21 Dec '04

21 Dec '04

No, certainly not. ''Test :: Name is a lexically-scoped Name, indissolubly connected to whatever Test means in the original lexical scope it appeared in mkName "Foo.Test" :: Name is a dynamically-scoped Name, captured by whatever context it's spliced into. They aren't the same, and shouldn't be! Simon | -----Original Message----- | From: template-haskell-bounces(a)haskell.org [mailto:template-haskell-bounces@haskell.org] On | Behalf Of Keean Schupke | Sent: 21 December 2004 10:06 | To: Simon Peyton-Jones | Cc: template-haskell(a)haskell.org | Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | | I think I am already using that... I wasn't sure if the 'module-part' of the | name was the same as in a global name generated from ''... is: | | module Foo | data Test | | mkName "Foo.Test" == ''Test -- is this true? | | Keean | | Simon Peyton-Jones wrote: | | >Keean | > | >I was half way through implementing this when I realised it's there | >already. If you say | > | > mkName "Foo.baz" | > | >you'll get a dynamically scoped qualified name, just as you want. | > | >Is that OK? | > | >Simon | > | >| -----Original Message----- | >| From: template-haskell-bounces(a)haskell.org | >[mailto:template-haskell-bounces@haskell.org] On | >| Behalf Of Keean Schupke | >| Sent: 03 December 2004 14:38 | >| To: Simon Peyton-Jones | >| Cc: template-haskell(a)haskell.org | >| Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | >| | >| Simon Peyton-Jones wrote: | >| | >| >do you mean | >| > n <- newName s | >| >or | >| > let n = mkName s | >| >? | >| > | >| > | >| I mean the latter... or perhaps: "n <- return $ mkName s"... | >| | >| >If the latter, you will get | >| > data Foo = Foo Foo | >| > | >| >... and then I see what you want. You want a version of mkName that | >is | >| >like giving a qualified name in Haskell. It's dynamically scoped | >just | >| >like mkName, but it obeys the usual rules for qualified names in | >| >Haskell. | >| > | >| >To be totally explicit, suppose I have | >| > | >| > foo = [| mkQualName "Foo" "baz" |] | >| > | >| >then if I call foo thus | >| > | >| > wibble = \baz -> $foo | >| > | >| >the $foo splice will expand to "Foo.baz", and that won't see the | >\baz; | >| >it'll see whatever Foo.baz is in scope. | >| > | >| > | >| This is exactly what I am after... | >| | >| >Is that what you seek? I can see it's reasonable. I'd need to add | >| > | >| > mkQualName :: String -> String -> Name | >| > | >| >That'd mean an extra form of Name. Currently, if you say | >| > 'Foo.baz | >| >you'll get a Name whose nameModule isn't necessarily Foo... it'll be | >the | >| >module that baz was actually defined in. | >| > | >| > | >| I'll leave that up to you, I don't mind if this translation happens, | >as | >| long as it refers to the Foo.baz in dynamic scope the actuall name | >| can change. | >| | >| Keean | >| _______________________________________________ | >| template-haskell mailing list | >| template-haskell(a)haskell.org | >| http://www.haskell.org/mailman/listinfo/template-haskell | > | > | | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

2 1

RE: [Template-haskell] Re: template-haskell names in 6.3
by Simon Peyton-Jones 21 Dec '04

21 Dec '04

Keean I was half way through implementing this when I realised it's there already. If you say mkName "Foo.baz" you'll get a dynamically scoped qualified name, just as you want. Is that OK? Simon | -----Original Message----- | From: template-haskell-bounces(a)haskell.org [mailto:template-haskell-bounces@haskell.org] On | Behalf Of Keean Schupke | Sent: 03 December 2004 14:38 | To: Simon Peyton-Jones | Cc: template-haskell(a)haskell.org | Subject: Re: [Template-haskell] Re: template-haskell names in 6.3 | | Simon Peyton-Jones wrote: | | >do you mean | > n <- newName s | >or | > let n = mkName s | >? | > | > | I mean the latter... or perhaps: "n <- return $ mkName s"... | | >If the latter, you will get | > data Foo = Foo Foo | > | >... and then I see what you want. You want a version of mkName that is | >like giving a qualified name in Haskell. It's dynamically scoped just | >like mkName, but it obeys the usual rules for qualified names in | >Haskell. | > | >To be totally explicit, suppose I have | > | > foo = [| mkQualName "Foo" "baz" |] | > | >then if I call foo thus | > | > wibble = \baz -> $foo | > | >the $foo splice will expand to "Foo.baz", and that won't see the \baz; | >it'll see whatever Foo.baz is in scope. | > | > | This is exactly what I am after... | | >Is that what you seek? I can see it's reasonable. I'd need to add | > | > mkQualName :: String -> String -> Name | > | >That'd mean an extra form of Name. Currently, if you say | > 'Foo.baz | >you'll get a Name whose nameModule isn't necessarily Foo... it'll be the | >module that baz was actually defined in. | > | > | I'll leave that up to you, I don't mind if this translation happens, as | long as it refers to the Foo.baz in dynamic scope the actuall name | can change. | | Keean | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

2 1

RE: FW: [Template-haskell] TH restrictions
by Simon Peyton-Jones 15 Dec '04

15 Dec '04

Ian, I don't think I understand what you are saying here. What is your number 1? Simon | -----Original Message----- | From: Ian Lynagh [mailto:ian@matrix.chaos.earth.li] On Behalf Of Ian Lynagh | Sent: 05 December 2004 03:03 | To: Simon Peyton-Jones; template-haskell(a)haskell.org | Subject: Re: FW: [Template-haskell] TH restrictions | | On Thu, Nov 18, 2004 at 01:15:25AM +0000, Ian Lynagh wrote: | > | > For what it's worth, I think one of the biggest irritations I've had | > with writing stuff using TH is the need to split things across module | > boundaries. | | Actually, I think this is number 2 on my list, below portability. | | As I said recently, I've hacked something I wrote to print out TH | generated code for use on a platform without ghci+TH, which is a pain | but not a showstopper for myself. | | However, I wanted to unroll the inner loop in SHA1 (essentially says "do | this for n = 0..79") for darcs, and really don't want it to get into the | situation of having to maintain generated code for platforms without | ghci+TH. So for the time being I have to leave it either in a slower HO | form or the fast but ugly hand-unrolled form. | | I appreciate that this is also a hard problem to solve; I just thought | I'd let you know it is an issue. | | (In general portability to other implementations may be an issue too, of | course, although darcs currently doesn't support them anyway) | | | Thanks | Ian

1 0

RE: [Template-haskell] Re: Show Info
by Simon Peyton-Jones 15 Dec '04

15 Dec '04

Jim You need to update to the latest HEAD. Then it works. (But there was a missing case in pretty-printing for PrimTyConI which I fixed this morning, so "latest" means any time from now.) Simon | -----Original Message----- | From: template-haskell-bounces(a)haskell.org [mailto:template-haskell-bounces@haskell.org] On | Behalf Of Jim Apple | Sent: 14 December 2004 06:00 | To: template-haskell(a)haskell.org | Subject: [Template-haskell] Re: Show Info | | Jim Apple wrote: | > Is there any reason why Info isn't deriving(Show) ? | | Ok, I see from the testsuite examples and some code that I can print it | with pprint. I am now getting this new error: | | >$(do {display ''(->); [| () |]}) | | ghc-6.3.20041123: panic! (the `impossible' happened, GHC version | 6.3.20041123): | algTyConRhs GHCziPrim.ZLzmzgZR{(w) tc 3D} | | with display as in | | testsuite/tests/ghc-regress/th/TH_reifyDecl1.hs | | Jim | | _______________________________________________ | template-haskell mailing list | template-haskell(a)haskell.org | http://www.haskell.org/mailman/listinfo/template-haskell

1 0