Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
  Reporter:  augustss|  Owner:  simonmar
  Type:  bug | Status:  closed  
  Priority:  high|  Milestone:  7.2.1   
 Component:  libraries/base  |Version:  7.1 
Resolution:  fixed   |   Keywords:  
  Testcase:  |  Blockedby:  
Difficulty:  | Os:  Unknown/Multiple
  Blocking:  |   Architecture:  Unknown/Multiple
   Failure:  None/Unknown|  
-+--
Changes (by simonmar):

  * status:  new => closed
  * resolution:  => fixed


Comment:

 Now fixed; see 52ddeaaedfed3135a739585b41fdd2ad4f1f8e81 and following
 commits, and 903c1afa3bdd205dc5473c0fca1065d598145311 in packages/base.

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  simonmar
Type:  bug   |   Status:  new 
Priority:  high  |Milestone:  7.2.1   
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by simonmar):

 strictly speaking, in 7.1 `tyConString` changed only for the built-in
 types for which instances were declared in `Data.Typeable`, because we
 switched from using CPP macros to declare the instances to using
 standalone deriving.  In 7.2 we'll be changing the `tyConString` for
 derived instances to match the behaviour of `tyConString` in the CPP macro
 instances in 7.0.

 So for the built-in types, we'll be reverting to the 7.0 behaviour, and
 for all other types the `tyConSting`s will change.

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  simonmar
Type:  bug   |   Status:  new 
Priority:  high  |Milestone:  7.2.1   
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by malcolm.wallace@…):

 As I understand it, the value of tyConString has already changed, between
 the released 7.0 and the development head 7.1.  The request is to change
 it back, before the release of 7.2, to the same behaviour as in 7.0 and
 before.

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  simonmar
Type:  bug   |   Status:  new 
Priority:  high  |Milestone:  7.2.1   
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by igloo):

 So if I understand correctly, the value of `tyConString` for derived
 `Typeable` instances will change between 7.0 and 7.2?

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  simonmar
Type:  bug   |   Status:  new 
Priority:  high  |Milestone:  7.2.1   
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by simonmar):

 Replying to [comment:4 batterseapower]:

 > I don't think Lennart wanted a way to deconstruct a TyCon. He just
 wanted equality on TyCon strings to be a) backward compatible and b) not
 reveal implementation details about where "built-in" types like Int were
 actually declared. Do you plan to do something separate to address that
 point?

 The specific problem is that `tyConString` returns a string that is not
 portable and reveals implementation details.  Making `tyConString` return
 just the name of the !TyCon and not its implementation module would fix
 that.  We also need a way to uniquely identify !TyCons, and the hash
 provides that - the hash would be the Md5 of "package-1.0:Module.TyCon",
 but computed at compile-time because it's easier.  The other fields are
 just for information (we'll probably provide GHC-specific ways to get at
 them).

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  simonmar
Type:  bug   |   Status:  new 
Priority:  high  |Milestone:  7.2.1   
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by batterseapower):

 Replying to [comment:2 simonpj]:
 > Simon and I talked about this.  We propose to change `Typeable.TyCon`
 from this:
 > {{{
 > data TyCon = TyCon !Key String
 > }}}
 > to this:
 > {{{
 > data TyCon = TyCon !Key
 >String-- Package name
 >String-- Module name
 >String-- TyCon name
 > }}}

 I don't think Lennart wanted a way to deconstruct a TyCon. He just wanted
 equality on TyCon strings to be a) backward compatible and b) not reveal
 implementation details about where "built-in" types like Int were actually
 declared. Do you plan to do something separate to address that point?

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  simonmar
Type:  bug   |   Status:  new 
Priority:  high  |Milestone:  7.2.1   
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--
Changes (by simonpj):

  * owner:  => simonmar
  * priority:  normal => high
  * milestone:  => 7.2.1


-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  
Type:  bug   |   Status:  new 
Priority:  normal|Milestone:  
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by simonpj):

 Simon and I talked about this.  We propose to change `Typeable.TyCon` from
 this:
 {{{
 data TyCon = TyCon !Key String
 }}}
 to this:
 {{{
 data TyCon = TyCon !Key
String-- Package name
String-- Module name
String-- TyCon name
 }}}
 and add accessors for these fields. Then you can get the bits you want.

 It's not enough to keep just the loacl name (like Integer) because there
 might be many types with the same name.

 A the same time we propose to get rid of the side-effected cache and use
 MD5 checksums instead.

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Re: [GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |Owner:  
Type:  bug   |   Status:  new 
Priority:  normal|Milestone:  
   Component:  libraries/base|  Version:  7.1 
Keywords:| Testcase:  
   Blockedby:|   Difficulty:  
  Os:  Unknown/Multiple  | Blocking:  
Architecture:  Unknown/Multiple  |  Failure:  None/Unknown
-+--

Comment(by simonmar):

 I agree.  Is there some reason we use the qualified original name here?
 It's not even necessarily unique, because we don't include the package
 name.

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

[GHC] #5275: Data.Typeable not backwards compatible

[GHC]

#5275: Data.Typeable not backwards compatible
-+--
Reporter:  augustss  |   Owner:
Type:  bug   |  Status:  new   
Priority:  normal|   Component:  libraries/base
 Version:  7.1   |Keywords:
Testcase:|   Blockedby:
  Os:  Unknown/Multiple  |Blocking:
Architecture:  Unknown/Multiple  | Failure:  None/Unknown  
-+--
 In ghc 7.1 the type name returned for prelude types has changed.  It used
 to be Int, Integer, etc, but in ghc 7.1 it's GHC.Types.Int,
 GHC.Integer.Type.Integer, etc.

 This has many drawbacks:[[BR]]
  * The names are impossible to guess[[BR]]
  * The names are not backwards compatible[[BR]]
  * The names are not portable to other compilers[[BR]]
  * The names reveal implementation details that should be kept hidden

-- 
Ticket URL: 
GHC 
The Glasgow Haskell Compiler

___
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs