Changes between Version 2 and Version 3 of KindFact


Ignore:
Timestamp:
Mar 9, 2011 10:52:31 PM (3 years ago)
Author:
pigworker
Comment:

fixed link to context aliases, added link to class aliases

Legend:

Unmodified
Added
Removed
Modified
  • KindFact

    v2 v3  
    11Proposal: extend the kind system with a kind '''Fact''' to cover constraints as well as types, in order to reuse existing abstraction mechanisms, notably '''type synonyms''', in the constraint language. 
    22 
    3 Much of the motivation for this proposal can be found in [http://www.cs.kuleuven.be/%7Etoms/Research/papers/constraint_families.pdf Haskell Type Constraints Unleashed] which identifies the shortage of abstraction mechanisms for constraints relative to types. See ticket #788 for the resulting '''constraint synonym''' proposal, which seeks to fill some of the gaps with new declaration forms. Here, however, the plan is to extend the kind system, empowering the existing mechanisms to work with constraints. [http://blog.omega-prime.co.uk/?p=61 Max Bolingbroke], commenting on [wiki:ContextAlias context aliases] makes a similar suggestion, remarking that a new kind would probably help. The claim here is that the new kind obviates the need for other new syntax. 
     3Much of the motivation for this proposal can be found in [http://www.cs.kuleuven.be/%7Etoms/Research/papers/constraint_families.pdf Haskell Type Constraints Unleashed] which identifies the shortage of abstraction mechanisms for constraints relative to types. See ticket #788 for the resulting '''constraint synonym''' proposal, which seeks to fill some of the gaps with new declaration forms. Here, however, the plan is to extend the kind system, empowering the existing mechanisms to work with constraints. [http://blog.omega-prime.co.uk/?p=61 Max Bolingbroke], commenting on [http://www.haskell.org/haskellwiki/Context_alias context aliases] (in turn based on John Meacham's [http://repetae.net/recent/out/classalias.html class alias] proposal) makes a similar suggestion, remarking that a new kind would probably help. The claim here is that the new kind obviates the need for other new syntax. 
    44 
    55Concretely, the proposal is to