[GHC] #10725: Figure out how to support type synonym implementions of abstract data

[GHC]

#10725: Figure out how to support type synonym implementions of abstract data
-------------------------------------+-------------------------------------
        Reporter:  ezyang            |                Owner:  ezyang
            Type:  task              |               Status:  closed
        Priority:  normal            |            Milestone:
       Component:  Compiler (Type    |              Version:  7.10.2
  checker)                           |
      Resolution:  fixed             |             Keywords:  backpack
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Changes (by ezyang):

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


Comment:

 It was really easy, and I've implemented it on my branch. Just a few minor
 complications:

 1. When we recompile an hsig file with the actual backing implementation,
 we need to notice when we define an abstract type "data A" which is
 implemented with a type synonym, and "swap in" the correct type synonym
 immediately. If we don't, when we subsequently refer to A we'll get a
 TyCon that is not decorated as a type synonym (despite being the same
 original name as a type synonym), and then type equality chokes.

 2. There was a bug when serializing out hsig interface files where I was
 incorrectly grabbing fingerprints locally rather than from the backing
 implementation. That's fixed now.

 https://github.com/ezyang/ghc/commit/4d59c74b2381eb8e2b9c1d31111a921ed95ebd23

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10725#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler