[GHC] #14254: The Binary instance for TypeRep smells a bit expensive

GHC ghc-devs at haskell.org
Wed Oct 11 09:12:19 UTC 2017


#14254: The Binary instance for TypeRep smells a bit expensive
-------------------------------------+-------------------------------------
        Reporter:  dfeuer            |                Owner:  (none)
            Type:  bug               |               Status:  patch
        Priority:  normal            |            Milestone:  8.4.1
       Component:  Compiler          |              Version:  8.2.1
      Resolution:                    |             Keywords:  Typeable
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):  Phab:D3998,
       Wiki Page:                    |  Phab:D4078, Phab:D4082
-------------------------------------+-------------------------------------

Comment (by dfeuer):

 One more thing: while I don't think it's a good idea, we ''could'' go with
 a version of Phab:D4085 that only caches kind typereps in `TrTyCon`
 constructors. That would eliminate the allocation problem and alleviate
 but not eliminate the time problem. The reason I favor the fully cached
 version is that it doesn't strike me as really that expensive to add one
 more pointer to a constructor that already has four words of payload (two
 for the `Fingerprint`, one for the function, and one for the argument).
 However, that's somewhat conditional on being able to eliminate the extra
 laziness. It seems there's a knot that needs tying for `Type :: Type`,
 which makes things a tad fussy.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14254#comment:12>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler


More information about the ghc-tickets mailing list