[GHC] #14880: GHC panic: updateRole

[GHC]

#14880: GHC panic: updateRole
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                Owner:  goldfire
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.8.1
       Component:  Compiler (Type    |              Version:  8.2.2
  checker)                           |
      Resolution:                    |             Keywords:  TypeInType
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  crash or panic                     |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #15076            |  Differential Rev(s):  Phab:D4769
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by tdammers):

 So I manually retrofit the -accum change onto the wip/T14880 branch
 (pushed to `wip/T14880-reengineered`); however, I'm still getting stat
 failures. I'm currently running a full `./validate` on a clean build tree
 to confirm, but maybe you could take a look in the meantime to see if I've
 missed anything. I have taken the liberty to change a few things around a
 bit; I believe they shouldn't make a fundamental difference, but maybe I'm
 wrong.

 I also noticed that there is an error in the -accum patch; lines 1664-1667
 say:

 {{{#!haskell
 {- 1664 -} ty_co_vars_of_co_var v is acc
 {- 1665 -}   | v `elemVarSet` is = acc
 {- 1666 -}   | v `elemVarSet` is  = acc
 {- 1667 -}   | otherwise         = ty_co_vars_of_type (varType v) is
 (extendVarSet acc v)
 }}}

 ...but line 1666 is redundant and should probably rather be:

 {{{#!haskell
 {- 1666 -}   | v `elemVarSet` acc = acc
 }}}

 So I changed that.

 The above hasn't been rebased onto master yet btw.; I also have a branch
 around that is the result of actually merging the `wip/T1448-accum` branch
 onto a rebased `wip/T14880` branch - that one also has stat failures, but
 it still contains the error described above. I'll also run a validate on
 this one after fixing that.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14880#comment:105>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler