RE: [commit: ghc] master: Improve error messages around kind mismatches. (8e15e3d)

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

RE: [commit: ghc] master: Improve error messages around kind mismatches. (8e15e3d)

GHC - devs mailing list
|      typecheck/should_fail/T7368) Solution: track the provenance of
|      Derived constraints and refuse to float out a constraint that has
|      an insoluble Derived.

For the record, that's not the solution that the patch implements.  Instead we just don’t float heterogeneous equalities.

I think Richard just forgot to update the commit message.

Simon



|  -----Original Message-----
|  From: ghc-commits [mailto:[hidden email]] On Behalf
|  Of [hidden email]
|  Sent: 27 July 2017 12:50
|  To: [hidden email]
|  Subject: [commit: ghc] master: Improve error messages around kind
|  mismatches. (8e15e3d)
|  
|  Repository : ssh://[hidden email]/ghc
|  
|  On branch  : master
|  Link       :
|  https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fghc.ha
|  skell.org%2Ftrac%2Fghc%2Fchangeset%2F8e15e3d370e9c253ae0dbb330e25b72cb
|  00cdb76%2Fghc&data=02%7C01%7Csimonpj%40microsoft.com%7C9993301a429e467
|  e29b008d4d4e6350f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C6363675
|  32584262688&sdata=bY%2Bruj0hhXfC23hfMA1ivuAs2Av2%2BY4%2BJK2yyCoPzbA%3D
|  &reserved=0
|  
|  >---------------------------------------------------------------
|  
|  commit 8e15e3d370e9c253ae0dbb330e25b72cb00cdb76
|  Author: Richard Eisenberg <[hidden email]>
|  Date:   Thu Jun 1 17:27:14 2017 -0400
|  
|      Improve error messages around kind mismatches.
|  
|      Previously, when canonicalizing (or unifying, in uType) a
|      heterogeneous equality, we emitted a kind equality and used the
|      resulting coercion to cast one side of the heterogeneous equality.
|  
|      While sound, this led to terrible error messages. (See the bugs
|      listed below.) The problem is that using the coercion built from
|      the emitted kind equality is a bit like a wanted rewriting a
|  wanted.
|      The solution is to keep heterogeneous equalities as irreducible.
|  
|      See Note [Equalities with incompatible kinds] in TcCanonical.
|  
|      This commit also removes a highly suspicious switch to
|  FM_SubstOnly
|      when flattening in the kinds of a type variable. I have no idea
|      why this was there, other than as a holdover from pre-TypeInType.
|      I've not left a Note because there is simply no reason I can
|  conceive
|      of that the FM_SubstOnly should be there.
|  
|      One challenge with this patch is that the emitted derived
|  equalities
|      might get emitted several times: when a heterogeneous equality is
|      in an implication and then gets floated out from the implication,
|      the Derived is present both in and out of the implication. This
|      causes a duplicate error message. (Test case:
|      typecheck/should_fail/T7368) Solution: track the provenance of
|      Derived constraints and refuse to float out a constraint that has
|      an insoluble Derived.
|  
|      Lastly, this labels one test (dependent/should_fail/RAE_T32a)
|      as expect_broken, because the problem is really #12919. The
|      different handling of constraints in this patch exposes the error.
|  
|      This fixes bugs #11198, #12373, #13530, and #13610.
|  
|      test cases:
|      typecheck/should_fail/{T8262,T8603,tcail122,T12373,T13530,T13610}
|  
|  
|  >---------------------------------------------------------------
|  
|  8e15e3d370e9c253ae0dbb330e25b72cb00cdb76
|   compiler/typecheck/TcCanonical.hs                  | 296
|  ++++++++++++---------
|   compiler/typecheck/TcErrors.hs                     |  75 ++++--
|   compiler/typecheck/TcEvidence.hs                   |   8 +-
|   compiler/typecheck/TcFlatten.hs                    |  31 ++-
|   compiler/typecheck/TcRnTypes.hs                    |  30 ++-
|   compiler/typecheck/TcSimplify.hs                   |  34 ++-
|   compiler/typecheck/TcType.hs                       |  10 +-
|   compiler/typecheck/TcUnify.hs                      |  28 +-
|   compiler/types/Type.hs                             |   4 +-
|   testsuite/tests/dependent/should_fail/T11471.hs    |   2 +-
|   .../tests/dependent/should_fail/T11471.stderr      |  11 +-
|   testsuite/tests/dependent/should_fail/all.T        |   2 +-
|   testsuite/tests/gadt/gadt7.stderr                  |   6 +-
|   .../tests/ghci.debugger/scripts/break012.stdout    |  14 +-
|   .../indexed-types/should_fail/ExtraTcsUntch.stderr |   6 +-
|   .../tests/indexed-types/should_fail/T5934.stderr   |  13 -
|   testsuite/tests/polykinds/T12593.stderr            |  56 ++++
|   testsuite/tests/polykinds/T13555.stderr            |  21 +-
|   testsuite/tests/polykinds/T7438.stderr             |   6 +-
|   testsuite/tests/polykinds/T8566.stderr             |   2 +-
|   testsuite/tests/polykinds/T9017.stderr             |  10 +-
|   testsuite/tests/typecheck/should_fail/T12373.hs    |  10 +
|   .../tests/typecheck/should_fail/T12373.stderr      |   8 +
|   testsuite/tests/typecheck/should_fail/T13530.hs    |  11 +
|   .../tests/typecheck/should_fail/T13530.stderr      |   7 +
|   testsuite/tests/typecheck/should_fail/T13610.hs    |  11 +
|   .../tests/typecheck/should_fail/T13610.stderr      |  14 +
|   testsuite/tests/typecheck/should_fail/T5691.stderr |  10 +-
|   testsuite/tests/typecheck/should_fail/T7368.stderr |   6 +-
|   .../tests/typecheck/should_fail/T7368a.stderr      |   2 +-
|   testsuite/tests/typecheck/should_fail/T7453.stderr |  48 +---
|   testsuite/tests/typecheck/should_fail/T7696.stderr |   4 +-
|   testsuite/tests/typecheck/should_fail/T8262.stderr |   6 +-
|   testsuite/tests/typecheck/should_fail/T8603.hs     |   4 +
|   testsuite/tests/typecheck/should_fail/T8603.stderr |  13 +-
|   testsuite/tests/typecheck/should_fail/all.T        |   3 +
|   .../tests/typecheck/should_fail/tcfail090.stderr   |   4 +-
|   .../tests/typecheck/should_fail/tcfail122.stderr   |   8 +-
|   .../tests/typecheck/should_fail/tcfail123.stderr   |  13 +-
|   .../tests/typecheck/should_fail/tcfail200.stderr   |   6 +-
|   40 files changed, 537 insertions(+), 316 deletions(-)
|  
|  Diff suppressed because of size. To see it, use:
|  
|      git diff-tree --root --patch-with-stat --no-color --find-copies-
|  harder --ignore-space-at-eol --cc
|  8e15e3d370e9c253ae0dbb330e25b72cb00cdb76
|  _______________________________________________
|  ghc-commits mailing list
|  [hidden email]
|  https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fmail.h
|  askell.org%2Fcgi-bin%2Fmailman%2Flistinfo%2Fghc-
|  commits&data=02%7C01%7Csimonpj%40microsoft.com%7C9993301a429e467e29b00
|  8d4d4e6350f%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C6363675325842
|  62688&sdata=JkrPJNfzdjx2Rz%2FfcMq2HlIP%2BlsFIE%2FT68a%2FSUae%2Fzo%3D&r
|  eserved=0
_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [commit: ghc] master: Improve error messages around kind mismatches. (8e15e3d)

Richard Eisenberg-4

> On Jul 27, 2017, at 8:01 AM, Simon Peyton Jones via ghc-devs <[hidden email]> wrote:
>
> I think Richard just forgot to update the commit message.

Indeed. Bah. I don't suppose there's a way of fixing that now. Sorry!

Richard
_______________________________________________
ghc-devs mailing list
[hidden email]
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
Loading...