[GHC] #8809: Prettier error messages?

[GHC]

#8809: Prettier error messages?
-------------------------------------+-------------------------------------
        Reporter:  joelteon          |                Owner:
            Type:  feature request   |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  7.9
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:  13122
 Related Tickets:                    |  Differential Rev(s):
  #8809,#10073,#10179,#12906         |
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by bgamari):

 alanz, indeed much of the motivation for providing these annotations to
 begin with is to enable tooling like Idris'.

 mpickering, your assessment is pretty much dead on. I agree that we can
 and should move ahead with (1) independently from (2). `ErrMsg` already
 includes the information that we would need to do this; all that needs to
 be done is to ensure that we don't lose it and potentially add some output
 format for tooling consumers (e.g. JSON).

 However, I think we should refrain from trying to solve the weakened form
 of (2) that comment:52 tackles. This approach appears to introduce a lot
 of accidental complexity for what is easily accomplished using an existing
 annotated pretty-printing library. The only thing holding us back from
 moving ahead here is #10735.

 As far as I can tell, the problem in #10735 is that both the
 [[https://github.com/ekmett/wl-pprint-extras/issues/16|Wadler-Leijen]] and
 [[https://github.com/haskell/pretty/issues/32|Hughes]] pretty-printers can
 become quadratic in some cases. This hurts a great deal in the case of
 GHC, where we use the pretty-printer to emit massive quantities of code.
 Mpickering has advocated that we work around the issue by simply
 eliminating the use of the pretty-printer for code generation. I would
 prefer that we not do this if possible. Ultimately I believe that the root
 cause of the non-linearity is back-tracking due to the document exceeding
 the ribbon width. However, when we are producing code we configure
 `pretty` with an infinite ribbon width, meaning that it should never be
 necessary to backtrack. Unfortunately the `pretty` implementation doesn't
 take advantage of this fact. I argue that this should be fixed.

 The other related issue is the inability to make use of `FastString` with
 the upstream `pretty` since the `text` combinator has type `String ->
 Doc`. This is actually a very poor interface as `text` needs the length of
 the string, which is an O(n) operation. Really it should at least allow
 usage of a more efficient string representation. niteria and I discussed
 this a few months ago and agreed that this could be resolved by simply
 generalizing `text` over its string type with a typeclass providing an
 `unpack` and `length` function.

 goldfire, does any of the above sound like a reasonable task for your
 student?

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8809#comment:60>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler