[GHC] #8809: Prettier error messages?
GHC
ghc-devs at haskell.org
Fri Jul 3 08:59:56 UTC 2015
#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:
Type of failure: None/Unknown | Unknown/Multiple
Blocked By: | Test Case:
Related Tickets: | Blocking:
| Differential Revisions:
-------------------------------------+-------------------------------------
Comment (by bgamari):
goldfire, indeed the an ADT-per-compiler-phase is exactly what I was
thinking (and I have the beginnings of a branch looking at `TcErrors` in
particular. In my case though, I was thinking of at least starting by
merely annotating a few semantically-important elements of the message
(e.g. `Name`s, `Type`s, `TyVar`s, etc.). This would enable, for instance,
allow links to the definition span of a symbol, printing an expanded
representation of a type, etc.
That being said, there is no reason why one couldn't go further with this
same approach and encode the entire error as a value. This certainly
offers further advantages, although also implies a bit more work (which is
why I'm starting with the atoms listed above).
As far as the indexing issue goes, I was thinking we would give `Doc` a
`Monad` instance. This would allow a number of quite convenient patterns.
For instance, have `msgs :: Doc TcErrDoc` containing some errors you'd
like to print: If you have `pprTcErrDoc :: TcErrDoc -> Doc Void`, you
could trivially flatten the document with `msgs >>= pprTcErrDoc`.
Further if you want to combine a `Doc TcErrDoc` with a `Doc ParserErrDoc`,
you'd simply lift them both into an ADT `data GhcErrDoc = TcErrDoc
TcErrDoc | ParserErrDoc ParserErrDoc` with `Applicative`. Alternatively,
if you'd rather keep the universe of error types open, you could opt to
lift them into a universally quantified `newtype`, roughly like you
suggest.
--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8809#comment:10>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
More information about the ghc-tickets
mailing list