I am using the new existential datatype support (great!) - but now when I patch in fundep support (more or less as I posted before) adding an extra field to ClassD, I have to modify GHC to use the extra field in ClassD... Now when I include a definition: [d| class Test x |] I get the error: ghc-6.3: panic! (the `impossible' happened, GHC version 6.3): deSugar/DsMeta.hs:925:4-41: Irrefutable pattern failed for pattern [elt_ty] Any idea why this might be happening (why is the error in DsMeta, when the TH stuff is in hsSyn/Convert.lhs and typecheck/TcSplice.lhs? Keean.
On Wed, Nov 17, 2004 at 05:12:38PM +0000, Keean Schupke wrote:
I am using the new existential datatype support (great!) - but now when I patch in fundep support (more or less as I posted before) adding an extra field to ClassD, I
I've just committed support for this - sorry for the delay.
Now when I include a definition:
[d| class Test x |]
I get the error:
ghc-6.3: panic! (the `impossible' happened, GHC version 6.3): deSugar/DsMeta.hs:925:4-41: Irrefutable pattern failed for pattern [elt_ty]
Any idea why this might be happening (why is the error in DsMeta, when the TH stuff is in hsSyn/Convert.lhs and typecheck/TcSplice.lhs?
The bit in TcSplice is for the reify keyword. DsMeta handles the conversion from GHC's internal representation to the TH representation. Thanks Ian
participants (2)
-
Ian Lynagh -
Keean Schupke