Better treatment of signatures in cls/inst
authorSimon Peyton Jones <simonpj@microsoft.com>
Thu, 30 Jul 2015 14:06:55 +0000 (15:06 +0100)
committerSimon Peyton Jones <simonpj@microsoft.com>
Thu, 30 Jul 2015 14:35:22 +0000 (15:35 +0100)
commit72d23c3e0244b1163d7806e40128ad51cc959f7f
treeb9ac60715798fc6571e170d5c41b80854f148d3f
parent2e33b9c87615d3cc26ad188d5ca565e8a981208e
Better treatment of signatures in cls/inst

The provoking cause for this patch is Trac #5001, comment:23.  There
was an INLINE pragma in an instance decl, that shouldn't be there.
But there was no complaint, just a  mysterious WARN later.

I ended up having to do some real refactoring but the result is,
I think, simpler and more robust.
14 files changed:
compiler/rename/RnBinds.hs
compiler/rename/RnEnv.hs
compiler/rename/RnSource.hs
compiler/typecheck/TcBinds.hs
compiler/typecheck/TcDeriv.hs
testsuite/tests/module/mod48.stderr
testsuite/tests/partial-sigs/should_fail/WildcardInInstanceSig.hs
testsuite/tests/partial-sigs/should_fail/WildcardInInstanceSig.stderr
testsuite/tests/patsyn/should_fail/T9705-1.stderr
testsuite/tests/patsyn/should_fail/T9705-2.stderr
testsuite/tests/rename/should_fail/T5001.hs [new file with mode: 0644]
testsuite/tests/rename/should_fail/T5001.stderr [new file with mode: 0644]
testsuite/tests/rename/should_fail/all.T
testsuite/tests/typecheck/should_fail/tcfail021.stderr