Current known failures
~~~~~~~~~~~~~~~~~~~~~~

deSugar/should_compile/ds052
		-- inaccuracy in non-exhaustive pattern warnings

lib/IO/hReady001
		-- IO.hReady doesn't detect the end of a file

lib/IO/openFile005, openFile007, readFile001 [Windows]
                -- file locking is broken on Windows

lib/IOExts/hTell001, hTell002 [Windows]
		-- seeking on text handles not implemented on Windows

lib/net/net001 [Windows]
                -- getProtocolByName seems not to be working on Windows

lib/system/system001 [Windows]
		-- probably a failure due to System.system using Cmd.exe?

lib/should_run/enum04
		-- enumFrom on basic numeric types should be strict
		-- (possibly a bug in the Haskell Report: it specifies that
		-- these ops should be strict in the section on Enum, but the
 		-- sample code in the Prelude doesn't agree, at least for
		-- Float and Double).

parser/should_fail/read002
		-- we don't disallow certain characters in strings (eg. newline)

parser/should_fail/read024
		-- parser bug related to layout processing

rename/should_fail/rnfail034
		-- duplicated "in" in an error message

typecheck/should_fail/tcfail080
		-- ambiguity test: ambiguity isn't detected early enough
		-- (behaviour is sound but could be more helpful, according
		-- to SPJ)

typecheck/should_run/tcrun007
		-- tests for generics, which are currently broken in 5.02

programs/okeefe_neural
		-- causes the compiler to run out of heap in the simplifier
		-- (problem with large constant data structures?)

-- These tests fail the 'optasm' way, due to inconsistent floating
-- point results on x86.  The inconsistency is caused 

numeric/should_run/arith008
numeric/should_run/arith012
programs/barton_mangler_bug
programs/cholewo-eval
programs/galois_raytrace
