GitLab

The report says that it isn't meant to.

They now get read as Ident's, and the Read Double/Float instances
(via convertFrac) handle that Ident specially.

This doesn't change the behaviour yet, but I think it's a step in the
right direction.

This eliminates a SOURCE import

I think it was all to work around problems in old versions of haddock.

Bytestring uses it, so I've moved it into that package.

This is necessary now that we no longer allow CTYPE pragmas on type
synonyms.

Now that we have the CTYPE pragma we can do this in such a way that
it doesn't break the build on OS X.

This makes exceptions more predictable.

Added testBit, bit, and popCount to the required set.  They no longer
have default implementations as the Num constraint was removed from the
Bits class.

Also add an INLINABLE pragma so that the function can be specialized at
the call site.

gwern authored Nov 27, 2011 and Simon Marlow committed Feb 13, 2012

System.Environment: improve 'getEnv' doc to mention closest 'setEnv' we have ('System.Posix.putEnv')

Added missing Functor, Applicative, Alternative and MonadPlus instances Added Applicative and Alternative instances for ReadP and ReadPrec Added Functor, Applicative, Alternative and MonadPlust instances for ArrowMonadx

This test checks that calling trig functions via the FFI gives the
same results as the Prelude versions.  But it uses an extreme value to
test: 1e20, and on Windows this gives slightly different results (for
unknown reasons).  However, using less extreme values gives reasonable
results, so I've added more values to the test to check that the
discrepancy is limited to the extreme - indeed it first goes wrong
around 1e19, values below that seem to be fine.