GitLab

This was causing the bad behaviour in Trac #7507,
because 'sum' wasn't getting specialised to Int64.

It also deals with Trac #4321, which had the same cause.

This has a big effect on some nofib programs too:

--------------------------------------------------------------------------------
        Program           Allocs   Runtime   Elapsed  TotalMem
------------------------------------------------------------------------
     bernouilli            -2.6%     -2.0%     -2.0%     +0.0%
           fft2           -23.8%      0.09      0.09    -16.7%
          fluid            -4.4%      0.01      0.01     +0.0%
         hidden            -3.2%     +2.1%     +1.8%     +0.0%
      integrate           -38.0%    -47.7%    -47.7%     -1.0%
           x2n1           -30.2%      0.01      0.01    -50.0%
------------------------------------------------------------------------
            Min           -38.0%    -47.7%    -47.7%    -50.0%
            Max            +0.4%    +11.2%    +11.8%     +6.9%
 Geometric Mean            -1.3%     +0.2%     +0.2%     -0.8%

They are now used by TcGenDeriv

Replaced the previous implementation due to licensing concerns. The new
implementation is a reimplementation by Clark Gaebel, based on the
public domain implementation at
http://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2

We decided not to break existing users without providing an easy
migration path.  For the time being I've made these implementations,
which fix the bugs in the old versions and perform reasonably well.

In due course we should move the concurrency functionality, including
these modules, out of base and into a separate package.

Right now, we only have

data AsyncException
  = StackOverflow
  | HeapOverflow
  | ThreadKilled
  | ...

so it is not possible to add another async exception.  For instance,
the Timeout exception in System.Timeout should really be an async
exception.

This patch adds a superclass for all async exceptions:

data SomeAsyncException = forall e . Exception e => SomeAsyncException e
  deriving Typeable

and makes the existing AsyncException and Timeout children of
SomeAsyncException in the hierarchy.

With the direct top level recursion, `until` cannot be inlined by GHC, so no strictness
analysis and consequently no unboxing occurs at the call site.
With the recursion delegated to the worker with the condition and update function
static arguments, it can be inlined, and strictness analysis can happen.

AndreasVoellmy authored Oct 17, 2012 and ian@well-typed.com committed Dec 02, 2012

This allows the calling thread to unregister interest in the file, e.g. in the case
of an exception in the thread.

Needed for #7342

We now keep the HEAD version numbers as values which would be suitable
for immediate release.

Patch from dreixel.

This changes the output of throwGetLastError to include the system error
message, rather than the message of our fictitious errno.

It also adds several definitions to GHC.Windows, mostly from the Win32 package.
The exceptions are:

 * getErrorMessage: returns a String, unlike in System.Win32.Types,
   where it returns an LPWSTR.

 * errCodeToIOError: new

 * c_maperrno_func: new

It no longer refers to a particular version of the language standard.

It now includes a note about how it is implicitly imported, and
it looks like the other documentation from the language report
is already included in the haddock docs.