GitLab

This fixes #9355.

in order to make its RULES semantics preserving. This fixes #9495.

...several modules in `base` recently touched by me

as spotted by David Feuer and reported as #9585.
The comment was right until ceb68b91,
which is only 13 years ago :-).

This is a follow-up commit to
e428b5b8 (refs D195 & #9550)

Summary:
As discussed in
http://www.haskell.org/pipermail/libraries/2014-July/023314.html and
submitted at #9550.

Test Plan: Submit to phab, see what happens.

Reviewers: austin

Subscribers: simonmar, ezyang, carter

Differential Revision: https://phabricator.haskell.org/D195

GHC Trac Issues: #9550

Summary:
Make last a good consumer simply by implementing it as foldl. This fixes Trac: #9339.
Thanks to David Feuer for bringing it up.

Test Plan: perf/should_run/T9339 + general validation

Reviewers: austin

Reviewed By: austin

Subscribers: phaskell, simonmar, relrod, carter

Differential Revision: https://phabricator.haskell.org/D86

Trac Issues: #9339

together with the call arity analysis and the following patch (about inlining
maximum), we get nice benefits from fusing foldl and foldl' with good
producers:

                Min          -0.1%    -74.5%     -6.8%     -8.3%    -50.0%
                Max          +0.2%      0.0%    +38.5%    +38.5%      0.0%
     Geometric Mean          -0.0%     -4.1%     +7.7%     +7.7%     -0.8%

Because this depends on a compiler optimisation, we have to watch out for cases
where this is not an improvements, and whether they occur in the wild.

This helps with the changes in #7994, but might also generally be a good
idea (ignore the runtime):

--------------------------------------------------------------------------------
        Program           Size    Allocs   Runtime   Elapsed  TotalMem
           fft2          -0.1%     -1.5%      0.07      0.07     +0.0%
       fibheaps          +0.0%    -17.2%      0.03      0.03     +0.0%
          fluid          +0.5%     -0.7%      0.01      0.01     +0.0%
      integrate          +0.0%     -0.9%      0.16      0.16     +0.0%
        rewrite          +0.0%     -1.1%      0.02      0.02     +0.0%
--------------------------------------------------------------------------------
            Min          -0.1%    -17.2%     -1.6%     +0.0%     +0.0%
            Max          +0.5%     +0.0%   +107.7%   +106.2%    +11.3%
 Geometric Mean          +0.0%     -0.2%    +23.7%    +23.9%     +0.1%

The now obsolete (and redundant) `#hide` pragmas have been superseded by
`{-# OPTIONS_HADDOCK hide #-}` pragmas which are used by most of the
affected modules anyway.

This commit also adds proper `{-# OPTIONS_HADDOCK hide #-}` pragmas to
`GHC.Desugar` and `GHC.IO.Encoding.Iconv` which had only the ineffective
`#hide` annotation.
Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

Trac #876 is the oldest ticket I have fixed in a long time.
I finally figured out how to make foldr behave in a non
space-leaky way for length.

Thanks to Andy for re-opening.

Add explicit {-# LANGUAGE xxx #-} pragmas to each module, that say
what extensions that module uses.  This makes it clearer where
different extensions are used in the (large, variagated) base package.

Now base.cabal doesn't need any extensions field

Thanks to Bas van Dijk for doing all the work.

It gets 2 arguments in the "zipWith" rule but its arity was higher and the new
inliner didn't inline it sometimes, for instance here:

mpp ::  [Double] -> [Double] -> [Double] -> [Double] -> [Double]
mpp as bs cs ds = zipWith (*) (zipWith (+) as bs) (zipWith (+) cs ds)
 
This was a regression vs. 6.10.

GHC.Base SOURCE-imports GHC.Err, and re-exports 'error'.  So 
other modules need only import GHC.Base.

This doesn't change the fact that these other modules are all compiled
before GHC.Err, so they are all part of the module loop that starts with
GHC.Base and finishes with GHC.Err.  But it does reduce the occurrence
of those SOURCE imports.

Make length into a good consumer.  Fixes Trac bug #707.

(Before length simply didn't use foldr.)

Make length into a good consumer.  Fixes Trac bug #707.

(Before length simply didn't use foldr.)

This allows take to be deforested, and improves performance of
replicate and replicateM/replicateM_.  We have a separate problem that
means expressions involving [n..m] aren't being completely optimised
because eftIntFB isn't being inlined but otherwise the results look
good.  

Sadly this has invalidated a number of the nofib benchmarks which were
erroneously using take to duplicate work in a misguided attempt to
lengthen their runtimes (ToDo).

hide GHC internals from Haddock