| Safe Haskell | Safe-Inferred |
|---|---|
| Language | Haskell2010 |
Evaluation.Builtins.Costing
Contents
deriving stock Eq. We don't want to do the
latter, because the semantics of a CostStream are those of the sum of its elements and the
derived SatInt belongs. E.g.>>>toRange 42(11,100)>>>toRange 1234(1001,10000)
chooseSatInt :: (addCostStream equals the sum of the sums of its two arguments.
test_minCostStreamIsMin :: minCostStream equals the minimum of the sums of its two arguments.
test_zipCostStreamIsZip :: ExBudget containing the sums of its two arguments.
test_mapCostStreamReasonableLength :: addCostStream equals the sum of the lengths of its two arguments.
test_minCostStreamReasonableLength :: zipCostStream equals the maximum of the lengths of its two arguments.
test_mapCostStreamHandlesBottom :: minCostStream preserves the laziness of its two arguments.
postAlignWith :: a -> [a] -> [a] -> ([a], [a]) #
Pad the shortest of the given lists by appending the given element to it until the length of the result matches the length of the other list.
>>>postAlignWith 'a' "bcd" "e"("bcd","eaa")>>>postAlignWith 'a' "b" "cdef"("baaa","cdef")
test_zipCostStreamHandlesBottom :: CostRose #
Return a finite balanced tree with each node (apart from the leaves) having exactly 3 children. The parameter is the depth of the tree. Named after https://en.wikipedia.org/wiki/Sierpi%C5%84ski_triangle
test_flattenCostRoseIsLinearForSierpinskiRose :: TestTree #
Traverse a sierpinskiRose of the given depth and display the total amount of elements
processed. See test_flattenCostRoseIsLinear for why we do this.
test_flattenCostRoseIsLinear :: CostRose takes _linearly_ more time. The actual test can only be done with eyes unfortunately, because the tests are way too noisy for evaluation times to be reported even remotely accurately.
genCostRose :: CostRose from the given list by splitting the list into sublists and generating
a memoryUsage called over a value of a built-in type never returns a stream
containing a negative cost.
test_costing :: SatInts with exponentially lower chance. This is in order to
make the generator of CostStream produce streams whose sums are more or less evenly distributed
across magnitudes.