ox-arrays - Nested, compositional struct-of-arrays orthotope arrays

	Commit message (Collapse)	Author	Age
*	Improve an error message	Tom Smeding	4 days
\|
*	Compatibility liftVEltwise1 (TODO remove)	Tom Smeding	9 days
\|
*	Some re-exports for compatibility	Tom Smeding	9 days
\|
*	Separate arith routines into a library	Tom Smeding	9 days
\| \| \| \|	The point is that this separate library does not depend on orthotope.
*	Arith statistics collection from C	Tom Smeding	10 days
\|
*	arith: Don't FFI-import unused dotprod_*_strided ops	Tom Smeding	10 days
\|
*	Optimise reductions and dotprod with more vectorisation	Tom Smeding	2025-03-14
\| \| \| \| \| \| \| \| \|	Turns out that if you don't supply -ffast-math, the C compiler will faithfully reproduce your linear reduction order, which is rather disastrous for parallelisation with vector units. This changes the summation order, so numerical results might differ slightly. To wit: the test suite needed adjustment.
*	arith: Correct rank arguments to C wrapper functions	Tom Smeding	2025-03-13
\|
*	Export IntElt	Tom Smeding	2025-03-13
\|
*	Rename arithPromote* to lift* (unrelated to arith ops, really)	Tom Smeding	2025-03-13
\|
*	Add atan2	Tom Smeding	2025-03-13
\|
*	Implement quot/rem	Tom Smeding	2025-03-13
\|
*	Remove redundant constraints on arith instances	Tom Smeding	2025-03-12
\|
*	Slight cleanup of arith instances for Mixed/Ranked/Shaped	Tom Smeding	2025-03-12
\| \| \| \|	Reformulate error messages; add NumElt superclass to FloatElt
*	Binary ops without normalisation	Tom Smeding	2025-03-12
\| \| \| \| \| \| \| \| \|	Before: > sum() Double [1e6] stride 1; -1: OK > 68.9 ms ± 4.7 ms After: > sum() Double [1e6] stride 1; -1: OK > 1.44 ms ± 50 μs
*	More consistent equality functions on shapes	Tom Smeding	2025-03-11
\|
*	arith: Unary float ops on strided arrays without normalisation	Tom Smeding	2025-03-05
\|
*	All arrays are Eq and Ord (they compare shape first, then values)	Tom Smeding	2025-02-18
\| \| \| \| \|	This comparison policy is lifted straight from orthotope (and indeed the instances are all just auto-derived).
*	Less pointless standalone Ord instances	Tom Smeding	2025-02-18
\|
*	arith: Fix unary op stride bugs	Tom Smeding	2025-02-18
\|
*	Fix a typo in an error message	Mikolaj Konarski	2025-02-17
\|
*	arith: Only strided unary int ops	Tom Smeding	2025-02-16
\| \| \| \| \|	This should have negligible overhead and will save a whole bunch of C code duplication when the FUnops are also converted to strided form.
*	arith: Unary int ops on strided arrays without normalisation	Tom Smeding	2025-02-16
\|
*	arith: Fix cover calculation in stridesDense	Tom Smeding	2025-02-16
\| \| \| \| \|	The calculation was incorrect for windowed arrays (as produced by RS.window)
*	Remove 'type data' for now (GHC 9.10 breaks)	Tom Smeding	2025-02-15
\|
*	mcastSafe, castCastable	Tom Smeding	2025-02-01
\|
*	Eliminate the lame shaped fromInteger implementation	Mikolaj Konarski	2025-01-24
\|
*	Export IShX from Nested	Mikolaj Konarski	2025-01-09
\|
*	Export ListX from Nested	Mikolaj Konarski	2025-01-07
\|
*	Add some spuriously missing Eq/Generic instances	Tom Smeding	2024-12-18
\|
*	Uniformise NFData instance (by putting rnf in Elt)	Tom Smeding	2024-12-18
\| \| \| \| \| \|	This now depends on: https://github.com/augustss/orthotope/pull/14 My sincere apologies.
*	Remove some missed overlapping Show (Mixed) instances	Tom Smeding	2024-12-17
\|
*	Uniform Show instance for Mixed	Tom Smeding	2024-12-17
\| \| \| \|	The previous code was needlessly complicated.
*	rcastToMixed	Tom Smeding	2024-12-11
\|
*	Simpler API to mcast	Tom Smeding	2024-12-11
\|
*	Fix scastToMixed	Tom Smeding	2024-12-11
\| \| \| \| \|	The previous implementation failed when any of sh' was Unknown, which makes no sense.
*	Improve scastToMixed's implementation	Tom Smeding	2024-11-29
\|
*	scastToMixed, weakly-typed version of stoMixed	Tom Smeding	2024-11-29
\|
*	[mrs]{un,}zip	Tom Smeding	2024-11-22
\|
*	Fix X.empty	Tom Smeding	2024-11-22
\| \| \| \|	S.constant always evaluates its argument.
*	Improve error message for X.empty	Tom Smeding	2024-11-22
\|
*	[rsm]emptyArray	Tom Smeding	2024-11-21
\|
*	Add the criminally absent singletons to numEltMinIndex and numEltMaxIndex	Mikolaj Konarski	2024-11-19
\|
*	Improve error message	Tom Smeding	2024-11-14
\|
*	Decent, consistent Num for Shaped, but with type classes :(	Tom Smeding	2024-11-14
\|
*	snatMinus	Tom Smeding	2024-11-13
\|
*	Add withKnownSh{S,X}	Tom Smeding	2024-11-13
\|
*	Incoherent Shaped '[] instances (Num/Fractional)	Tom Smeding	2024-11-13
\|
*	Remove now-unused import :D	Tom Smeding	2024-11-12
\|
*	Make fromInteger error for all array types	Tom Smeding	2024-11-12
\| \| \| \|	They made no sense anyway, and horde-ad should now not use them any more