Changelog for bv-sized-1.0.4
Changelog for bv-sized
package
1.0.4 March 2022
- Deprecates trunc' and adds two alternatives, sresize and zresize
- Support for GHC 9.2
- BV is now a newtype
1.0.3 April 2021
- New instances (
NFData
,Random
) - New functions for
BV
that create uniformly random bitvectors - Fix: Adds
asBV
forSignedBV
(should have been there to begin with)
1.0.2 August 2020
- Allows tasty 1.3 for test suite
- Fixes bug in signedClamp function which made it possible to violate the nonnegative invariant on the internal representation of BVs
- Fixes divide by zero error in rotateL and rotateR
- Enhances test suite to test well-formedness of all operators that return a BV
- Fixes some documentation
1.0.1 May 2020
This fixed a subtle bug in the test suite which occasionally caused a divide by zero exception.
1.0.0 May 2020
This is a completely breaking change and it is completely incompatible with any previous use for this library.
- Bitvectors no longer track their own width. Every operations that relies on runtime awareness of the width (for instance, truncations) requires an expicit 'NatRepr' argument.
- Bitvectors do not support any typical instances you might hope for (Num, Bits, etc.). This is because they are not interpreted by default as signed or unsigned, so any class that requires such an interpretation is not supported. We do provide wrapper types that supply those instances when the bitvector width is known (SignedBV/UnsignedBV).
- Every operation has been renamed. Most are pretty straightforward (e.g. bvAdd ==> add).
- Several previously unsupported operations have been added (e.g. count leading zeros, conversion to/from bit/bytestrings)
- The App and BitLayout modules have been removed entirely. Both are potentially useful, but are out of date and probably should be in a different package anyway.
- New modules
- Data.BitVector.Sized.{Signed,Unsigned}: wrappers for BV that provide many instances
- Data.BitVector.Sized.Overflow: wrappers for operations that provide overflow information as part of the output
0.7.0 April 2019
- extractWithRepr now takes a NatRepr as an argument to specify the index, which it always should have.
- Updated to recent parameterized-utils hackage release, which fixes the build failures in the previous bv-sized release.
0.6.0 March 2019
- changed WithRepr functions to '
- added Num, Bits instances
- bitVector now takes arbitrary Integral argument
- add 'bitLayoutAssignmentList' function (see haddocks for details
- Hid BV constructor, exposed BitVector as pattern
0.5.1 August 2018
- fixed github URL
0.5.0 August 2018
- Added a lot of better support for the App module, including a type class for embedding BVApp expressions along with associated smart constructors
0.4.0 April 2018
- Added App module for BitVector expression languages and evaluation
0.3.0 April 2018
- fixed bug with bvShiftR, it was doing a left shift!
- Division now rounds to zero for negative integers, bvDiv -> bvQuot
- added Ix instance for BitVector w
- added bv0, zero-width vector
- bvConcatMany, bvGetBytesU (conversion to/from list of bytes)
0.2.1 March 2018
- bvMulFSU
- bvDivU, bvDivS
- Added Read instance, fixed Show to be compatible. Using prettyclass for pretty printing. (I guess this is semi-breaking, but whatever.)
0.2 March 2018
- bv -> bitVector, so this is very much a breaking change
- bvShiftL, bvShiftRL, bvShiftRA
- bvLTU, bvLTS
0.1.1.1 March 2018
- added BitLayout
0.1.1.0 March 2018
- added functions
bvMulFS
/bvMulFU
for full bitvector multiplication without truncation - removed Internal module, now export all those functions in Data.BitVector.Sized
- fixed the bv*WithRepr functions, which were not truncating the inputs properly
0.1.0.0 March 2018
- First release