1.3.14

• Use more compact heap object representation which saves 16 bytes on 64bit platforms.
• Add toWords64/fromWords64 functions

1.3.13

• Optimize V4.nextRandom (~3x speed increase)
• Optimize UUID V3 & V5 generation (~2x speed increase)
• Use cryptohash-md5/cryptohash-sha1/entropy instead of memory/cryptonite for better performance and stability, but GHCJS is now no longer supported.
• Update cabal-spec to version 1.10

1.3.12

• Update package dependencies
• Use cryptonite for crypto
• The function V4.nextRandom is now implemented with functions from Crypto.Random in package cryptonite. This does slow-down random UUID generation but provides for far greater randomness. The Random instance for UUID can be used to re-gain the old behavior.

1.3.11

• Add toText/fromText

1.3.10

• Update dependencies in tests and benchmarks.

1.3.9

• Split definition of UUID data type into separate Cabal package to enable 3rd party libraries to avoid some of the larger dependencies.

1.3.8

• Allow building against newer deepseq.

1.3.7

• Allow building against newer time.

1.3.6

• Move to GitHub.

1.3.5

• Allow building against newer random.

1.3.4

• Allow building against newer QuickCheck.

1.3.3

• More complex version constraints on hashable, to avoid building against versions less than 1.1.1.0 or equal to 1.2.0.*.

1.3.2

• Fix for building against bytestring 0.9.*

1.3.1

• Allow building against hashable 1.1.* in addition to 1.2.*

1.3.0

• New functions for parsing and printing UUIDs to and from ASCII BytesStrings
• New module Data.UUID.Util. This module includes the type UnpackedUUID, whose fields correspond to the UUID fields described in RFC 4122.
• The Storable instance now stores a UUID in host byte-order instead of big endian.
• There is now an instance for Hashable UUID.

1.2.13

Benchmark only changes:

• Allow criterion 0.8.*

1.2.12

Test only changes:

• Allow QuickCheck 2.6.*

1.2.11

• Allow binary 0.7.*

1.2.10

• Allow cryptohash version 0.9.*
• Cleanup tests

1.2.9

• Bumped dependency on cryptohash.

1.2.8

• Bumped various dependencies and cleaned up dependencies in general.

1.2.7

• Added stricter constraints on random package.

1.2.6

• Add module V4 to direct attention to our Random instance

• In module V1 seed the generator with a random number if the hardware MAC address could not be discovered.

• Fix and cleanup various haddocks.

• In module docs, warn about MD5 use in Data.UUID.V3 and encourage the reader to use Data.UUID.V5 instead.

1.2.5

• Use cryptohash package for MD5 and SHA1 instead of Crypto

1.2.4

• Unpack Word32 values into UUID constructor.

• Update test suite to QuickCheck 2

• Bump other dependencies in tests/benchmarks

1.2.3

• The Read instance now drops leading spaces in the string to be parsed. Thanks to Marc Ziegert for reporting this bug.

• The tests have moved over to the new Cabal test running framework.

1.2.2

• Add functions fromWords/toWords

The goal was to have a total function that can be used to construct a UUID, primarily for use by uuid-th.

1.2.1

• Fix concurrency bug in Data.UUID.V1 (thanks to Neil Mitchell for reporting and a test case)

1.2.0 (Contributors: Antoine Latter & Mark Lentczner)

• added functions toByteString and fromByteString
• added nil UUID
• added unit tests and benchmarks, built when configured -ftest
• major speed up of to/from functions (as well as in general)
• added version-3 generation (deterministic based on MD5)
• major changes to internal representation
  • now uses four strict Word32 values
  • internal ByteSource classes for easy construction (see Builder.hs)
• Storable instance now stores in memory as system libraries in C do: 16 bytes derived from the network order of the fields, no matter what the host native endianess is.
• fixed bugs in V1 time and clock stepping, and V1 generated values
• builds cleanly under GHC's -Wall
• added CHANGES file

1.1.1

• no longer exporting null from the prelude
• add null predicate on UUIDs
• documentation fix (thanks Mark Lentczner)