LLNL Units Versions Save

Fix some string output issues related to units with various prefixes of meters and seconds, and add a UNITS_CMAKE_PROJECT_NAME variable to cmake to allow the CMAKE project name to be customized.

Fixed

• fixed an issue with string output related to mm/s outputting as mHz*m and mm/s^2 outputting as Gy/km #308

Added

• Added a CMAKE variable to change the project name, to resolve some packaging conflicts. UNITS_CMAKE_PROJECT_NAME #310

A few user suggested tweaks, and support additional unit string conversions supported by the google unit converter. Cleaner output for Hz and prioritization of SI base units in the to_string operation.

Changed

• Modified the eflag to work a little differently than the iflag and moved a couple units to use the eflag instead to support things like sqrt(degC.pow(2))=degC #275
• Updated copyright date to 2023 #266
• Updated to new version of google test and added some additional units and test from the google unit conversion #283
• Fixed and corrected some documentation to match the latest code #284
• Altered the match_flags to 64 bits and corresponding updates to the flag constants#294
• adjusting the partitioning algorithms to have a minimum size and have some controlling flags #294,#297
• split the non-english units into a separate array that can be controlled independently in the build #296

Fixed

• fix fuzzing issue with arb. unit #275
• Format arb.unit nicely and fix a hardcoded namespace #272
• Fix quad unit, and clean up some unit conversion strings involving Hz #265
• Update clang tidy and fix resulting issues #267
• Fixed an issue where scalar unit values in measurements using to_string could cause a crash #289
• Fixed some issues with prioritization of base units in the string generation #303

Added

• Added multiplies and divides operations which take into account perunit operations #299
• Added application notes on strain #299
• Added support for addition in unit string, the units on both sides must have the same units #298
• Updated coverage to hit 100% again #291
• Added several unit strings and cleaned up the typography units #282
• Added a dependabot configuration for updating the actions #274
• Added removeUserDefinedUnit method #273
• Added `addUserDefinedOutput method to specify an output string for a unit #270
• Added support for robi and qubi prefixes

Added several math operations for units, restored coverage to 100%, added uncertain constants and cleaner string generation and interpretation around the use of '.' and uncertain measurements, and added support for new SI prefixes.

Changed

• Changed the interpretation of '.' in some unit strings to allow more consistent use as abbreviation #250
• Made the string generation for uncertain measurements obey rules for significant figures more closely #249

Fixed

• Get code coverage back to 100% #256, #254, #253, #252
• Fixed hard coded namespaces in unit_conversion maps #241
• Fixed a bug found by fuzzing #240

Added

• math operations from the standard library including: trunc, ceil, floor, round, fmod, sin, cos, tan. #235
• added support for new SI prefixes: rotta, quetta, ronto, quecto. #255
• Added a number of uncertain constants #248

This release has a number of fixes for specific unit strings, adds some other units and a new unit_conversion_maps structure to reduce stack usage.

Changed

• Convert default branch to "main"
• The string maps are new defined in a constexpr array in units_conversion_maps.hpp this reduces the stack usage significantly and clears up some warnings about that on some compilers [#217][],#226
• The country specific codes for China and Canada were changed from "ch" and can to cn and ca to match country top level domains #229

Fixed

• The "hartree" string produced an incorrect value #163
• Fixed an issue with inverse multipliers for inverse conversions #166
• changed the string conversions to better align with SI recommendations #173
• Fix an issue with the string "0.2" #188
• Fix asterisk operator typo #194
• Update the docker images to use more recent base images #206,#212
• User defined units were not being used in the string conversion operations properly #213
• Update third party library links #216
• Clean up some if constexpr warnings on some platforms #227
• Fix an issue of misinterpreting short strings with spaces such as kg m #224

Added

• Added a series of constants to make the available constants match the Nist Common Constants list #163
• added "Eh" as a valid string value for "hartree" #163
• Added support for larger power factors in units when using larger base #184
• Added support for some climate related units #210,#179
• Added mass to weight conversions #229
• added domains to allow some domain specific units that might mean different things in different contexts #173

This release includes some changes to the CMake builds and Targets. Also includes support for 64 bit base types for units with expanded power support, and support for a custom namespace for better integration with other applications. And additional units for natural gas and particle physics.

Changed

• update the physical constants to latest values and updated related documentation on Physical constants [#56][]
• CMake targets units::units and units::header_only are the only two public targets [#155][]
• Updated code checks [#85][]

Fixed

• Fixed -Wshadow warnings for GCC 4.9 and earlier [#66][]
• Documentation Fixes and updates [#152][], [#146][], [#111][], [#82][]
• Fixed pow_overflows not handling 0 correctly [#139][], [#134][]
• Fixed a few issues with the output strings and some SI units [#130][]
• Some cleanup for multiplier string generation [#93][]
• Fixed an issue with plural two character units like lb, oz [#91][]
• Fix identified Fuzzing issue [#84][]
• Fixed some shadow warnings on some compilers [#72][]
• Fixed some issues with std::abs usage [#70][]

Added

• Added version output to converter app and the correct version info on the webserver [#54][]
• Added markdown linter to pre-commit [#64][]
• CMake format check to pre-commit [#65][]
• Added tests for the shared library build [#155][]
• Added support for 64 bit base unit [#150][], [#142][]
• CMake Variable for manipulating output namespace [#136][], [#141][]
• Added some support for compiling with C++20 [#125][]
• Added some additional units for supporting particle physics [#120][],[#115][], [#112][], [#109][]
• Added support for exposing the main unit maps for string conversions [#113][]
• Added support for natural gas related units [#74][]

Removed

• Travis-CI builds have been migrated to Azure and Circle-CI [#159][]
• removed units::units-static, and units::units-shared CMake targets[#155][]
• Removed Becquerel from the generated unit strings as it was producing some very odd units at times [#82][]

Add a converter command line application and fix a few slow conversion issues and some other fuzz issues that came up recently, add isnormal operation for measurements, better test coverage for fixed_precise_measurement, some additional tests and static analysis checks.

Changed

• Added several tests run under Azure to remove deprecated image and add some new tests #40
• clears up several warnings from clang-tidy #41
• the fuzzer now uses fuzz_measurement #42
• update clang format to limit line length to 80 and allow reflowing comments #43
• add clang-tidy checks for the tests and fix a few issues. #47

Fixed

• A few timeouts on the fuzzer- the fix was to generalize the multiplier insertion after ^ to accept multiple digits after the ^ instead of just ignoring it after more than one. #34
• An asymmetry was observed in the unit equality from on the fuzzers, this resulted in some modifications of the cround_equal and cround_precice_equals functions. Also noted that the functions weren't aborting on exact floating point equality so were doing quite a bit of extra calculations. #34
• A timeout issue from fuzzing having to do with not injecting multiplies after [] in some circumstances. The fix was to be a little more refined as to which point to not inject the * and to do it in multiple stages so as to not rely on the partitioner so much. #35
• fixed_measurement and fixed_precise_measurement had incorrect subtraction operation in a few overloads. #39
• fixed a few initial issues from fuzz_measurement #42
• Some more fuzzing generated issues with cascading powers #45
• A number of additional clang-tidy checks were added and the resulting warnings fixed #46
• An issue from the fuzzer dealing with equation type units #51

Added

• added a converter command line application that can convert units on the command line #35
• Added a file operation that can load user defined units from a file #36
• Added is_valid methods for all measurement types #36
• Added addUserDefinedInputUnit to add user defined units only on the input #36
• The webserver gained a to_string option to use the internal to_string operations to simplify the measurement and units #37
• The webserver and the converter gained an ability to handle * and <base> as the input unit to convert the measurement to base units. #37
• Added to_string operation for uncertain_measurements #38
• Added isnormal operation for measurement types #39
• Added UNITS_CLANG_TIDY option to run tests with Clang tidy #41
• Added fuzz_measurement fuzzer to test measurement_from_string #42
• Added cpplint test to azure #43
• Added a number of additional units from UDunits #44
• Added several CI checks including CPPlint, and a series of CMAKE checks, and formatting checks #48,#49
• Added a series of additional tests for coverage #50

Continued work on cleaning up the library and starting to add main documentation, as well as adding more units and cleaning up string conversions and some additional tests. Additional fuzzing fixes and add a webserver for exploring conversions.

Changed

• Change the unit_data operators from '+', '-' to '*' and '/' so they actually match the operation they are performing #12
• Pow on measurements is a free function instead of operator #12
• Update cmake policy configuration so it works with newer CMakes #31

Fixed

• Several issues that came from the fuzzer tests (#14, #18, #16, #19, #24, #28, #30)
• Fixed the UNITS_HEADER_ONLY target so it actually works and add some tests for it #23
• Update the cmake code so it correctly deals with and uses the CMAKE_CXX_STANDARD option #22
• Some strict aliasing warnings on gcc 6 #26

Added

• Added pow and root functions to measurements #7
• Add sqrt function which is a wrapper function around the root function for measurements and units #8
• Added uncertain measurment class for dealing with uncertainties #9, later modified in the primary method of uncertainty propagation#32
• Added a webserver for doing conversions through an html based interface #11
• Added a docker file for doing fuzzing #16
• Added initial set of documention on readthedocs.io #25,#27

Removed

• member methods of pow and root for measurements

Baseline release before some additional components get added