| David Ghandehari | 9e5b587 | 2016-07-28 09:50:04 -0700 | [diff] [blame] | 1 | Name: David M. Gay's floating point routines |
| 2 | URL: http://www.netlib.org/fp/ |
| 3 | License: MIT-like |
| 4 | |
| 5 | Original dtoa.c file can be found at <http://www.netlib.org/fp/dtoa.c>. |
| 6 | Original g_fmt.c file can be found at <http://www.netlib.org/fp/g_fmt.c>. |
| 7 | |
| 8 | List of changes made to original code: |
| 9 | - wrapped functions in dmg_fp namespace |
| 10 | - renamed .c files to .cc |
| 11 | - added dmg_fp.h header |
| 12 | - added #define IEEE_8087 to dtoa.cc |
| 13 | - added #define NO_HEX_FP to dtoa.cc |
| 14 | - made some minor changes to allow clean compilation under g++ -Wall, see |
| 15 | gcc_warnings.patch. |
| 16 | - made some minor changes to build on 64-bit, see gcc_64_bit.patch. |
| 17 | - made minor changes for -Wextra for Mac build, see mac_wextra.patch |
| Andrew Top | 0d1858f | 2019-05-15 22:01:47 -0700 | [diff] [blame] | 18 | - fixed warnings under msvc, see msvc_warnings.patch |
| 19 | - fixed parsing of long exponents, see exp_length.patch and crbug.com/542881 |
| 20 | - made hexdig array const |
| 21 | - removed deprecated `register` keyword |
| 22 | - #undef Long so that it won't change Long in other files in jumbo builds |
