This directory contains a copy of the Intel Decimal Floating-Point Math Library
(inteldfp) version 2.0 Update 1.

The upstream library can be obtained from
https://software.intel.com/en-us/articles/intel-decimal-floating-point-math-library

The following changes have been made from the upstream:

- The "EXAMPLES" and "TESTS" directories have been removed, mostly because they
  occupied about 20MB in size combined.

- Moved directory "LIBRARY/src" to "LIBRARY/src/inteldfp" to allow packages
  depending on inteldfp to use relative include paths (e.g., "#include
  <inteldfp/bid_functions.h>" instead of "#include <bid_functions.h>") be built
  in place.

- The line breaks have been converted to use Unix Style from DOS style.

- The mode of the source files have been changed to be no longer executable.

- Created waf 'wscript' files to build using waf, because inteldfp has a
  makefile which is difficult to port to multiple platforms.

Note that the set of C source files linked into the inteldfp library (defined
in the wscript is the same as those compiled by the Makefile, which originally
came with inteldfp, on Linux.  The Makefile conditionally compiles other source
files (mostly providing certain transcendental functions) depending on the
platform, but they are not used by BDE's bdldfp package, so we decided to not
spend additional effort to include them in the wscript.  If the functions
provided by the unincluded source files are desired, the wscript must be
updated accordingly.
