GitLab

Removed cpack code entirely for non-Windows/OSX.

Fixed install directories on everything else.

The source package isn't really a proper source package if it relies on
external data; this commit imports the upstream data being used, and
includes a script to update it as needed.

Switch to GNUInstallDirs for proper install locations.

MathJax CDN wasn't being properly used: when Doxyfile has
MATHJAX_RELPATH = (blank) it assumes MathJax is in cwd: the line
shouldn't be there at all when the CDN is desired.

Copied from eris.

Also removed a set of duplicate (and apparently unused) so version
variables.

I believe that the upstream citation *requirement* is legally void, as
a program cannot claim copyright over its output (in most cases).
Since the copyright itself only applies to the use (and copying) of the
program itself, any restriction on what you do *after* using the
program is impossible to enforce and, thus, void.

Ethically, however, the citation is required--and thus I interpret
upstream's intentions as a citation request, rather than the
(unenforceable) citation requirement, which *is* compatible with the
GPL.

They are symlinked into the build directory, which really isn't right.

This is supposed to be slightly more accurate, and is (ignoring the
work Eigen is doing) simpler code.

Also added BOOST_FORCE_LOCAL option (just like EIGEN_FORCE_LOCAL) to
force use of embedded boost copy.

These deprecation warnings were supposedly fixed in 3.2.5, but
apparently not completely:

http://eigen.tuxfamily.org/bz/show_bug.cgi?id=872

The devel branch commit is simple: ifdef out the use under C++11 or
later:

https://bitbucket.org/eigen/eigen/commits/ff283951217/

Also print out the version of Eigen found and used.

I want to change one of the decompositions, which needed more of
Eigen--having the whole Eigen code is considerably more convenient, and
not all that much bigger than what was there already.

- Added --linear flag to fdpval and fdcrit that uses linear B
  interpolation between the two nearest dataset B values instead of the
  5-9 nearest points for a quadratic interpolation.
- Added version and licence info to fdpval and fdcrit output
- Added version variables to library (in fracdist/version.hpp, .cpp)
- Removed md5sum check from mn-files.zip download; it broke every time
  the data set download changed (from the recent fracdist.f fixes),
  making the package unbuildable.
- Renamed parse-vals.hpp to cli-common.hpp, and moved some more
  fdpval.cpp/fdcrit.cpp common code into it.

Usage is simple: ./compare.pl usually works.

Relying on CMAKE_HOST_SYSTEM_PROCESSOR isn't entirely reliable: it
gives x86_64 in an i386 chroot, for instance.  It now also queries
perl's longsize and recognizes arch=i386 if longsize=4 but
CMAKE_HOST_SYSTEM_PROCESSOR reports x86_64.

Otherwise doxygen becomes required despite the code to avoid that in
CMakeLists.txt.

The previous version was using a generic LU inverse, which ran into
numerical stability problems for large q values.  For example,

q=7, c=0, b=(1.40, 1.41, ..., 1.50) looked like:

0.1767019
0.1741197
0.1640931
0.1330334
0.2003187
0.1792207
0.1792029
0.175216
0.1836156
0.2131243
0.2235138

because the source data files have only slight variation in that range.
Using a cholesky decomposition inverse fixes this, and gives numbers
agreeing exactly with JGM's fortran version (which also uses Cholesky
decomposition):

0.1744341
0.1780252
0.181262
0.1841298
0.1866257
0.1887341
0.1904548
0.1917758
0.1926928
0.1932153
0.1933418

which agrees precisely with JGM's (trimmed) fracdist output of:

 P value = 0.174
 P value = 0.178
 P value = 0.181
 P value = 0.184
 P value = 0.187
 P value = 0.189
 P value = 0.190
 P value = 0.192
 P value = 0.193
 P value = 0.193
 P value = 0.193

It isn't doing anything useful anymore, since the libgsl dep is gone.

The 1.0.0 release binary packages were installing headers in the wrong
place (include/common.hpp, etc. instead of include/fracdist/common.hpp),
and were missing the fracdist/data.hpp header.

Added component configurations for Windows installers so that Windows
users can elect to not install the headers and/or docs when using the
.exe installer.

This way the project doesn't actually contain it, which is better.

There's no reason for QCache to be in the header as its entirely
private to common.cpp.  Moved it (and unnamed it).

Eigen3 is getting something wrong with the matrix reuse (clang doesn't
seem to have a problem).  Forcing Eigen to build the result is
worthwhile anyway (lazy evaluation here is probably not helpful).