Have source release available for use with R package

[gothub]

NCEAS and rOpenSci are eager to integrate the C++ version of YW into an R package
that will wrap YW functionality to make it easily available to R users, and for use into
our own R packages. No repo has been setup for this yet.

One of the restrictions of distributing an R package (via CRAN) is that
no pre-compiled libraries can be included with the R package (no pre-compiled YW C++)
library. The build process for the R package in CRAN howwever, can perform all the linking
(sqlite, XSB).

So, it would be very helpful to have 'source' releases available that could be used by the R package.

[tmcphillips]

Great! I was hoping that a C++ implementation of YW could facilitate development of an R package distributed through CRAN. It's good to know about the restrictions on distribution. I, too, am distrustful of precompiled shared libraries (and dependence on OS-level package managers) where scientific computing (and reproducibility) is concerned. The bar is even higher for software aiming to enable transparency, reproducibility, and provenance management generally.

Fortunately, building the new YW sources and essential dependencies (SQLite and the ANTLR4 runtime) is very easy. I expect XSB to be relatively straightforward. Graphviz is more complicated due to extensive dependencies on 3rd-party libraries for font and graphics formats. It's not clear how many output graphic file formats YW will be able support right away without dependence on shared libraries (SVG and postscript easily because they are built into the Graphviz sources themselves, PNG and PDF not so easily). It'd be great to sort all this out together!