disable inclusion of modules code with RCPP_NO_MODULES

[kevinushey]

Closes #993.

[codecov-io]

Codecov Report

Merging #995 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #995   +/-   ##
=======================================
  Coverage   82.46%   82.46%           
=======================================
  Files          63       63           
  Lines        3166     3166           
=======================================
  Hits         2611     2611           
  Misses        555      555

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 92b6207...286a45f. Read the comment docs.

[eddelbuettel]

What did you timing tests reveal in terms of net change? 10%? 20%? More?

[eddelbuettel]

The PR looks really clean and straightforward but to be on the safe side I think I will turn on a full rev.dep check. (Not that we expect anything with an opt-in #define but doesn't hurt to run a full set and it has been a few weeks since I last did it...)

[kevinushey]

Using roxygen2 as an example:

# regular build ----
$ time R CMD INSTALL --preclean roxygen2 &> /dev/null

real    0m13.648s
user    0m12.796s
sys     0m0.773s

# RCPP_NO_MODULES ----
$ time R CMD INSTALL --preclean roxygen2 &> /dev/null

real    0m9.479s
user    0m8.800s
sys     0m0.594s

So, almost 50% faster, which isn't too shabby.

[eddelbuettel]

Ok, took a moment to go over rev.deps as some unrelated change (possibly in package sf) seems to have invalidated a bunch of the spatial packages and their tests failed -- nothing to do with us, mostly just needing a recompilation on that box it seems. Will merge in due course.