benchmarks of major functionality and fix LTO check/testthat errors

[slwu89]

hey @giovannic!

I added benchmarking scripts to tests/performance using the bench package. I don't know what's the best way to set up these scripts so please let me know what you think after you have a chance to run them. Anyway, we probably won't be running benchmarks for every PR so it might not be a big deal.

I think I fixed the testthat error contributing to the LTO check problem on CRAN. Upgrading testthat to the latest version and setting up the C++ testing files again using testthat::use_catch() seems to have made changes suggested by OP in r-lib/testthat#1230. So maybe they fixed it and didn't remember to update people following the issue. After making that change I ran the r-hub builders, making sure --enable-lto flags were enabled and none of the build platforms had that specific error we were getting.

I also added #include <unordered_set> to Event.h as some of the r-hub platforms (Solaris .........) errored out without that header, and also deleted the ; from the ctor of TargetedEvent to get rid of that -Wpedantic compiler warning.

[codecov]

Codecov Report

Merging #151 (1ad60f3) into dev (70967d8) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##              dev     #151   +/-   ##
=======================================
  Coverage   94.68%   94.68%           
=======================================
  Files          26       26           
  Lines        1185     1185           
=======================================
  Hits         1122     1122           
  Misses         63       63           

Impacted Files	Coverage Δ
R/categorical_variable.R	100.00% <100.00%> (ø)
inst/include/Event.h	92.68% <100.00%> (ø)
src/categorical_variable.cpp	100.00% <100.00%> (ø)

---

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 b5667e0...1ad60f3. Read the comment docs.

[giovannic]

Looks great!

Some suggestions:

It would be good to add bench to DESCRIPTION > Suggests.

Do we really need tidyr? It takes a while for me to install

It would be tidier to have build_grid, simplify_bench, create_random... in a tests/performance/utils.R file.

Thank you so much for the testthat fix!

[slwu89]

Thanks @giovannic.

Yeah, let me see if I can work out a version of simplify_bench_output not relying on tidyr. I agree it would be much tidyr to put all that in a utils.R file as well.

[giovannic]

Great. Thanks!