It passed the following:
(1) WIn-builder
(2) Travis-ci
(3) Mac
(4) Window7
(5) Linux where R was build with long-double disabled
The A-B comparison on "(5) Linux - long-double disabled" shows that the package will fail "R CMD check --as-cran" if there is no such special R statement to change accuracy based on the long.double flag.
Here is the difference between A and B
19d18
< accuracy <- ifelse(capabilities("long.double"), accuracy, 0.05)
Here is the difference between R CMD check --as-cran on A and B
1c1
< * using log directory ‘/home/hubertj/FinalTest/A/CBPS.Rcheck’
using log directory ‘/home/hubertj/FinalTest/B/CBPS.Rcheck’
54c54
< * checking tests ... OK
checking tests ... ERROR
55a56,72
Running the tests in ‘tests/testthat.R’ failed.
Last 13 lines of output:
1/1 mismatches
[1] 0.00852 - 0.0121 == -0.00356
── 2. Failure: tests CBMS on the Lalonde data (@test-all.R#37) ────────────────
x["re74", "re75"] not equal to -0.03111142.
1/1 mismatches
[1] 0.0058 - -0.0311 == 0.0369
It passed the following: (1) WIn-builder (2) Travis-ci (3) Mac (4) Window7 (5) Linux where R was build with long-double disabled
The A-B comparison on "(5) Linux - long-double disabled" shows that the package will fail "R CMD check --as-cran" if there is no such special R statement to change accuracy based on the long.double flag.
Here is the difference between A and B
19d18 < accuracy <- ifelse(capabilities("long.double"), accuracy, 0.05) Here is the difference between R CMD check --as-cran on A and B
1c1 < * using log directory ‘/home/hubertj/FinalTest/A/CBPS.Rcheck’