Closed pwep closed 5 years ago
Merging #23 into development will not change coverage. The diff coverage is
100%
.
@@ Coverage Diff @@
## development #23 +/- ##
============================================
Coverage 64.74% 64.74%
============================================
Files 56 56
Lines 7457 7457
============================================
Hits 4828 4828
Misses 2629 2629
Impacted Files | Coverage Δ | |
---|---|---|
R/StSignificanceTesting.R | 72.93% <100%> (ø) |
:arrow_up: |
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 cd2e0b2...b9c8d2f. Read the comment docs.
Worked! Thanks Peter; I will check other files where this type of change might be appropriate; I must say the student's coding held up! It turned out to be a R bug - I would never have figured that.
Still having problems with test
Incorporated Peter proposed changes (as.vector()) to StdError and DF in StSignificanceTestingCrossedModalities.R; did not alter StSingleFixedFactor.R
Changed goodValues directory to goodValues361, meaning the values are for R version 3.6.1
Regenerated good values on laptop, which is running 3.6.0;
Moved goodvalues for SigTest and SigTestCad from values from Peter's FTP website to RJafroc testthat directory; after adding .rds extension to all of the files
── Test failures ───────────────────────────────────────────────── testthat ────
library("testthat") library("RJafroc")
test_check("RJafroc") ── 1. Failure: SignificanceTestingAllCombinations (@test-significance-testing CurrentValues[[listMem]] not equal to GoodValues[[listMem]]. Component "Source": Attributes: < Component "levels": 3 string mismatches > List member = 2, Dataset = dataset02, FOM = Wilcoxon, method = DBMH
── 2. Failure: SignificanceTestingAllCombinations (@test-significance-testing CurrentValues[[listMem]] not equal to GoodValues[[listMem]]. Component "Source": Attributes: < Component "levels": 3 string mismatches > List member = 2, Dataset = dataset05, FOM = HrAuc, method = DBMH
══ testthat results ════════════════════════════════════════════════════════ OK: 283 SKIPPED: 0 WARNINGS: 0 FAILED: 2
Need help Peter! These look identical to me, yet it fails R CMD check (but passes dev tools::test) and passes all Travis checks; strange
GoodValues <- readRDS(fn) CurrentValues <- StSignificanceTesting(dataset, FOM = FOM_arr[i],method = method_arr[j]) str(GoodValues[[2]]) 'data.frame': 8 obs. of 4 variables: $ Source: Factor w/ 8 levels "C","R","RC","T",..: 4 2 1 7 5 3 8 6 $ SS : num 0.547 1.749 44.846 0.251 11.283 ... $ DF : num 1 4 113 4 113 ... $ MS : num 0.5468 0.4373 0.3969 0.0628 0.0998 ... str(CurrentValues[[2]]) 'data.frame': 8 obs. of 4 variables: $ Source: Factor w/ 8 levels "C","R","RC","T",..: 4 2 1 7 5 3 8 6 $ SS : num 0.547 1.749 44.846 0.251 11.283 ... $ DF : num 1 4 113 4 113 ... $ MS : num 0.5468 0.4373 0.3969 0.0628 0.0998 ... attributes(GoodValues[[2]]) $names [1] "Source" "SS" "DF" "MS"
$class [1] "data.frame"
$row.names [1] 1 2 3 4 5 6 7 8
attributes(CurrentValues[[2]]) $names [1] "Source" "SS" "DF" "MS"
$class [1] "data.frame"
$row.names [1] 1 2 3 4 5 6 7 8
A bug labelled Type inconsistency when converting arrays to data frames was fixed in R v3.6.1
https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17570
The fix changes some simple array structure to a vector when included in a
data.frame
. Older versions <3.6.1 did not do this as expected. This appears to be the source of the errors in the test ofStSignificanceTesting.R
. Do have a look at the simple example code on the bug fix page for more detail.This pull request adds a
as.vector
call to the relevant parts of the output inStSignificanceTesting.R
- it should not have any impact for v3.6.1, but bring earlier versions into line. I have added it toStSignificanceTesting.R
so sensible output is generated, rather than try to catch the individual issues in the test itself.Travis should pass on all three versions (versions currently:
oldrel: 3.5.3
;release: 3.6.0
;devel:
is a recent pull from R dev tree which will become R 3.6.1). The goodValues (not changed in this pull request) have all been generated previously in R 3.6.1.