understanding false-negative outcomes

bbolker commented 3 years ago

I'm currently doing a revdep_check() run of the lme4 package with an up-to-date version of revdepcheck (1.0.0.9001), but for reasons I can't fathom I'm missing some package errors. Any help would be greatly appreciated!

When I run revdep_summary("lme4") it looks like lmerTest is OK:

✔ lmerTest 3.1-2                         ── E: 0     | W: 0     | N: 0

yet: tail lme4/revdep/checks/lmerTest/old/lmerTest.Rcheck/00check.log gives

  Running ‘test_ranova_step.R’
  Running ‘test_summary.R’
  Running ‘test_zerovar.R’
  Running ‘zlmerTest_zeroDenom.R’
* DONE
Status: OK

while

tail lme4/revdep/checks/lmerTest/new/lmerTest.Rcheck/00check.log gives

 fm1    4 1774.5 1787.3 -883.26   1766.5                    
  fm0    4 1794.5 1807.2 -893.23   1786.5     0  0           
  > stopifnot(
  +   an2[2L, "Pr(>Chisq)"] == 1
  + )
  Error: an2[2L, "Pr(>Chisq)"] == 1 is not TRUE
  Execution halted
* DONE
Status: 1 ERROR

In case they're useful here are the full log files and the script I'm running (with output so far):

lmerTest_OLD_00check.log lmerTest_NEW_00check.log revdepcheck.R.txt revdepcheck.Rout.txt

bbolker commented 3 years ago

bump? any thoughts about this?

bbolker commented 3 years ago

any thoughts? (sorry to nag; I will shut up and go away if requested to do so). (The behaviour reported above is accurate with respect to the current CRAN version of lmerTest but has been fixed on GitHub.)

r-lib / revdepcheck

understanding false-negative outcomes #280