[R-pkg-devel] help with CRAN failures (broom.mixed, rstan-related)

[Ben Bolker]

   broom.mixed is failing tests on Fedora.

With apologies for the mangled formatting, here is the test failure 
(full properly formatted details at 
https://www.r-project.org/nosvn/R.check/r-devel-linux-x86_64-fedora-gcc/broom.mixed-00check.html).

   ══ Failed tests 
════════════════════════════════════════════════════════════════   ── 
Error ('test-alltibbles.R:57:21'): (code run outside of `test_that()`) 
──────   Error in `.requirePackage(package)`: unable to find required 
package 'rstan'   Backtrace:        ▆     1. └─broom.mixed (local) 
testf("glance", z) at test-alltibbles.R:62:13     2. 
├─testthat::expect_is(get(fn_str)(obj), "tbl_df") at 
test-alltibbles.R:57:21     3. │ └─testthat::quasi_label(enquo(object), 
label, arg = "object")     4. │ └─rlang::eval_bare(expr, 
quo_get_env(quo))     5. ├─get(fn_str)(obj)     6. 
└─broom.mixed:::glance.brmsfit(obj)     7. 
└─broom.mixed:::glance_stan(...)     8. ├─stats::sigma(x)     9. 
└─broom.mixed:::sigma.brmsfit(x)    10. ├─"sigma" %in% 
brms::variables(object)    11. ├─brms::variables(object)    12. 
└─brms:::variables.brmsfit(object)    13. └─methods (local) 
dimnames(x$fit)    14. └─methods (local) `<fn>`(`<list>`, `<stndrdGn>`, 
`<env>`)    15. └─methods:::.findInheritedMethods(classes, fdef, mtable) 
    16. └─methods::getClass(cl1, where = where, .Force = TRUE)    17. 
└─methods::getClassDef(Class, where)    18. 
└─methods:::.requirePackage(package)      [ FAIL 1 | WARN 4 | SKIP 1 | 
PASS 237 ]   Error: Test failures   Execution halted


   This is in a test section where I loop over fitted examples from 
different modeling packages. *If* the relevant package is installed, I 
test the various methods defined in broom.mixed for extracting 
information from that kind of object.

   This appears to be failing when testing objects from 'brms', because 
'rstan' isn't available. However, I don't see how brms can be available 
when rstan isn't, since brms Imports: rstan?

   Weirdly, I can't get this to fail on my local system even if I remove 
rstan -- maybe rstan needs to be present for brms at install time but 
not at runtime ... ???

   Any hints/insights/guesses/suggestions for further diagnostics are 
welcome.

   cheers
     Ben




-- 
Dr. Benjamin Bolker
Professor, Mathematics & Statistics and Biology, McMaster University
Director, School of Computational Science and Engineering
* E-mail is sent at my convenience; I don't expect replies outside of 
working hours.