[R-sig-ME] bug/problem with update for merMod

[Ben Bolker]

Alexandra Kuznetsova <alku at ...> writes:

> 
> Dear all,
> 
> I have an issue with update.merMod function. I would like
> to call it within a function and to send some extra
> parameters to it.
> The code below fails with an error:
> 
updateModel <- function(model, reml, l)
{
 
   nfit <- update(object=model, formula.=as.formula(paste(".~."))
                  , REML=reml ,contrasts=l, evaluate=FALSE)
   env <- environment(formula(model))
   nfit <- eval(nfit, envir = env)
   #nfit <- eval.parent(nfit)
   #nfit <- eval(nfit, parent.frame())
 }
 
> m <- lmer(Reaction ~ Days +(1|Subject), data=sleepstudy)
> updateModel(m, FALSE, "contr.SAS")
> 
> Error in model.matrix.default(fixedform, fr, contrasts) : object 'l' not found
> 
> eval.parent(nfit) fails as well.
> 
> If I use evaluate=TRUE, then there is no problem, 
> but I need to use evaluate=FALSE
> 
> > sessionInfo()
> R version 3.0.2 (2013-09-25)
> Platform: x86_64-w64-mingw32/x64 (64-bit)
> 
[snip]

> other attached packages:
> [1] lme4_1.1-2   Matrix_1.1-0
> 
 
[snip]

> Is there some way to overcome this problem/bug?
> 
> Thank you in advance for any suggestion!
> 
> With best regards,
> Alexandra
> 

  It's not at all surprising that update fails when you do fancy
things with environments: it's a big headache trying to keep everything in
sync.  If you get a chance could you post this at 
https://github.com/lme4/lme4/issues ?

  Ben Bolker