[R] Latest revision of lme4 requires 'Matrix' >= 0.995.2 but that version is not available on CRAN

[Uwe Ligges]

Søren Højsgaard wrote:

> I see that Matrix, lme4 and SASmixed in zip format have been 'rolled back' to the previous versions - which I have now installed. Running 
>  
> library(SASmixed)
> data(Demand)
> example(Demand)
>  
> gives
> ....
>  
> Demand> fm1Demand <- lmer(log(d) ~ log(y) + log(rd) + log(rt) + 
>     log(rs) + (1 | State) + (1 | Year), Demand)
> Error: object "d" not found
> Error in log(d) : unable to find the argument 'x' in selecting a method for function 'log'
> 
> - so it is not a complete 'roll back' - unless I've misunderstood something? Help will be appreciated...

Yes, I rolled back one version. This is another error in the former 
version. But since Matrix 0.995-3 is on its way to CRAN, there are 
hopefully completely working versions available within 24 hours.

Uwe Ligges


> / Søren
> 
>  
> 
> [...]
> 
> 
> Yes, but implementing this would require an even higher computational
> effort to build and check Windows binary packages.
> 
> Since it looks like the Matrix problems are persisting, I will play back
> the "old" versions of both packages to CRAN shortly.
> 
> Uwe
> 
> 
> 
> 
>>Anyway, that sort of things happen... I've now downloaded the
>>previous version of lme4 as a tar.gz file - and that works. I wonder
>>if it would be an idea to have a function rollback('pkg-name') for
>>that sort of situations???
>>
>>Best regards
>>
>>Søren Højsgaard
>>
>>
>>
>>
>>
>>Moreover, now that I have updated to the latest release of lme4, I
>>wonder how to get get a version of lme4 that actually works with the
>>version of Matrix which is currently available.
>>
>>
>>
>>
>>
>>
> 
> 
>