[R-pkg-devel] debian-clang error with class

[Fernando Roa]

Take a look at
https://stat.ethz.ch/pipermail/r-package-devel/2019q4/004675.html

Best,

On Wed, Jan 8, 2020 at 3:09 PM Christian Martin Hennig <
christian.hennig using unibo.it> wrote:

> Hi there,
>
> I have an ERROR in my fpc package on debian-clang that I currently
> cannot reproduce due to the fact that rhub is apparently affected by a
> bug on that platform that doesn't allow me to install Bioconductor
> packages. The description is "--- failure: the condition has length > 1
> ---". I have googled for this and have found that there are a lot of
> packages that have such an error on debian-clang only. In all cases this
> seems to be associated with lines that compare a class of some object
> with a string such as
>
> if (class(xxx)!="try-error"), if (class(m)=="matrix"), if (class(y) %in%
> c("onething,"anotherthing")) etc.
>
> (The fpc package apparently achieved this with a "try-error" check.)
>
> Now on my machine (which isn't debian-clang) the class of xxx is a
> harmless thing, either "try-error" or another simple string (in the
> example that caused the error it is actually "matrix"), so it's not
> clear to me what's wrong with this on debian-clang. However, looking at
> all these other cases in other packages, I'm pretty sure that the error
> comes from comparing class(something) with a string.
>
> Can anybody shed some light on this? What goes wrong with
> class(xxx)!="try-error" and the like on debian-clang that works elsewhere?
>
> Thanks,
>
> Christian
>
>
> --
> Christian Hennig
> Dipartimento di Scienze Statistiche "Paolo Fortunati",
> Universita di Bologna, phone +39 05120 98163
> christian.hennig using unibo.it
>
> ______________________________________________
> R-package-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>


-- 
Fernando Roa

	[[alternative HTML version deleted]]