[Rd] Possible bug in 'new()' for Reference Classes

[John Chambers]

Right, thanks for the catch.  Actually, field names "s", "se", "sel" 
would also produce the bug.  Partial matching of argument names bites 
again.  This should be fixed in r-devel and 2.14 patched, as of SVN rev. 
57842.

Do try to follow the API in the documentation and use generator objects 
for reference classes.  It's simpler than using S4 new() and makes it 
clear that the example is of a reference class.

John

On 12/7/11 7:36 AM, Janko Thyson wrote:
> Dear list,
>
> I think I stumbled across a little bug with respect to the standard
> initialization routine for Reference Classes.
>
> It seems that a field 'self' is treated as if it's name would be
> '.self' (which we know is reserved for the self reference of the
> instantiated object itself) and thus an error is thrown.
> If the field value is assigned in an explicit call after the
> instantiation via 'new()', everything works just fine:
>
> setRefClass("ClassInfo",
>         fields=list(
>             self="character", super="character", sub="character"
>         )
>     )
>     new("ClassInfo", self="B", super="A", sub="C")    # Error
>
>     x <- new("ClassInfo", super="A", sub="C")
>     x
>     x$self <- "B" # Works
>     x
>
> Best regards,
> Janko
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>