[Rd] more problems when using "installWithVers"

[Prof Brian Ripley]

On Tue, 4 Oct 2005, Deepayan Sarkar wrote:

> On Tuesday 04 October 2005 07:48, Prof Brian Ripley wrote:
>> Tony,
>>
>> The problem is specific to your example:
>>> install.packages("ash")
>>
>> ...
>>
>>> install.packages("ash", installWithVers=TRUE)
>>> library("ash", version="1.0-9")
>>> search()
>>
>>   [1] ".GlobalEnv"        "package:ash_1.0-9" "package:methods"
>>   [4] "package:stats"     "package:graphics"  "package:grDevices"
>>   [7] "package:utils"     "package:datasets"  "Autoloads"
>> [10] "package:base"
>>
>> and lies in lattice's .onLoad (as the message says, and hence the Cc: to
>> Deepayan).  That has
>>
>> .onLoad <- function(lib, pkg)
>> {
>>      library.dynam(pkg, pkg, lib )
>>      lattice.options(.defaultLatticeOptions())
>>      lattice.setStatus(.defaultLatticeStatus())
>> }
>>
>> which is incorrect: it should be library.dynam("lattice", pkg, lib)
>> OTOH, 'ash' has it correctly and so works.
>
> I'll upload a fixed version. There was a related discussion a while back about
> having a PACKAGE= argument in .Call()'s, but I don't remember the details.
> What's the current recommendation?

I would recommend that you do.  I think in your case you could also use 
useDynlib in the NAMESPACE file (since it is called before .onLoad), and 
only then is it reasonably safe to omit PACKAGE=.  The issue is about if 
you can load two versions simultaneously, and I don't see that as very 
safe with lattice (since it is tied to a grid version).

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595