[R] R CMD check message that I can't figure out (.onLoad failed)

[Prof Brian Ripley]

On Wed, 6 Aug 2008, Uwe Ligges wrote:

>
>
> stephen sefick wrote:
>> I have been able to track and fix everything so far with R CMD check.
>> How do I make this reproducible and I will do it.
>> 
>> this is the message that I got:
>> 
>> Error in fun(...) : couldn't connect to display ":0.0"
>> Error : .onLoad failed in 'loadNamespace' for 'tcltk'
>> 
>> 
>
>
> The first line is the really relevant one. I guess you have no X available 
> (or not forwarded via ssh) or whatever, but the package starts some GUI 
> stuff.

(It looks like the package or a dependency depends on tcltk, so that does 
get loaded for the codetools checks and fails.)

If this is a vanilla Unix-alike this only happens if you have DISPLAY set 
and no X running on the setting (at least in tolerably recent R).

However, I've just seen a message from ssefick on R-sig-mac, so that he 
forgot to tell us the minimum information asked for in the posting guide 
is perhaps extremely relevant.  If this is a Mac, the exact OS matters as 
it is supposed to auto-start an X server.  Maybe unsetting DISPLAY will 
work, otherwise seek advice from an expert on the unstated OS.

-- 
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