R help - Dec 2002 - Loading libraries: Nas introduced

Hi all,

I am trying to package a library in R 1.6.1 (Windoze XP).

I have read the document "Writing R extensions" and think I
have done things correctly (though apparently not).  I have
searched the mail archives for help to no avail.

When I try to attach the library using, eg

	> library( libname, lib.loc=path.to.library)

I get this message:

	Warning message: 
	NAs introduced by coercion 

and the library does not work.

Points to note:

* I have no FORTRAN, only R code (getting FORTRAN working is
  the next project...!  Stay tuned!)

* If I simply source the R code, the functions all work with no
  problems.  So at present I am just sourcing the R code that is
  living in the R directory of my library and using it with no 
  problems.  This seems to suggest that the R code is not at fault
  since I am using the same R code that the library would use.

I can't figure out where or why NAs need to be introduced
(by coercion or otherwise!).  What is the reason, where is
it happening and why, and what is the solution?

Any help much appreciated.

Thanks as always.

P.

Dr Peter Dunn          (USQ CRICOS No. 00244B)
  Web:    http://www.sci.usq.edu.au/staff/dunn
  Email:  dunn @ usq.edu.au
<Insert favourite worthless disclaimer here>

"Peter Dunn" <dunn at usq.edu.au> writes:
> Hi all,
> 
> I am trying to package a library in R 1.6.1 (Windoze XP).
> 
> I have read the document "Writing R extensions" and think I
> have done things correctly (though apparently not).  I have
> searched the mail archives for help to no avail.
> 
> When I try to attach the library using, eg
> 
> 	> library( libname, lib.loc=path.to.library)
> 
> I get this message:
> 
> 	Warning message: 
> 	NAs introduced by coercion 
> 
> and the library does not work.
> 
> Points to note:
> 
> * I have no FORTRAN, only R code (getting FORTRAN working is
>   the next project...!  Stay tuned!)
> 
> * If I simply source the R code, the functions all work with no
>   problems.  So at present I am just sourcing the R code that is
>   living in the R directory of my library and using it with no 
>   problems.  This seems to suggest that the R code is not at fault
>   since I am using the same R code that the library would use.
> 
> I can't figure out where or why NAs need to be introduced
> (by coercion or otherwise!).  What is the reason, where is
> it happening and why, and what is the solution?
Hmm... that Warning message could come from lots of things --
as.numeric(c(1,2,"a")) is prototypical. 

First, did you install the package properly? Rcmd INSTALL, etc.

I see three places where things could go wrong:

1.  in the library() function itself
2.  while sourcing /path/to/yourlibrary/yourpackage/R/yourpackage
3.  during .First.lib processing

if /path/to/yourlibrary/yourpackage/R/yourpackage has nothing but
function definitions, something is structurally wrong. I think I'd try
sticking in a few test printouts between the functions in that file
and try to home in on the trouble spot that way.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907