Skip to content

Namespace File

9 messages · Simon Urbanek, Fabio Mathias Corrêa

#
Dear colleagues,


Problem solved with:


".First.lib" <- function(lib, pkg)
{
  library.dynam("Bayesthres", package = pkg, lib.loc = lib)
  return(invisible(0))
}





 
              F?bio Mathias Corr?a
         Departamento de Estat?stica
   Universidade Estadual de Santa Cruz
#
On Dec 15, 2010, at 6:57 AM, Fabio Mathias Corr?a wrote:

            
Well, that's shooting at a clay pigeon with a missile ;). You could have done the same simply using
useDynLib(Bayesthres)
in the NAMESPACE.

Your issue is very likely just a mixup of cases as there is no "vuA" symbol in Fortran as all symbols are lower-case (at the object file level).

Cheers,
Simon
#
Dear Simon

By using useDynLib (Bayesthres) in NAMESPACE file. Appeared another error:

Error in dyn.load("Bayesthres.so") : 
  unable to load shared object '/home/fmc/Bayesthres/Bayesthres.so':
  /home/fmc/Bayesthres/Bayesthres.so: cannot open shared object file: No such 
file or directory
ERROR: lazy loading failed for package ?Bayesthres?
* removing ?/home/fmc/Bayesthres.Rcheck/Bayesthres?

 
Now he is not recognizing the file Bayesthres.so

Grateful for the help.



              F?bio Mathias Corr?a
         Departamento de Estat?stica
   Universidade Estadual de Santa Cruz




----- Mensagem original ----
De: Simon Urbanek <simon.urbanek at r-project.org>
Para: Fabio Mathias Corr?a <fabio.ufla at yahoo.com.br>
Cc: r-devel at r-project.org
Enviadas: Quarta-feira, 15 de Dezembro de 2010 13:41:34
Assunto: Re: [Rd] Namespace File
On Dec 15, 2010, at 6:57 AM, Fabio Mathias Corr?a wrote:

            
Well, that's shooting at a clay pigeon with a missile ;). You could have done 
the same simply using
useDynLib(Bayesthres)
in the NAMESPACE.

Your issue is very likely just a mixup of cases as there is no "vuA" symbol in 
Fortran as all symbols are lower-case (at the object file level).

Cheers,
Simon
#
On Dec 15, 2010, at 1:31 PM, Fabio Mathias Corr?a wrote:

            
... but that's an entirely wrong place - I suspect you have some real issues in the way your package is constructed. Are you sure you have the correct layout and are using the *packaged* sources and not your working tree? (You didn't tell us anything about the package so there are many things that may be wrong - also you omitted the most basic details such as R version and platform...)

Cheers,
Simon
#
Dear,

I am using Ubuntu linux and R2.12.0
The trial version is in my home.

\home\Bayesthres~$

To perform the R CMD check

\home~$ R CMD check Bayesthres

Within the directory Bayesthres, have:

Directories: inst, man, R and src

Files: DESCRIPTION AND NAMESPACE

Best wishes

F?bio




... but that's an entirely wrong place - I suspect you have some real issues in 
the way your package is constructed. Are you sure you have the correct layout 
and are using the *packaged* sources and not your working tree? (You didn't tell 
us anything about the package so there are many things that may be wrong - also 
you omitted the most basic details such as R version and platform...)

Cheers,
Simon


 
              F?bio Mathias Corr?a
         Departamento de Estat?stica
   Universidade Estadual de Santa Cruz
#
On Dec 15, 2010, at 1:56 PM, Fabio Mathias Corr?a wrote:

            
That is in general not a good idea, especially if you ever run things inside by hand for testing (because that will pollute the build). Make sure you clean your tree and use R CMD build followed by R CMD check on the resulting tar ball. Does the issue persist?
What is in inst? Make sure you have no binaries there.

Cheers,
Simon
#
The inst has CITATION only.

Best wishes



              F?bio Mathias Corr?a
         Departamento de Estat?stica
   Universidade Estadual de Santa Cruz





What is in inst? Make sure you have no binaries there.

Cheers,
Simon
#
See the complete log

fmc at raquel-laptop ~ $ R CMD check Bayesthres
* using log directory ?/home/fmc/Bayesthres.Rcheck?
* using R version 2.12.0 (2010-10-15)
* using platform: i486-pc-linux-gnu (32-bit)
* using session charset: UTF-8
* checking for file ?Bayesthres/DESCRIPTION? ... OK
* this is package ?Bayesthres? version ?0.1-0?
* checking package dependencies ... OK
* checking if this is a source package ... WARNING
Subdirectory ?Bayesthres/src? contains object files.
* checking for executable files ... OK
* checking whether package ?Bayesthres? can be installed ... ERROR
Installation failed.
See ?/home/fmc/Bayesthres.Rcheck/00install.out? for details.


File 00install.out

Error in dyn.load("Bayesthres.so") : 
  unable to load shared object '/home/fmc/Bayesthres/Bayesthres.so':
  /home/fmc/Bayesthres/Bayesthres.so: cannot open shared object file: No such 
file or directory
ERROR: lazy loading failed for package ?Bayesthres?
* removing ?/home/fmc/Bayesthres.Rcheck/Bayesthres?

Best wishes



 
              F?bio Mathias Corr?a
         Departamento de Estat?stica
   Universidade Estadual de Santa Cruz




----- Mensagem original ----
De: Simon Urbanek <simon.urbanek at r-project.org>
Para: Fabio Mathias Corr?a <fabio.ufla at yahoo.com.br>
Cc: R-devel <r-devel at r-project.org>
Enviadas: Quarta-feira, 15 de Dezembro de 2010 17:10:21
Assunto: Re: [Rd] Namespace File
On Dec 15, 2010, at 1:56 PM, Fabio Mathias Corr?a wrote:

            
That is in general not a good idea, especially if you ever run things inside by 
hand for testing (because that will pollute the build). Make sure you clean your 
tree and use R CMD build followed by R CMD check on the resulting tar ball. Does 
the issue persist?
What is in inst? Make sure you have no binaries there.

Cheers,
Simon
#
Fabio,

you must have
dyn.load("Bayesthres.so")
somewhere in your R code according to your output - that is wrong and have to remove it as that will happen automatically via NAMESPACE.
Also, please, do read what I wrote in the last e-mail - you're still running it on your working copy.

Cheers,
Simon

BTW: in order to avoid us having to guess what you did wrong it would be much easier if you simply put the package somewhere so we can check it - that would save us and you quite a bit of time.
On Dec 15, 2010, at 3:34 PM, Fabio Mathias Corr?a wrote: