R CMD check

[pingou]

Hi all,

Making some packages, and having "some" problems with the dependencies,
I was wondering if it possible to increment a sort of R CMD check2 that
do all the steps of the R CMD check except the "* checking package
dependencies".

That would help to process all the check command and avoid the circular
references problem, which is quite common.
This command would obviously be used only if the normal command can not
be used.

What do you think of ?
Is it possible ?

Thanks in advance
~pingou

	

	
		
___________________________________________________________________________ 
Yahoo! Mail réinvente le mail ! Découvrez le nouveau Yahoo! Mail et son interface
révolutionnaire.
http://fr.mail.yahoo.com

[Patrice Dumas]

On Thu, Aug 02, 2007 at 11:28:01PM +0200, pingou wrote:
>  Hi all,
>  
>  What do you think of ? 
I agree that it would be very usefull. But I don't have any idea about 
the feasability.

--
Pat

[Tom Callaway]

On Thu, 2007-08-02 at 23:33 +0200, Patrice Dumas wrote:
>  On Thu, Aug 02, 2007 at 11:28:01PM +0200, pingou wrote:
>  > Hi all,
>  > 
>  > What do you think of ?
>  
>  I agree that it would be very usefull. But I don't have any idea about 
>  the feasability. 
I'm buried in licensing at the moment, but if someone wanted to
implement this, I'd be willing to look at a patch and throw it at
upstream to get their thoughts.

~spot

[pingou]

Tom "spot" Callaway wrote:
>  On Thu, 2007-08-02 at 23:33 +0200, Patrice Dumas wrote:
> > On Thu, Aug 02, 2007 at 11:28:01PM +0200, pingou wrote:
> >> Hi all,
> >>
> >> What do you think of ?
> > I agree that it would be very usefull. But I don't have any idea about 
> > the feasability.
>  
>  I'm buried in licensing at the moment, but if someone wanted to
>  implement this, I'd be willing to look at a patch and throw it at
>  upstream to get their thoughts.
>   
Looking at the code I have found the paragraph following under in 
/usr/lib/R/bin/check

I believe there are tow ways to do what we would like,

*either make a copy of the files in a check2 file and patch it to avoid 
the check on suggested libraries ( I think it is a bit "dirty" to do it 
in such a way)
*or we can also either patch or sed the DESCRIPTION files in the spec to 
  comment the lines "Suggests"

I do not know what you think about these propositions and I do not know 
if there are better ones but I believe these are not the optimal solution.
My problems is that my knowledge are too low to be able to propose 
directly the correct patch/solution.

I hope this help, if one of you  would like to do it.

Best regards,
~pingou



$log->checking("package dependencies");
             ## Everything listed in Depends or Suggests or Imports
             ## should be available for successfully running R CMD check.
             ## \VignetteDepends{} entries not "required" by the package 
code
             ## must be in Suggests.  Note also that some of us think that a
             ## package vignette must require its own package, which OTOH is
             ## not required in the package DESCRIPTION file.
             ## Namespace imports must really be in Depends.
             my $Rcmd = "options(warn=1,warnEscapes=FALSE); 
tools:::.check_package_depends(\"${pkgdir}\")\n";
             my @out = R_runR($Rcmd, "${R_opts} --quiet",
                              "R_DEFAULT_PACKAGES=NULL");
             @out = grep(!/^\>/, @out);
             if(scalar(@out) > 0) {
                 $log->error();
                 $log->print(join("\n", @out) . "\n");
                 $log->print(wrap("", "", @msg_DESCRIPTION));
                 exit(1);
             } else {
                 $log->result("OK");
             }
         }


	

	
		
___________________________________________________________________________ 
Yahoo! Mail réinvente le mail ! Découvrez le nouveau Yahoo! Mail et son interface
révolutionnaire.
http://fr.mail.yahoo.com