Skip to content

Correct use of Depends, Imports and ::

10 messages · Simon Urbanek, Brian Ripley, Yanming Di +2 more

#
Dear R developers,

Taking advantage of the yesterday discussion about the use of 
Depends/Import/Suggests/Enhances, I would like to add a related question.

Let's assume, in the DESCRIPTION file of my package foo0, I have:

Depends: foo1
Imports: foo2

while in the NAMESPACE file of my package I have

importFrom("foo2", f2)


and within my package I use the following two external functions:

f1, from package foo1
f2, from package foo2


If I want to use the two previous functions within a function of my 
package foo0, is it correct to call them as follow ( assuming 'x' is the 
only argument of 'f1' and 'f2'):

f1(x)
foo2::f2(x)

OR should I use

foo1::f1(x)
f2(x)

OR

f1(x)
f2(x)

OR

something else (the correct way of doing it in R) ?



Finally, can I use the following call in the body of a function of my 
package foo0:

foo3::f3(x)

without declaring the package 'foo3' in the Imports section of my 
DESCRIPTION file ?


Thanks in advance for your help,


Mauricio Zambrano-Bigiarini
#
On Nov 7, 2012, at 4:04 AM, Mauricio Zambrano-Bigiarini wrote:

            
Yes, this one. You have imported f2 so the whole point of that is that it will be inserted as "f2" into your imports environment, so you want to use it that way. When you use foo2:: you are skipping over imports and addressing the namespace of foo2 directly which sort of defeats the purpose (and is slower).

As for f1, you have attached the package so it is available. I would not use foo1:: for the same reason as above, but there is a risk that it may get clobbered since the call will go through the search path so you may not get f1 from foo1 necessarily - this may or may not be a good thing depending on your intentions. You could import it via import(foo1) to make sure you don't need to go through the search path - that's what I would do if I don't intend overrides. (Whether you want to use Depends or Imports in that case is related to what you intend for the user to see as I was saying in my last comment - if I expect users to use foo1 as well, I would use Depends, otherwise Imports is good enough.)
No, it has to be at least in Enhances or Suggests (see R-exts 1.1.1).

Cheers,
Simon
#
On Nov 7, 2012, at 9:25 AM, Simon Urbanek wrote:

            
This should have been "Yes, but it has ..."
#
On 11/07/12, Simon Urbanek <simon.urbanek at r-project.org> wrote:

            
Thanks Simon for making the point related to f2 clearer for me.
So, if I'm only using a few functions of foo1, I shouldn't declare them in Depends, but only in Imports, which also will allow me to be sure that the functions come from foo1 and not from another package that have functions with the same, right ?
Thanks again,

Mauricio

=====================================
Linux user #454569 -- Ubuntu user #17469
=====================================
"If you torture any data set long enough, 
it will confess anything!" (Murray Lark)
#
On 07/11/2012 17:52, mauricio zambrano wrote:
Or declare in Depends but also import in the NAMESPACE.
Yes, but it would be better to use importsFrom() with just the functions 
you need (the list of exported functions from foo1 could change over time).

I've just been through this advising an author of a package who needs to 
use intersect() from rgeos, not intersect() from base.
#
Hi all,

In a package I maintain, I use the "dqrls" from the base package to fit 
regression models. However, I was informed that directly calling functions 
from the base package is no longer allowed:

"CRAN packages should use only the public API. Hence they should not use
entry points not declared as API in installed headers nor .Internal()
nor .Call() etc calls to base packages. Such usages can cause packages
to break at any time, even in patched versions of R."

Prof. Ripley suggested that one can replace such calls with lm.wfit, which 
will work.

However, in my package, the call will need to be implemented hundreds of 
thousands times. Using lm.wfit is much slower. I am wondering whether there 
is alternative ways that are more efficient. I can program in C if needed.

Many thanks.

Best regards,
Yanming
#
On 07/11/2012 20:26, Yanming Di wrote:
And dqrls is no longer available via .Fortran in R-devel.
If the licence of your package allows it, you can copy the Fortran code 
into your package and write your own interface via .Fortran, .C or .Call.

The .Fortran interface is itself quite costly, especially in R < 2.15.1: 
in some cases lm.wfit is actually faster as it uses .Call.

  
    
#
On 11/7/12 12:00 PM, Prof Brian Ripley wrote:
This is especially a good idea if there is any chance that anyone else 
will ever want to import your package.  If a package is in Depends but 
the functions you use are not explicitly imported in the NAMESPACE (with 
imports or importsFrom), it can cause problems as described in this thread:

https://stat.ethz.ch/pipermail/r-devel/2011-September/062043.html
Stephanie Gogarten
Research Scientist, Biostatistics
University of Washington
#
2012/11/7 Prof Brian Ripley <ripley at stats.ox.ac.uk>:
That is what I wanted to say (although I was not explicit):
to declare the dependency of package foo1 in the Imports section of
the DESCRIPTION file, and then import the specific functions in the
NAMESPACE file by using importFrom("foo1", f11,...,f1n)

Thank you very much for this useful advise.


Mauricio Zambrano-Bigiarini
--
=====================================
Water Resources Unit
Institute for Environment and Sustainability
Joint Research Centre, European Commission
webinfo    : http://floods.jrc.ec.europa.eu/
=====================================
DISCLAIMER:\ "The views expressed are purely those of th...{{dropped:15}}
#
2012/11/7 Stephanie M. Gogarten <sdmorris at u.washington.edu>:
So, no matter if the package providing the functions I'm using within
my package (e.g., foo1) are declared in the Depends or Imports section
of the DESCRIPTION file, it is always a "good practice" to specify all
the imported functions with import(foo1) or importFrom("foo1",
f11,...,f1n) in the NAMESPACE file of my package, in order to avoid
problems as the one described in the previous link.

Thanks Prof. Ripley and Stephanie for the advice,

All the best,

Mauricio
--
=====================================
Water Resources Unit
Institute for Environment and Sustainability
Joint Research Centre, European Commission
webinfo    : http://floods.jrc.ec.europa.eu/
=====================================
DISCLAIMER:\ "The views expressed are purely those of th...{{dropped:11}}