Re: [Rd] Namespace dependency not required

[Patrick Giraudoux]

Le 18/03/2012 19:30, Uwe Ligges a écrit :


On 18.03.2012 19:27, Patrick Giraudoux wrote:
Le 18/03/2012 18:02, Uwe Ligges a écrit :


On 18.03.2012 16:08, Patrick Giraudoux wrote:
Le 18/03/2012 14:57, Uwe Ligges a écrit :


On 18.03.2012 11:21, Patrick Giraudoux wrote:
Hi,

I am working at adding namespace to my packages, carefully
following the
doc "Writing R extensions" and some threads on the web. However I
cannot
find clear explanation about how to best deal with the import or
importFrom functions in the name space. To make it short:

To declare dependencies in the description file either after 
Depends:
(packages including functions that are called from a package 
function)
or after Suggests: (packages that are called eg from the 
doucmentation
examples), works well. In this case one does not need to declare
import
in the namespace file.

However, declaring dependencies sometimes for using only one
function in
a given package looks a bit over the top, and I though I could use
importFrom in the namespace to deal with that. So removing the
corresponding declaration in Depends (description file) and 
declaring
eg importFrom(splancs, inout) in the namespace always led to:

* checking package dependencies ... ERROR
Namespace dependency not required: 'splancs'

Same thing with import(sp). I have the same trouble if I do not 
remove
the declarations in Depends.

So, I can hardly follow the meaning of the "Writing R extension" 
doc:

Packages implicitly import the base namespace. Variables exported 
from
other packages with namespaces need to be imported explicitly using
the
directives |import| and |importFrom|. The |import| directive
imports all
exported variables from the specified package(s). Thus the 
directives

import(foo, bar)

specifies that all exported variables in the packages *foo* and 
*bar*
are to be imported. If only some of the exported variables from a
package are needed, then they can be imported using |importFrom|. 
The
directive

importFrom(foo, f, g)

specifies that the exported variables |f| and |g| of the package 
*foo*
are to be imported.

Can anybody tell us the conditions for which import and importFrom
commands should be used in the NAMESPACE rather than in the Depends
and
Suggests declarations of the description file, and the conditions of
their applicability (without error message... eg with a working
example) ?

If you import from another namespace (which is the recommended way
rather than just relying on search order), do not forget to 
declare it
as "Imports" in your DESCRIPTION file. Your package depends on the
other one: It could not be loaded without having the other one
installed now.

Uwe Ligges


OK. got it (more or less).

With this in the description file:

Depends: boot (>= 1.3-4), nlme(>= 3.1-64), rgdal (>= 0.7-8), sp (>=
0.9-97), spdep (>= 0.5-43), splancs (>= 2.01-31)

importFrom(splancs, inout) is accepted in the NAMESPACE file and I 
have
no error message with rcmd check.

Irr works also with:

Depends: boot (>= 1.3-4), nlme(>= 3.1-64), rgdal (>= 0.7-8), sp (>=
0.9-97), spdep (>= 0.5-43)
Imports: splancs (>= 2.01-31)

Practically, I a am safe (thanks a lot), but I am however still hungry
to learn a bit more. I hardly understand what R does when reading the
suggests and imports in the DESCRIPTION file compared to what it does
reading the NAMESPACE (my mistake was to believe that one was a
substitute for the other). From a practical point of view, why is it
important to use import or importFrom in the NAMESPACE when everything
looks like working well when declarations are done in the DESCRIPTION
file only (eg in depends or imports)? It is not crystal clear to me 
even
reading carefully "Writing R extension". DESCRIPTION file is used 
to get
the "list of packages that will be attached (...) before the current
package" (indeed, now, I no longer needs to use 'require' within my
package functions). In simple words, what may add the NAMESPACE to
that ?

Apologize to be so poorly aware (= so much stupid) about the role of
each of those declaration spaces.

Please read some literature about NAMESPACES, e.g. documentation in
Writing R Extensions plus Luke Tierney's articles in R News.

ok. Will do it with Luke Tierney's articles. I read Writing R extensions
already and also S programming (Venables and Ripley), but indeed should
come down to more basic text for dudes :-) .

S Programming is excellent, but it was written before Namespaces were 
introduced to R.

Uwe Ligges





Best,

Patrick



A NAMESPACE import imports functionality of another namespace into
your own namespace and hence avoids problems caused by name clashes.
The DESCRIPTION file is required to derive dependency structure of
packages, i.e. derive which package has to be *installed* first etc.

Uwe Ligges





Great ! Luke Tierney's article in R news is particularly clear (to me) 
with nice examples on how namespace avoids variable shadowing. I give 
the link here for any user/developper who may follow the thread on this 
issue: http://journal.r-project.org/archive.html Volume 3/1 June 2003, 
Name Space Management for R by Luke Tierney.

Thanks Uwe for the hint.

Best,

Patrick

______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel