Skip to content

[Bioc-devel] importing packages without a name space

7 messages · Laurent Gatto, Seth Falcon

Original
Laurent Gatto
# 
Dear list,

I have a problem with my NAMESPACE file. I need to import a package that
does not define a name space, and I always get the same error on
import(), specifying that the package I try to load 'does not have a
name space'. 
Note that I also need to import a class from that package with
importClassFrom() which seems to work.

I tried different syntaxes like import(package),
importFrom(package,variable), included .onLoad hook (as specified in
R-ext) in the zzz.R file and also modified the Depends line in
DESCRIPTION, which did not help. 

I am using the latest R 2.5.0 version.

Am I missing something or is it not possible to import a package without
name space?

Thank you very much in advance,

Laurent



--
Laurent Gatto
www.dnavision.be
Seth Falcon
# 
Hi Laurent,

"Laurent Gatto" <L.Gatto at dnavision.be> writes:
Actually, it is simply not possible to import from a package that does
not have a name space of its own.
That is news to me; I would not expect it to work.

I think the best you can do when you are using code from a package
that does not have a NAMESPACE file is to put this package in the
Depends field.  

The next thing to do is send a friendly email to the package's
maintainer and ask that they add a name space to their package :-)

+ seth
2 days later
Laurent Gatto
# 
Hello again,

Thanks to Seth and Holger for their answers, but I still can not get it
working. I will provide more details.

I have to load two packages, geneplotter and simpleaffy and my package
(yaqcaffy) defines YAQCStats, a subclass of simpleaffy's QCStats class.

As geneplotter provides a name space, I load the required function with
'importFrom(geneplotter,smoothScatter)' in NAMESPACE. I also added 
.onLoad <-  function(lib,pkg) require(methods) 
in R/zzz.R and geneplotter on the 'Imports' line in DESCRIPTION.

For simpleaffy, I do not import it in NAMESPACE, but add it as a
dependency in DESCRIPTION. Here is the .Fist.lib function as defined in
R/zzz.R:
.First.lib <-  function(lib,pkg,where) { 
    require(simpleaffy,quietly=TRUE);
    # Find what position in the search path this package is 
    where <- match(paste("package:", pkg, sep=""), search());
    .initClassesAndMethods(where);
    cacheMetaData(as.environment(where));
    .createYAQCEnvironment()
}

When I check the src of my package (or try to load it), I get the
following error:

Error in setIs(Class, class2, classDef = classDef, where = where) :
        Unable to find package environment for class "QCStats" to revise
subclass information	
Error in setClass("YAQCStats", contains = "QCStats",
representation(average.noise = "numeric",  :
        error in contained classes ("QCStats") for class "YAQCStats";
class definition removed from "yaqcaffy"
In addition: Warning message:
Couldn't find superclass "QCStats" to clean up when removing subclass
references to class "YAQCStats" in: .removeSuperclassBackRefs(Class,
classDef, classWhere)
Error : unable to load R code in package 'yaqcaffy'
Error: package/namespace load failed for 'yaqcaffy'
Execution halted

If I add 'importClassesFrom(simpleaffy, QCStats)' in the NAMESPACE file,
I get

Error : package 'simpleaffy' does not have a name space
Error: package/namespace load failed for 'yaqcaffy'
Execution halted

Note that I get the same error message if I ask to import simpleaffy in
NAMESPACE.

Below is my sessionInfo() output after loading simpleaffy and
geneplotter.

Thank you very much in advance,

Laurent

--
Laurent Gatto
www.dnavision.be
R version 2.5.0 beta (2007-04-11 r41127)
i686-pc-linux-gnu

locale:
LC_CTYPE=en_US.UTF-8;LC_NUMERIC=C;LC_TIME=en_US.UTF-8;LC_COLLATE=en_US.U
TF-8;LC_MONETARY=en_US.UTF-8;LC_MESSAGES=en_US.UTF-8;LC_PAPER=en_US.UTF-
8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=en_US.UTF-8;LC_ID
ENTIFICATION=C

attached base packages:
[1] "splines"   "tools"     "stats"     "graphics"  "grDevices" "utils"
[7] "datasets"  "methods"   "base"

other attached packages:
geneplotter     lattice    annotate  simpleaffy  genefilter    survival
   "1.13.8"    "0.15-3"    "1.13.8"     "2.9.3"   "1.13.14"      "2.31"
       affy      affyio     Biobase
  "1.13.19"     "1.2.0"   "1.13.49"


-----Original Message-----
From: Seth Falcon [mailto:sfalcon at fhcrc.org] 
Sent: Friday, April 13, 2007 7:59 PM
To: Laurent Gatto
Cc: bioc-devel at stat.math.ethz.ch
Subject: Re: [Bioc-devel] importing packages without a name space

Hi Laurent,

"Laurent Gatto" <L.Gatto at dnavision.be> writes:
that
Actually, it is simply not possible to import from a package that does
not have a name space of its own.
That is news to me; I would not expect it to work.

I think the best you can do when you are using code from a package
that does not have a NAMESPACE file is to put this package in the
Depends field.  

The next thing to do is send a friendly email to the package's
maintainer and ask that they add a name space to their package :-)

+ seth
Seth Falcon
# 
"Laurent Gatto" <L.Gatto at dnavision.be> writes:
If your package has a NAMESPACE file then the package initialization
function that will be run is .onLoad and .First.lib will be ignored.
So I'm pretty sure all of the code in .First.lib is not getting run.

But are you sure you need all of that code anyhow?  It looks to me
like you have copy/pasted from another package (I'd like to know which
one because perhaps it is not needed there either).  There was a time
long ago when S4 using packages needed such an arrangement, but in
general they no longer do (there are a few cases, but I doubt your
package is among them).

So I'm guessing that .initClassesAndMethods() contains all of your
setClass and setMethod calls.  Can you try just putting those at the
package level in your R code?  

What does .createYAQCEnvironment do?

Have you exported all classes and methods in your NAMESPACE file?

Perhaps you want to post your DESCRIPTION file as well, R is fairly
picky about the syntax and I'd like to rule that out.

+ seth
Laurent Gatto
# 
Hello,
Yes indeed, I also tried to R CMD check the package without .Fist.Lib
and it had no impact on the error message. I now moved
.createYAQCEnvironment() in .onLoad.
You are right. I started to write the code using simpleaffy as an
example, as it does similar things. Everything worked fine (and still
does) without the NAMESPACE file.
All the setClass and SetMethods are in the main R file.
The function creates a new environment and loads some data that is
stored in a file in the data directory of the package:

.createYAQCEnvironment <- function() {
  if(!exists(".yaqcEnv")) {
    .yaqcEnv <- new.env()
     data(morespikes,envir =.yaqcEnv )
     assign(".yaqcEnv",.yaqcEnv,envir=globalenv())
  }
}
Yes, I export the class with 'exportClasses("YAQCStats")', the formal
methods with exportMethods() and the other function with export().
Package: yaqcaffy
Title: Affymetrix expression data quality control and reproducibility
analysis
Version: 0.0.12
Author: Laurent Gatto
Description: Quality control of Affymetrix GeneChip expression data  
             and reproducibility analysis of human whole genome chips 
             with the MAQC reference datasets. 
Maintainer: Laurent Gatto <l.gatto at dnavision.be>
Depends: R (>= 2.0), simpleaffy, methods
Imports: geneplotter
biocViews: Microarray,OneChannel,QualityControl
License: The Artistic License, Version 2.0
URL: http://www.bioconductor.org

Thank you,

Laurent
Seth Falcon
# 
"Laurent Gatto" <L.Gatto at dnavision.be> writes:
In general, that just isn't a good moel for how to do things.  I am
not surprised that you've encountered difficulties with using a
NAMESPACE given that setup.
It would be much cleaner to put this environment in your package name
space and not muck with the globalenv.  How about putting this in one
of your .R files:

   .yaqcEnv <- new.env(parent=emptyenv(), hash=TRUE)
   data(morespikes, envir = .yaqcEnv)

And then you can decide whether this is something the user should have
access to by exporting this object in the NAMESPACE file or not.
Good.
That looks reasonable.  Are you still seeing the same error messages
or are things working better/different now?

+ seth