[BioC] package structure

[Vincent Carey]

In addition to Valerie's suggestions, you might consider an
interesting approach to dynamic response to package loading in the
oligoClasses/oligo stack.  The packageEvent and setHook functions of
base R are used quite effectively there -- see oligoClasses/R/zzz.R

On Fri, Feb 25, 2011 at 5:43 PM, Valerie Obenchain <vobencha at fhcrc.org> wrote:
> Hi Robert,
>
> I don't think you need to create two packages. You can put Rgraphviz and
> Cairo in the Suggests field of your package DESCRIPTION file.
> Using the .onLoad function in a zzz.R file you can include a check if
> the user has the packages loaded and set a global varaiable.
> The global variable can be used to dictate if code dependent on
> Rgarphviz and Ciaro is run.
>
> For example,
>
> In a zzz.R file, define a global variable
>
>    .options <- new.env(parent = emptyenv())
>
> Use the .onLoad function to set the global variable,
>
>    .onLoad <- function(...) {
>        .options[["foo.available"]] <- require(foo, quietly = TRUE)
>    }
>
> In your package code, check the value of the global variable to execute
> code dependent on the graph packages,
>
> if(.options[["foo.available"]]) do something ... else do something else ....
>
>
> Valerie
>
>
>
> On 02/25/2011 12:15 PM, Robert M. Flight wrote:
>> Hi All,
>>
>> I'm looking for some advice on package structure and dependencies.
>>
>> I'm working on a package that allows for different types of
>> meta-analyses between different datasets. As part of the original
>> development leading up to creating a package, I also generated a
>> series of functions that do a lot of work to generate navigable
>> results using HTML imagemaps.
>>
>> However, the visualizations depend currently on Rgraphviz (layouts)
>> and Cairo (PNG figure generation). The results can be summarized in
>> different ways than the visualizations I have developed (potentially
>> using Cytoscape, or just as a table), and I want to make the package
>> as easy to install and use as possible. Therefore, if I can make the
>> analysis package not require Rgraphviz and Cairo, I think that would
>> be good.
>>
>> Would it be advisable to create a second package containing the
>> functions necessary for creating the visualizations? Otherwise I
>> introduce dependencies that may not even be used by the users of the
>> package. And I don't know of any other proper way (i.e. besides
>> "require", which I believe is to be avoided in packages) to call other
>> packages than having them in the DESCRIPTION and NAMESPACE files, and
>> I believe anything in NAMESPACE requires that the package be
>> installed.
>>
>> I would appreciate others thoughts on this.
>>
>> Thanks,
>>
>> -Robert
>>
>> Robert M. Flight, Ph.D.
>> University of Louisville Bioinformatics Laboratory
>> University of Louisville
>> Louisville, KY
>>
>> PH 502-852-1809 (HSC)
>> PH 502-852-0467 (Belknap)
>> EM robert.flight at louisville.edu
>> EM rflight79 at gmail.com
>>
>> Williams and Holland's Law:
>>        If enough data is collected, anything may be proven by
>> statistical methods.
>>
>> _______________________________________________
>> Bioconductor mailing list
>> Bioconductor at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
>
>
>        [[alternative HTML version deleted]]
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
>