[Bioc-devel] library checking: question on require/depends

Hi,

On Tue, Jul 29, 2008 at 2:43 PM, Laurent Gautier <lgautier at gmail.com> wrote:

Adding it in "Suggests" will keep R CMD check happy, as Sean told it.

Thanks, I just did not know that still I needed to have all the
"Suggests" library installed. Now the warning is gone.

One question, still:
Are you using the those packages in examples ? In that case, using a
smaller dataset might be a good idea.
If there is not any suitable smaller dataset currently avalaible, it
could be nice to create a demo-oriented one, with one chromosome for
example. This could also be used by other packages needing full
genome.

Yes, I'm using one of the suggested packages in the examples and I'm
releasing a small dataset with the library, just for demonstration
purposes.

2008/7/29 Sean Davis <sdavis2 at mail.nih.gov>:

On Tue, Jul 29, 2008 at 1:51 PM, mattia pelizzola
<mattia.pelizzola at gmail.com> wrote:

Hi,

I'm working on a new package. The 'build' is ok, but when running
'check' I get a warning that I'm not sure on how to eliminate it.
The warning is :

* checking for unstated dependencies in R code ... WARNING
'library' or 'require' calls not declared from:
 BSgenome.Hsapiens.UCSC.hg17 BSgenome.Hsapiens.UCSC.hg18

The reason of course is that I'm using the require function and I'm
not adding these libraries in 'Depends' or 'Suggests'. The point is
that I would not like to oblige users to install both hg17 and hg18
libraries, since they are huge (800Mb). This is going to be even more
important if I would support many other annotation libraries. The idea
was then to use require to load only the one requested from the user
(either hg17 or hg18).

I'm going to submit this library to Bioconductor. Can I ignore this
warning, should I add all the annotation libraries in 'Depends', or is
there another way to fix this warning?

This is what "Suggests" is for.

Sean