[Bioc-devel] developing R package for new release

[James W. MacDonald]

On Fri, Mar 17, 2017 at 3:45 PM, Nicholas Clark <nicholas.clark00 at gmail.com>
wrote:

> I’m planning on adding some new features to my package (GRmetrics) before
> this upcoming release and I have a few development questions.
>
>
> 1) Which version of R should I be using? I looked at this page (
> http://bioconductor.org/developers/how-to/useDevel/
> ), but I was still confused as to whether I should use the recently
> released R 3.3.3 or the R-devel 3.4.0 at http://r.research.att.com/


For the spring release you should use R-devel. For the fall release you
should use the most current version of R. This is because we release twice
a year, but R is only released in spring.

Looking at the useDevel page I see that I am simply repeating (almost
verbatim) what is written there. Is there something unclear about this that
we should address? It seems clear enough to me, but I've been doing this a
long time.


>
> .
>
>
> 2) What is the best/easiest way to work with github? Should I fork the
> repository from the read-only Bioconductor repo and work on that or
> maintain a separate repo? Or does it not matter?
>

I believe you should fork the repo from the Bioconductor mirror and use
that. There is some info here (
http://bioconductor.org/developers/how-to/git-mirrors/). But there are some
considerations to be made. Right now, we are using subversion for version
control, and anything you do in github has to be 'subversionized' in order
for your commits to be checked into the main version control repository.
It's far easier IMO to just use subversion to do all your version control,
because you don't have to worry about getting git to talk to subversion.

That said, after the April release we are transitioning to git, so we will
be using git soon enough. But for now I am still using SVN because it's a
direct commit to the main repo. My advice is to use whatever the main repo
is based on, because mixing and matching adds an extra level of complexity
that doesn't appear to have much to offer in return.


>
> 3) Should I make a “release-3.5” branch and commit changes there, or
> should it be “devel”, or just the “master” branch? http://bioconductor.
> org/developers/how-to/git-mirrors/
>  talks about this, but again, it’s not clear to me.
>

You never make your own branches. That's controlled by BioC core. Unless
you are fixing a bug in the release version of your package you should be
using the master branch (or if you use SVN, you should just be checking out
of the trunk,
https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/GRmetrics).


>
> 4) It looks like my package is failing the Windows build because it
> requires “SummarizedExperiment”, which is failing the Windows build. Is
> there anything I can do to fix this? Apologies if this has already been
> addressed. The error is: ERROR: dependency ‘SummarizedExperiment’ is not
> available for package ‘GRmetrics' (http://bioconductor.org/
> checkResults/release/bioc-LATEST/GRmetrics/tokay1-buildsrc.html
> )
>

If one of your dependencies is failing, and in particular if the dependency
is maintained by Bioc core, then the best thing to do is wait. If you are
relying on somebody else's package and they are consistently failing you
might contact them and find out if you can help.

Jim



>
> Best,
> Nick Clark
>
>
>
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> Bioc-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel




-- 
James W. MacDonald, M.S.
Biostatistician
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099

	[[alternative HTML version deleted]]