Skip to content

[Bioc-devel] BiocInstaller not supporting R 3.4.0 -- yes, it's early, but shouldn't it default to using Bioc-devel instead of Bioc-release in this scenario?

4 messages · Martin Morgan, Leonardo Collado Torres

#
Hi,

This is definitely 4 weeks early and things are working as intended.
Due to https://github.com/travis-ci/travis-ci/issues/5809 I noticed
that BiocInstaller does not yet support R 3.4.0. The main issue was my
assumption that R-travis using `r: devel` would install R 3.3.0 as I
describe in the issue I linked above.

I was failing to see this because builds for `regionReport` were
completing successfully. However, builds for `derfinderPlot` failed
and that's because `derfinderPlot` has Bioc-devel dependencies
(S4Vectors >= 0.9.38). Thanks to Jim Hester I realized that both
packages were being tested on R 3.4.0, and that BiocInstaller after
failing would then default on Bioc-release. See
https://travis-ci.org/leekgroup/derfinderPlot/builds/117500745#L419-L420
and https://travis-ci.org/leekgroup/regionReport/builds/117499393#L418-L419

Bioconductor version 3.2 (BiocInstaller 1.20.1), ?biocLite for help
Bioconductor does not yet support R version 3.4.0

As for BiocInstaller, I'm a bit curious why it falls back to
Bioc-release instead of Bioc-devel when trying to use it in a future R
version that is not yet supported.

In any case, I can fix my R-travis builds by specifying that I want
them to be run on R 3.3.0 for now.

I'm guessing that other developers who use R-travis might run into
this issue in these 4 weeks before R 3.3.0 is released.

Best,
Leo

Leonardo Collado Torres, PhD Candidate
Department of Biostatistics
Johns Hopkins University
Bloomberg School of Public Health
Website: http://lcolladotor.github.io/about.html
#
Check Jim Hester's thoughts on this issue
https://github.com/travis-ci/travis-ci/issues/5809#issuecomment-199858950

One way of solving it is by having BiocInstaller allow using
Bioc-devel with R 3.4.0.

On Tue, Mar 22, 2016 at 10:43 AM, Leonardo Collado Torres
<lcollado at jhu.edu> wrote:
#
the biocLite.R script has been changed to use bioc-devel when the R 
version is 'too new'.

Worth pointing out perhaps that Bioconductor will not use R-devel again 
for another six months -- after our next release, both release and devel 
versions of Bioconductor will use the R-3-3-branch, because that will be 
the 'user' version of R available for both the 3.3 and 3.4 releases of 
Bioconductor.

Thanks for the request, Leo.

Martin
On 03/22/2016 11:24 AM, Leonardo Collado Torres wrote:
This email message may contain legally privileged and/or confidential information.  If you are not the intended recipient(s), or the employee or agent responsible for the delivery of this message to the intended recipient(s), you are hereby notified that any disclosure, copying, distribution, or use of this email message is prohibited.  If you have received this message in error, please notify the sender immediately by e-mail and delete this email message from your computer. Thank you.
#
Thanks for the new feature Martin! I also have to thank Jim Hester for
adding support for R version 3.3.0 on Travis.

I know that we'll use R-3-3-branch for Bioc 3.4, but it's good to
mention it on the thread for reference.

Best,
Leo

On Tue, Mar 22, 2016 at 3:49 PM, Martin Morgan
<martin.morgan at roswellpark.org> wrote: