Skip to content

[R-pkg-devel] Package Rejection Error Help

10 messages · Andrew Dhawan, Patrick Giraudoux, Robert M. Flight +2 more

#
Dear all,

I am a bit puzzled by this error for the sigQC package being uploaded to
CRAN. Checks on my end pass without issue. There is one error regarding the
bioConductor dependency RankProd that comes up only on the Debian OS when
checking. See here for further details:
https://win-builder.r-project.org/incoming_pretest/sigQC_0.1.22_20210406_110708/

I am not sure how to resolve this - it seems to me that the dependency
*should* be available across platforms, and short of removing the
dependency (which we rely on a fair bit), I'm not sure what  we can do.

Any thoughts or suggestions would be massively appreciated.

Thank you so kindly!
Andrew
#
Two Notes:


Found the following (possibly) invalid URLs:
   URL: http://imagemagick.org/script/download.php (moved to 
https://imagemagick.org/script/download.php)
     From: inst/doc/vignette.html
     Status: 200
     Message: OK


Please change http --> https, add trailing slashes, or follow moved 
content as appropriate.

* checking LazyData ... NOTE
   'LazyData' is specified without a 'data' directory

Simply omit the field if you have no data.

Best,
Uwe Ligges
On 06.04.2021 16:27, Andrew Dhawan wrote:
#
I had a similar problem with intepreting "notes" in pgirmess checks. 
Some are lethal and lead to rejection and some not (just informative 
comments). Fortunately Uwe has been kind enough to tell me what I missed 
(I thought that all notes were just "informative").
Would it be possible to call lethal notes "errors" to keep us aware that 
we must really do something with them and fix what is actually a problem?
Best,
Patrick


Le 06/04/2021 ? 17:33, Uwe Ligges a ?crit?:

  
  
#
On 06.04.2021 17:39, Patrick Giraudoux wrote:
These are Notes as it may not be a serious issue or may be a false 
positive, nevertheless, we like to get clean packages to CRAN.

WARNINGs and ERRORs will never be accepted.

The policies clearly say that significant Notes have to be addressed.
The only Note that is always acceptable is the CRAN maintainer Note 
about the package maintainer's address. All other Notes should be fixed 
generally or need manual attention in case of the really very few 
exceptions.

Best,
Uwe Ligges
#
Thank you both very kindly - I have addressed both notes. I am still
wondering though about the RankProd dependency error - any insight into
this?

Thanks!
Andrew

On Tue, Apr 6, 2021 at 11:58 AM Uwe Ligges <ligges at statistik.tu-dortmund.de>
wrote:

  
  
#
Andrew,

I think if you want to include a Bioconductor dependency, then I think you
add the "biocViews:" line to your description file to enable bioconductor
package installation.

See this answer on Bioinformatics stackexchange:
https://bioinformatics.stackexchange.com/a/3375/51

Hope that helps!

-Robert
On Tue, Apr 6, 2021 at 1:14 PM Andrew Dhawan <adhawan at qmed.ca> wrote:

            

  
  
#
Hi Robert,

Thank you for replying! I already had the biocViews: line on my description
file - it seems it must be something else contributing to the issue.

Andrew
On Tue, Apr 6, 2021 at 13:26 Robert M. Flight <rflight79 at gmail.com> wrote:

            
*Andrew Dhawan, MD, **DPhil, **B.Math*
#
On 06.04.2021 19:26, Robert M. Flight wrote:
Not needed for CRAN.

You simply need to declare the dependency, but BioC 3.13 is what we 
should use to check with R-devel and that was recetnly still pretty 
iunstable.

Best,
Uwe Ligges
#
Thank you again Uwe,

So, to clarify - is the error occurring because RankProd is not found in
BioC 3.13 (current release is 3.12)?  If so, I'm not sure how I can fix
this...

Thanks
Andrew


On Wed, Apr 7, 2021 at 5:45 AM Uwe Ligges <ligges at statistik.tu-dortmund.de>
wrote:

  
  
#
On 07/04/2021 10:50 a.m., Andrew Dhawan wrote:
I suspect the only thing you could do in the short term is to remove 
that dependency (or move it to Suggests, and check for it before using). 
  If that's not feasible, then you just need to wait for it to become 
available.  Based on its status page, it shouldn't be long.  Try 
submitting again tomorrow?

Duncan Murdoch