Skip to content

[R-pkg-devel] Advice on resubmitting when no response from CRAN

5 messages · Ben Bolker, David B. Dahl

#
Hi colleagues.  I am hoping to get some advice from you on how to get
a package "unstuck" at CRAN without antagonizing the busy CRAN
maintainers.

I submitted to CRAN an update to one of my packages on August 5.  The
submission passed the automatic tests on Debian, but not Windows.  The
automated response said, "If you are fairly certain the rejection is a
false positive, please reply-all to this message and explain."  I
replied-all and noted that the failure should be resolved by please
installing software as described in the SystemRequirements field and
the referenced INSTALL file.  After three weeks of not hearing back, I
sent a polite nudge (again via reply-all) on August 26 and suggested
an alternative if my original solution was not feasible.  I still have
not heard back and notice that now my package source is no longer on
https://cran.r-project.org/incoming/archive/.

How would you recommend that I proceed in a professional fashion?  Thanks!

-- David B. Dahl
#
One question and one suggestion.

    Is the software required/indicated by the SystemRequirements field 
unusual/possibly burdensome to install across platforms? That wouldn't 
justify not responding to your e-mails, but it might explain why the 
package didn't make it onto CRAN.

   For some context, can you tell us what the system 
requirement/alternative is?

   The suggestion would be to politely nudge again.
On 9/14/22 11:24 AM, David B. Dahl wrote:

  
    
#
Thanks for the reply, Ben.

The SystemRequirements is "Cargo (>= 1.56) for installation from source:
see INSTALL file".  The INSTALL notes that the package contains Rust source
code compiled by Cargo (the Rust package manager) which can be obtained
from the Rust project at https://rustup.rs.  On Windows, this involves
downloading an executable and following the onscreen instructions.

In my follow-up email, I wrote, "Alternatively, if you prefer, could these
packages please download
binary libraries produced by Cargo in accordance with the CRAN policy:
'Only as a last resort and with the agreement of the CRAN team should a
package download pre-compiled software.'"

So you suggest I politely nudge instead of resubmitting?  I don't think
that CRAN even has a copy of my submission anymore.  (It was here
https://cran.r-project.org/incoming/archive/ for several weeks, but is now
gone.)

Thanks!

-- David
On Wed, Sep 14, 2022 at 9:47 AM Ben Bolker <bbolker at gmail.com> wrote:

            

  
  
#
To be honest, I'm not sure about nudging vs resubmitting.

    I guess since they don't have a copy of the package handy, you might 
as well resubmit, with however much of the history as seems appropriate 
in your submission note.

   I guess the problem is for salso, bumping the required Cargo version 
from 1.51 to 1.56 ... ?
On 9/14/22 12:31 PM, David B. Dahl wrote:

  
    
#
Yes, the package in question is my salso package (but the issue
affects several of my other R packages that I'd like to update).

The problem is not so much the Cargo version number, but rather that
the previous version of my salso package downloaded a binary for a
static library if Cargo is not available during installation.  This is
the approach of other Rust-based packages (e.g., gifski, string2path,
ymd) but we were asked by a CRAN maintainer to submit updates which do
not do this (and that better documented authorship of depending
software) before 2022-08-10 to ensure the packages are not pulled from
CRAN.  That date has long passed and our packages are still on CRAN,
but I have not been able to update my package.  As an aside, the
author of string2path was able to update through the automatic checks
by still relying on the fallback of downloading a binary.  The updated
package is more clear about this fallback behavior, but his emails
requesting explicit permission have also not been answered.

-- David
On Wed, Sep 14, 2022 at 12:19 PM Ben Bolker <bbolker at gmail.com> wrote: