[R-pkg-devel] Errors from Suggests or Enhances not in mainstream repositories

[Uwe Ligges]

On 03.07.2018 00:41, Ben Bolker wrote:
> Thanks, Uwe.  Responses inline.
> 
> On Mon, Jul 2, 2018 at 6:23 PM Uwe Ligges
> <ligges using statistik.tu-dortmund.de> wrote:
>>
>>
>>
>> On 03.07.2018 00:13, Ben Bolker wrote:
>>> I got something similar.  I have a few thoughts:
>>>
>>> (1) you should use  "if (require(citrus)) { ... }" in your examples;
>>> "Suggests" and "Enhances" packages are supposed to be *optional*, i.e.
>>> examples and tests should be able to run even if they're not installed
>>> (I would imagine that "Writing R Extensions" and/or the CRAN policies
>>> say something to this effect, but I haven't checked)
>>> (2) I can imagine that win-builder is looking for *binary* versions of
>>> the package (I'm not sure why this would be the default, and why it
>>> couldn't fall back and install from source, but that would be my
>>> guess); do you have binary versions, for the appropriate major R
>>> version (i.e. 3.5 for release, maybe?? 3.4 for oldrelease??), in your
>>> repo. (In my example, I don't; I was going to try compiling a binary
>>> version and putting it in the repo. This is a bit of a chicken-and-egg
>>> problem if you can't get a binary built because the examples fail, but
>>> if you implement step #1 I think it should work).
>>>
>>> My point (3) is a follow-on question that I would love an answer to,
>>> or hints. I get a *message* similar to the one found in throm's
>>> win-builder logs, i.e. "Suggests or Enhances not in mainstream
>>> repositories: ..." - but since it's not labeled as a WARNING or an
>>> ERROR (or even a NOTE), I am wondering what its status is in the
>>> CRAN-maintainers' eyes - is it a CRAN-submission-blocking problem? Or
>>> a "we'd rather you didn't do that but we'll accept it this time?" or
>>> ...
>>
> Uwe> Ideally packages are in mainstream repositories.
> 
>    Agreed.
> 
> Uwe> For waek dependencies such as Suggests or Enhances, and if the package
> Uwe> is in some repository, you can declare it via the
> Uwe> Additional_Repositories field.
> 
>    I did that (so did the OP). Speaking for myself (since I'm not sure
> of the OP's exact situation), I know that the source tarball is in my
> repo, and it gets installed successfully on Linux and MacOS. On
> Windows (win-builder), the package is not detected in the repo. I can
> try adding a Windows binary to the repo to see if that solves the
> problem.

Ah, no, it won't be installed. We only isntall from mainstream repositories.



>    FWIW, one of the off-CRAN packages I Suggest: is INLA (which is
> beyond my control). The other is glmmADMB, which includes a binary
> blob and can't go on CRAN.  I would like to offer some support for
> these packages, if at all possible.
> 
> I just don't know if the message about the package *not* being found
> in the repo suggested by Additional_repositories: will be
> CRAN-blocking (and would prefer not to waste everyone's time by
> submitting if it is ...)

If it is correctly declared, it is not blocking.

Best,
Uwe

> Uwe> Finally, if not even in a repository, explain in the Descripton field
> Uwe> how to get it.
>>
>> Best,
>> Uwe Ligges
>>
>>>
>>>      I'm not asking anyone (except possibly CRAN maintainers) to speak
>>> for the CRAN maintainers, but if anyone has any experience or guesses
>>> (or pointers to parts of the CRAN policies/Writing R Extensions
>>> document that I overlooked) I'd appreciate hearing about it.
>>>
>>>     cheers
>>>        Ben Bolker
>>> On Mon, Jul 2, 2018 at 5:28 PM Throm, Allison <throm using wustl.edu> wrote:
>>>>
>>>> Hi all,
>>>>
>>>>
>>>> I'm trying to submit a package to the CRAN that further examines results from the "citrus" package as done for an accepted publication (available on Github: <https://github.com/nolanlab/citrus>).
>>>>
>>>>
>>>> I do not get any errors when using devtools::check() on my Mac, which already has the citrus package installed. However, I am getting errors during the CRAN submission check <https://win-builder.r-project.org/incoming_pretest/mineCitrus_1.0.0_20180702_172340/>.
>>>>
>>>>
>>>> I've tried searching Stack Overload and other CRAN packages with Additional_repositories fields, but none of the applicable solutions have worked. I'm pretty sure it's the loading of the citrus package that is causing the error and have added the Additional_repositories field with a link the package on Github, but still haven't been able to resolve the issue. There are very specific installation instructions for the package <https://github.com/nolanlab/citrus/wiki/Installing-Citrus>, which may be contributing to issues.
>>>>
>>>>
>>>> How should I properly specify dependencies on the nolanlab/citrus<https://github.com/nolanlab/citrus> Github package for my CRAN package?
>>>>
>>>>
>>>> Thank you in advance,
>>>>
>>>> Allison
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>           [[alternative HTML version deleted]]
>>>>
>>>> ______________________________________________
>>>> R-package-devel using r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>>>
>>> ______________________________________________
>>> R-package-devel using r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>>>