Re: [Rd] Feature request: display file name in R CMD check warning

From: Dan Tenenbaum <dtenenba_at_fhcrc.org>
Date: Fri, 18 Mar 2011 12:47:00 -0700

On Thu, Mar 17, 2011 at 6:49 PM, Duncan Murdoch <murdoch.duncan_at_gmail.com>wrote:

> On 11-03-17 12:33 PM, Dan Tenenbaum wrote:
>
>> On Thu, Mar 17, 2011 at 2:36 AM, Duncan Murdoch
>> <murdoch.duncan_at_gmail.com> wrote:
>>
>>> On 16/03/2011 7:55 PM, Dan Tenenbaum wrote:
>>>
>>>>
>>>> Hi,
>>>>
>>>> I came across the following warning in R CMD check (it only occurred on
>>>> Windows):
>>>>
>>>> The \usage entries for S3 methods should use the \method markup and not
>>>>
>>>>>
>>>>> their full name.
>>>>> See the chapter 'Writing R documentation files' in manual 'Writing R
>>>>> Extensions'.
>>>>>
>>>>
>>>>
>>>> The package I'm looking at is one that I did not write which has 34 .Rd
>>>> files. This warning does not tell me which file to look in. It would be
>>>> very
>>>> helpful if it did. Same goes for other warnings/errors produced by R CMD
>>>> check.
>>>>
>>>
>>> I was unable to duplicate this. When I tried it by messing up one of the
>>> man pages in the ellipse package, I got this:
>>>
>>> S3 methods shown with full name in documentation object 'ellipse.glm':
>>> ellipse.glm
>>>
>>> The \usage entries for S3 methods should use the \method markup and not
>>> their full name.
>>> See the chapter 'Writing R documentation files' in manual 'Writing R
>>> Extensions'.
>>>
>>> "Documentation object 'ellipse.glm'" tells me the \name{} inside the .Rd
>>> file, which is enough to uniquely identify the file. Are you not seeing
>>> this part of the message?
>>>
>>>
>>
>> No, I'm not.
>>
>> I still can't identify the offending file, but you can download the
>> whole package:
>>
>> svn --username readonly --password readonly export
>> https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/affyILM/
>>
>> Then run R CMD check on this package. You may need to install
>> dependencies.
>>
>
> I installed a lot of dependencies, but couldn't trust check because of this
> warning:
>
> Found the following significant warnings:
> Warning: running command 'C:\WINDOWS\system32\cmd.exe /c ftype perl' had
> status 2
>
> I'm not sure where this comes from; I don't think bgafun uses perl, so it's
> from one of the dependencies. But I think the error message you are seeing
> is spurious: something else is going wrong when check tries to check \usage
> sections, and so it reports that there was an error.
>
> As a sort of confirmation of this, I removed the dependencies from the
> DESCRIPTION file and tried to run check; I got lots of errors because of the
> missing dependencies now, but the one about S3 methods went away.
>
> Not sure what to suggest to diagnose this; I'm not familiar with most of
> those packages I just installed as dependencies. But I think it's safe to
> say that you shouldn't worry about the \usage sections.
>
> If you do figure out what's going wrong, please let us know because it
> would probably be a good idea to fix the usage checks so they give the right
> message.
>
>
You're right, it's a spurious warning. What causes it is the following line in findPerl.R in the package gdata (bgafun depends on made4 which depends on gplots which depends on gdata):

      perl.ftype <- shell("ftype perl", intern = TRUE)

This command returns the warning you saw. I don't think it's a reliable way to check if perl is installed, as it returned nothing even on a system where perl was installed.

Something about this shell command is screwing up R CMD build and producing the spurious warning about using \method markup. I'm not sure what the exact mechanism is, but I can reliably cause it (or eliminate it) by either leaving that line as is or changing it to

    perl.ftype <- ""

There is something weird about the ftype command and the way it interacts with the shell. If I do:

ftype | more

and press 'q' after the first screenful of info, I see this (many times): The process tried to write to a nonexistent pipe.

I don't see this behavior with other Windows commands.

I'm not sure if that helps you track this down or not, but that's what I am able to figure out.
Thanks
Dan

> Duncan Murdoch
>

        [[alternative HTML version deleted]]



R-devel_at_r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel Received on Fri 18 Mar 2011 - 19:55:04 GMT

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Fri 18 Mar 2011 - 21:50:33 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-devel. Please read the posting guide before posting to the list.

list of date sections of archive