[Rd] "bug report" field in DESCRIPTION file?

Duncan Murdoch murdoch at stats.uwo.ca
Thu Nov 26 16:00:03 CET 2009


On 26/11/2009 9:20 AM, Martin Maechler wrote:
> >>>>> "DM" == Duncan Murdoch <murdoch at stats.uwo.ca>
> >>>>>     on Thu, 26 Nov 2009 08:39:27 -0500 writes:
>
>     DM> On 26/11/2009 7:09 AM, Barry Rowlingson wrote:
>     >> A lot of R packages are now effectively maintained by several people
>     >> and so use sites like R-forge or google code for development. This
>     >> means the best way to report bugs or problems with these packages is
>     >> via the development site's bug tracking rather than emailing the
>     >> maintainer. Could we agree on a field in DESCRIPTION explicitly for
>     >> bug reports?
>     >> 
>     >> The DESCRIPTION file has an optional URL field which generally sends
>     >> the user to the 'home page' of the package. Whether or not there is a
>     >> bug tracker somewhere there isn't made explicit, so I don't think
>     >> overloading URL for a bug report address is a good idea. The Hmisc
>     >> package mentions its bug tracker in the Description field and also has
>     >> it in the URL field with three other URLs. This tells me a bug report
>     >> field might be a better idea.
>     >> 
>     >> I'd prefer an optional field called 'BugReports:', which would be a
>     >> URL, and this could either be the http: address of a web site bug
>     >> tracker or a mailto: URL (of a real live human or a mail-based
>     >> tracker).
>     >> 
>     >> If there's agreement on this then a further step may be to write a
>     >> 'bugreport(package)' function that would first look for a BugReport
>     >> field, then the URL or maintainer fields to give the poor confused
>     >> user some advice on what to do. Then when someone emails R-dev saying
>     >> they think there's a bug in package foo, we can say "Have you read the
>     >> output of 'bugreport("foo")'?" which might be more helpful than saying
>     >> 'bugs with 'foo' should be reported to the maintainer'.
>     >> 
>
>     DM> This sounds like a good idea, though I would add a "package" parameter 
>     DM> to the bug.report() function, rather than creating a new function.
>
> I agree (good idea;  use  bug.report() with arguments).
>
>     DM> Does the logic below sound right for bug.report() with the package 
>     DM> specified?
>
>     DM> If there's a BugReports field, bug.report() calls browseURL() on that page.
>
> I'm not sure if that's easy:  One main reason for bug.report() is
> to auto-collect the necessary info and put it into the body of
> an e-mail message.  So, I think the above only "works" when the
> BugReports fields is a 'mailto:' URL, but not, e.g., when it
> points to an R-forge bug tracking web page form.
>   

The idea was that we would not use the auto-collection at all if there 
was a BugReports field, because the package author(s) had indicated by 
it that they didn't want emailed bug reports.

Duncan Murdoch
>     DM> If not, it does more or less what it does now, but
>     DM> - it defaults "address" to the package maintainer. 
>     DM> - it adds a line in the intro to the message pointing to the URL field 
>     DM> if there was one.
>
> that sounds good.
> Martin
>
>
>     DM> Duncan Murdoch
>
>
>     >> Just an idea for a rainy morning...
>     >> 
>     >> Barry
>     >> 
>     >> ______________________________________________
>     >> R-devel at r-project.org mailing list
>     >> https://stat.ethz.ch/mailman/listinfo/r-devel
>     >> 
>
>     DM> ______________________________________________
>     DM> R-devel at r-project.org mailing list
>     DM> https://stat.ethz.ch/mailman/listinfo/r-devel
>



More information about the R-devel mailing list