R devel - Jul 2019 - install packages with missing pkg argument

Dear all,

The help for `?install.packages` decribes, in the `pkg` argument
description :
> If this is missing, a listbox of available packages is presented where
possible in an interactive R session.

In fact running it with a missing argument triggers an error :

install.packages()
> Error in install.packages : argument "pkgs" is missing, with no
default
What however doesn't trigger an error is callinginstall.packages on a zero
length character :

install.packages(character(0))

On my colleague's R 3.5.1 windows installation it shows the listbox of
available packages, on my 3.6.0 installation it pauses for a couple seconds
and doesn't do anything.

A character vector of length zero is what you get when you compute an empty
`setdiff` or `intersection`, so it was very surprising to us to see
something popup where we were expecting a vector of missing packages to be
installed (or none if there was no missing package).

I believe having the function work as advertised with a proper missing
argument, and having it do nothing silently when the argument is of length
zero, would make more sense.

Best regards,

Antoine

	[[alternative HTML version deleted]]

Are you running that command in RStudio? And do you get the documented
results when you run utils::install.packages() rather than just
install.packages()

If yes, then the function is likely working as advertised and you've
mixed up the R and RStudio versions


On 29/07/2019, Ant F <antoine.fabri at gmail.com>
wrote:
> Dear all,
>
> The help for `?install.packages` decribes, in the `pkg` argument
> description :
>
>> If this is missing, a listbox of available packages is presented where
> possible in an interactive R session.
>
> In fact running it with a missing argument triggers an error :
>
> install.packages()
>> Error in install.packages : argument "pkgs" is missing, with
no default
>
> What however doesn't trigger an error is callinginstall.packages on a
zero
> length character :
>
> install.packages(character(0))
>
> On my colleague's R 3.5.1 windows installation it shows the listbox of
> available packages, on my 3.6.0 installation it pauses for a couple seconds
> and doesn't do anything.
>
> A character vector of length zero is what you get when you compute an empty
> `setdiff` or `intersection`, so it was very surprising to us to see
> something popup where we were expecting a vector of missing packages to be
> installed (or none if there was no missing package).
>
> I believe having the function work as advertised with a proper missing
> argument, and having it do nothing silently when the argument is of length
> zero, would make more sense.
>
> Best regards,
>
> Antoine
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

Maybe related to this?

    https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17556
>>>>> "Antoine" == Ant F <antoine.fabri at
gmail.com> writes:
    Antoine> Dear all,
    Antoine> The help for `?install.packages` decribes, in the `pkg` argument
    Antoine> description :

    >> If this is missing, a listbox of available packages is presented
where
    Antoine> possible in an interactive R session.

    Antoine> In fact running it with a missing argument triggers an error :

    Antoine> install.packages()
    >> Error in install.packages : argument "pkgs" is missing,
with no default

    Antoine> What however doesn't trigger an error is
callinginstall.packages on a zero
    Antoine> length character :

    Antoine> install.packages(character(0))

    Antoine> On my colleague's R 3.5.1 windows installation it shows the
listbox of
    Antoine> available packages, on my 3.6.0 installation it pauses for a
couple seconds
    Antoine> and doesn't do anything.

    Antoine> A character vector of length zero is what you get when you
compute an empty
    Antoine> `setdiff` or `intersection`, so it was very surprising to us to
see
    Antoine> something popup where we were expecting a vector of missing
packages to be
    Antoine> installed (or none if there was no missing package).

    Antoine> I believe having the function work as advertised with a proper
missing
    Antoine> argument, and having it do nothing silently when the argument is
of length
    Antoine> zero, would make more sense.

    Antoine> Best regards,

    Antoine> Antoine


-- 
Enrico Schumann
Lucerne, Switzerland
http://enricoschumann.net

Indeed Hugh, apologies for the oversight, I've reported this to Rstudio, as
I believe it is problematic to get an irrelevant help file when calling
`?install.packages`.

https://github.com/rstudio/rstudio/issues/5154

Best regards,

Antoine

Le lun. 29 juil. 2019 ? 13:59, Hugh Parsonage <hugh.parsonage at
gmail.com> a
?crit :
> Are you running that command in RStudio? And do you get the documented
> results when you run utils::install.packages() rather than just
> install.packages()
>
> If yes, then the function is likely working as advertised and you've
> mixed up the R and RStudio versions
>
>
> On 29/07/2019, Ant F <antoine.fabri at gmail.com> wrote:
> > Dear all,
> >
> > The help for `?install.packages` decribes, in the `pkg` argument
> > description :
> >
> >> If this is missing, a listbox of available packages is presented
where
> > possible in an interactive R session.
> >
> > In fact running it with a missing argument triggers an error :
> >
> > install.packages()
> >> Error in install.packages : argument "pkgs" is missing,
with no default
> >
> > What however doesn't trigger an error is callinginstall.packages
on a
> zero
> > length character :
> >
> > install.packages(character(0))
> >
> > On my colleague's R 3.5.1 windows installation it shows the
listbox of
> > available packages, on my 3.6.0 installation it pauses for a couple
> seconds
> > and doesn't do anything.
> >
> > A character vector of length zero is what you get when you compute an
> empty
> > `setdiff` or `intersection`, so it was very surprising to us to see
> > something popup where we were expecting a vector of missing packages
to
> be
> > installed (or none if there was no missing package).
> >
> > I believe having the function work as advertised with a proper missing
> > argument, and having it do nothing silently when the argument is of
> length
> > zero, would make more sense.
> >
> > Best regards,
> >
> > Antoine
> >
> >       [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-devel at r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-devel
> >
>
	[[alternative HTML version deleted]]

On 29/07/2019 7:39 a.m., Ant F wrote:
> Dear all,
> 
> The help for `?install.packages` decribes, in the `pkg` argument
> description :
> 
>> If this is missing, a listbox of available packages is presented where
> possible in an interactive R session.
> 
> In fact running it with a missing argument triggers an error :
> 
> install.packages()
>> Error in install.packages : argument "pkgs" is missing, with
no default
> 
> What however doesn't trigger an error is callinginstall.packages on a
zero
> length character :
> 
> install.packages(character(0))
> 
> On my colleague's R 3.5.1 windows installation it shows the listbox of
> available packages, on my 3.6.0 installation it pauses for a couple seconds
> and doesn't do anything.
> 
> A character vector of length zero is what you get when you compute an empty
> `setdiff` or `intersection`, so it was very surprising to us to see
> something popup where we were expecting a vector of missing packages to be
> installed (or none if there was no missing package).
> 
> I believe having the function work as advertised with a proper missing
> argument, and having it do nothing silently when the argument is of length
> zero, would make more sense.
> 
The change you are requesting is listed as having been done in the NEWS 
for 3.6.0...

Duncan Murdoch

Indeed it's a RStudio issue as diagnosed by Hugh, and is still an issue
with the latest version.

Le lun. 29 juil. 2019 ? 15:02, Duncan Murdoch <murdoch.duncan at
gmail.com> a
?crit :
> On 29/07/2019 7:39 a.m., Ant F wrote:
> > Dear all,
> >
> > The help for `?install.packages` decribes, in the `pkg` argument
> > description :
> >
> >> If this is missing, a listbox of available packages is presented
where
> > possible in an interactive R session.
> >
> > In fact running it with a missing argument triggers an error :
> >
> > install.packages()
> >> Error in install.packages : argument "pkgs" is missing,
with no default
> >
> > What however doesn't trigger an error is callinginstall.packages
on a
> zero
> > length character :
> >
> > install.packages(character(0))
> >
> > On my colleague's R 3.5.1 windows installation it shows the
listbox of
> > available packages, on my 3.6.0 installation it pauses for a couple
> seconds
> > and doesn't do anything.
> >
> > A character vector of length zero is what you get when you compute an
> empty
> > `setdiff` or `intersection`, so it was very surprising to us to see
> > something popup where we were expecting a vector of missing packages
to
> be
> > installed (or none if there was no missing package).
> >
> > I believe having the function work as advertised with a proper missing
> > argument, and having it do nothing silently when the argument is of
> length
> > zero, would make more sense.
> >
>
> The change you are requesting is listed as having been done in the NEWS
> for 3.6.0...
>
> Duncan Murdoch
>
	[[alternative HTML version deleted]]