[Rd] Re: [R] adding option in the Windows installer for
Prof Brian Ripley
ripley at stats.ox.ac.uk
Mon Jul 26 18:07:08 CEST 2004
The direct connection provides you with better capabilities behind
firewalls if you set it up well enough. It also has the advantage of
working the same way on all R platforms.
I don't agree with the basic premise: --internet2 is only easier for
people whose sysadmins set up IE5/6 for them, and does not work with
authenticating proxies. As it loads an alternative DLL, the choice needs
to be made before internet functions are first used.
We could make this a configure option, but why is it so hard to point
people at the rw-FAQ and ask them to follow the instructions there?
On Thu, 15 Jul 2004, Duncan Murdoch wrote:
> (Moved from r-help)
> On Thu, 15 Jul 2004 19:41:54 -0400, "Liaw, Andy" <andy_liaw at merck.com>
> >The XEmacs and cygwin netinstaller has three install options: "Use IE5
> >settings", "Direct connection" (or something like that) and "Install from
> >local disk". Option #2 never work for me, but #1 does.
> Yes, I've seen that, but I never knew what it meant, and so it didn't
> give me confidence in the installation. (Did I make the right choice?
> How would I know if I didn't?)
> >I suspect the R
> >installer can do something similar: offer the two options (with and
> >without), and somehow test if the connection (to CRAN) works. If not,
> >prompt the user to try the other option.
> We don't have a netinstaller, so a connection to CRAN is not
> necessary. I would guess that a large number of people installing R
> don't have a net connection at the time, because a lot of students get
> it on a CD.
> > My take is that this option should
> >be (semi?) permanent: If it's needed, it's needed all the time for that
> It seems to me that if internet2 works at all, it should always be
> used, whether necessary or not. (It might be that your machine is
> behind a firewall sometimes, but not always: but if you've got the
> required library installed, you probably won't lose it.)
> Maybe internet2 should be the default now. I'd guess nowadays most
> Windows systems would support it.
> Does anyone know if there are any disadvantages to using the WinINet
> functions via internet2.dll instead of the WinSock functions via
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
More information about the R-devel