[R-sig-Debian] "Graphics history" in UNIX --> "R" in Ubuntu for students in applied sciences

[Scotti Roberto]

Many thank to everyone.
I really appreciate your kindness in making the point (or your point) on 
the topics that come up.
I am afraid that we completely abandoned the topic in the original 
subject line and shifted to a topic that is of great importance for me: 
<"R" in Ubuntu for students in applied sciences>.

For me the root question is related to Paul Johnson opinion.
 > 09/07/2010 00:36, Paul Johnson:
 > > "Personally, my opinion is if you want somebody else to do your 
work for you, you should have stayed with Windows." <<
To some extent it is necessarily true: choosing OpenSource software you 
take up the burden of taking care for yourself, if you prefer hanging 
over that burden to somebody else, you'd better pay him for the service!
On the other hand, I suppose we can agree that a tool for applied 
statistics as "R", is really useful if a non-statics-expert like a 
forestry student can use it having only basic knowledge of computers and 
informatics (and, obviously, something more than just basics in statistics).

What is the ideal compromise between these bounds?

A -        On 10 July 2010 at 11:54, Ross Boylan:
"The point of most distributions is to provide pre-compiled binary 
packages; particularly on Debian these are very good about indicating 
what other packages they require.  If you use a tool like aptitude and 
tell it you want package X, it will automatically pull in all other 
required packages.  Ubuntu is a Debian derivative."

B -        Date: Fri, 09 Jul 2010 18:37:38 +0100, Michael Dewey:
"My experience as a very naive Linux user with mostly Windows experience 
is that once you have the initial setup right subsequent installation of 
packages from source works very smoothly. I was quite nervous about the 
initial setup but the instructions on the welcome pages on CRAN were 
very helpful. I think when you look back in a few months time you will, 
like me, wonder why you were concerned in the first place."

I am very tempted to vote (B) but I am aware of my personal bias in 
favor of very general (though risky!) solutions.
I imagine that a common understanding on this point can help focus  
R-for-debian development.
Sincerely, Roberto.


Il 10/07/2010 21:53, Dirk Eddelbuettel ha scritto:
> On 10 July 2010 at 14:32, Dirk Eddelbuettel wrote:
> |
> | On 10 July 2010 at 11:54, Ross Boylan wrote:
> | | On Thu, 2010-07-08 at 16:41 +0200, Scotti Roberto wrote:
> | |>  stimulates a very basic question: "Is it possible to avoid building
> | |>  from
> | |>  source using R in Kubuntu?
> | |>  I shifted over from MS only recently. In Windows I used to easily
> | |>  download packages and use them.
> | |>  Now, every time I downloaded a new piece, I had the impression that
> | |>  building from source was necessary.
> | |>  Am I wrong?
> | |>  Many thanks Roberto
> | | Generally, building from source is not necessary.  I think the other
> | | responses agreeing that it is necessary were assuming you were
> | | interested in some slightly exotic R package (which may be the case).
> | |
> | | The point of most distributions is to provide pre-compiled binary
> | | packages; particularly on Debian these are very good about indicating
> | | what other packages they require.  If you use a tool like aptitude and
> | | tell it you want package X, it will automatically pull in all other
> | | required packages.  Ubuntu is a Debian derivative.
> | |
> | | Both R and many of its packages are in Debian.
> | |
> | | If you have a package that is not already built you can build it
> | | yourself or use cran2deb (assuming it's on CRAN).  As others have noted,
> | | cran2deb is only intended for Debian testing (aka squeeze) on some
> | | architectures.  It's also currently a bit broken.  You can try using the
>
> It is not broken; it is disabled for amd64 while we rewrite it. i386 works
> better than ever.  And this has been discussed on this list before.
>
> Dirk
>
>
>
> | | archive in another environment (e.g., Ubuntu), but if you want something
> | | that is not a gamble for R, running Debian testing (maybe in a VM) would
> | | be the safest route.  Of course, it's called testing rather than stable
> | | for a reason.
> | |
> | | I think the cran2deb packages, which are generated automatically, are
> | | less likely to have complete dependency information than those in the
> | | main distribution.
> |
> | Huh?  If the info wasn't complete, they'd break. They don't break, so the
> | info appears to complete. Please don't spread misinformation.
> |
> | D.
> |
> | | Finally, you expressed concern for your students.  You can build a
> | | binary package locally and then your students can use it.  Doing so for
> | | someone new to Linux is probably a heavy lift.
> | |
> | | Ross Boylan
> | |
> | | _______________________________________________
> | | R-SIG-Debian mailing list
> | | R-SIG-Debian at r-project.org
> | | https://stat.ethz.ch/mailman/listinfo/r-sig-debian
> |
> | --
> |   Regards, Dirk
> |
> | _______________________________________________
> | R-SIG-Debian mailing list
> | R-SIG-Debian at r-project.org
> | https://stat.ethz.ch/mailman/listinfo/r-sig-debian
>
>