[R] Upgrade process for libraries: can I use installed.packages on an old installation followed by install.packages in a new one

[Gabor Grothendieck]

Correction

On Thu, Apr 29, 2010 at 6:28 PM, Gabor Grothendieck
<ggrothendieck at gmail.com> wrote:
> The batchfiles distribution has a copydir.bat which copies files from
> one directory to another without overwriting anything and a
> movedir.bat which is similar but moves the files rather than copying
> them (which is much faster but you wont have the packages in your old
> installation any more).
>
> Get the batchfiles distribution from http://batchfiles.googlecode.com
> (this is not an R package) and from it get either copydir.bat or
> movedir.bat depending on whether you want to copy or move your
> packages from one library to another.  Either is a self contained
> batchfile so just put it anywhere in your PATH.  Neither of these will
> overwrite anything but can be used to copy or move the ones that do
> not already exist in the target library.  The example below assumes
> you use the R defaults for win-library:
>
>  :: first, download and install R-2.11.0
>  :: now perform the next two commands at the Windows console
>
>  cd %userprofile%\Documents\R\win-library
>  :: you can use movedir instead of copydir if you prefer to move them
>  copydir R-2.10.1\library R-2.11.0\library

Last line should be (assuming the defaults from R):

copydir 2.10\library 2.11\library

>
>  :: ... now startup new version of R (R 2.11.0 in this example)
>  # issue this command in R 2.11.0
>  update.packages(checkBuilt = TRUE, ask = "graphics")
>
> On Thu, Apr 29, 2010 at 6:12 PM, Ted Byers <r.ted.byers at gmail.com> wrote:
>> When doing a fresh install of a new version of R, using update.packages()
>> requires copying some of the contents of the library subdirectory to the new
>> installation.  While possible and viable, it can be problematic in being
>> tedious (more an irritation regarding how Windows handles copying
>> directories from one location to another when there are already things in
>> the target directory with the same names, than anything else), and there
>> exists the possibility that there are some old packages that are obsolete
>> and won't work properly in the new version.
>>
>> I don't suppose update.packages() will remove obsolete packages in the
>> library directory if it finds them, does it?
>>
>> I have a preference for trying to do a fresh install of a given product's
>> optional packages (so if a given package has a problem in the new version,
>> it just doesn't install - rather than cluttering its directory tree with
>> useless stuff); something that is trivially easy if looking at only a
>> handful of optional packages but very tedious when there are so many.  I
>> know from experience that repeatedly having a large, complex piece of
>> software, whether a major application (like R or MS Word, &c.) or an OS like
>> Windows, update/over-write key part of itself will eventually lead to hard
>> to diagnose problems.  It is often good to have more than one way to
>> accomplish a given task, and there are usually many options to choose from
>> when designing/implementing software.
>>
>> Actually, with the benefit of 20/20 hindsight, if I had been asked to write
>> a update.packages() function, I would have had it look in the registery on
>> Windows, or in the directory tree, for evidence of an older version of R
>> (perhaps a version that is used only during a fresh install of R), and have
>> it process the list of detected packages and install/upgrade any packages
>> that will work with the new version of R, and perhaps, if a given obsolete
>> package has been superceded by something else, make sure that 'something
>> else' is installed instead, just so the directory tree for the new install
>> is not cluttered with old, potentially broken, stuff.
>>
>> Thanks
>>
>> Ted
>>
>> On Thu, Apr 29, 2010 at 4:59 PM, Erik Iverson <eriki at ccbr.umn.edu> wrote:
>>
>>>
>>>
>>> Ted Byers wrote:
>>>
>>>> I tend to have a lot of packages installed, in part because of a wide
>>>> diversity of interests and a disposition of examining different ways to
>>>> accomplish a given task.
>>>>
>>>> I am looking for a better way to upgrade all my packages when I upgrade
>>>> the
>>>> version of R that I am running.
>>>>
>>>> On looking at support for installing and updating packages, I found these
>>>> two: installed.packages() and  install.packages() and it occurred to me
>>>> that
>>>> in principle I ought to be able to use the one in the original
>>>> installation
>>>> to get a list of packages I'm working with and and put its output into a
>>>> plain text file that I can read in the new installation and pass to the
>>>> other to ensure the new installation has a fresh installation of all the
>>>> packages I want to work with.
>>>>
>>>
>>> I must be missing the obvious, but what's wrong with update.packages() ?
>>>
>>
>>
>>
>> --
>> R.E.(Ted) Byers, Ph.D.,Ed.D.
>> TED at MERCHANTSERVICECORP.COM
>> CTO
>> Merchant Services Corp.
>> 350 Harry Walker Parkway North, Suite 8
>> Newmarket, Ontario
>> L3Y 8L3
>>
>>        [[alternative HTML version deleted]]
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>