[Rd] Installing R-2.2.0 package

Prof Brian Ripley ripley at stats.ox.ac.uk
Thu Oct 13 21:47:14 CEST 2005


On Thu, 13 Oct 2005, Dean, David P wrote:

>> From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
>> Sent: Thursday, October 13, 2005 12:21 PM
>> To: Dean, David P
>> Cc: Uwe Ligges; r-devel at r-project.org; Paul Gilbert
>> Subject: Re: [Rd] Installing R-2.2.0 package
>>
>> On Thu, 13 Oct 2005, Dean, David P wrote:
>>
>>>> From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
>>>> Sent: Thursday, October 13, 2005 11:42 AM
>>>> To: Uwe Ligges
>>>> Cc: Paul Gilbert; r-devel at r-project.org; Dean, David P
>>>> Subject: Re: [Rd] Installing R-2.2.0 package
>>>>
>>>> It works on Solaris 8 for me, and this is checked as part of the
>>>> alpha/beta process.
>>>>
>>>> The code is the same on Linux and Solaris, after all, so this would
>> have
>>>> to be a Solaris shell bug.
>>>
>>> I'm on Solaris 8 as well. R is compiled for 64-bits, but I wouldn't
>> expect
>>> that to be a factor.
>>
>> I tested this on a 64-bit setup, as it happens.
>>
>>> R was built and is run in an environment created with
>>> OpenPKG and containing GNU equivalents to many utilities -- so it's not
>> a
>>> stock Solaris setup.
>>
>> I hope /bin/sh is Solaris and not bash.
>>
>>> I'm not familiar with R internals but will be happy to try to debug if I
>>> can. Does the installer call out to the shell somehow? Where would I
>> start
>>> to look for the relevant code?
>>
>> INSTALL _is_ a shell script.
>>
>> Rather than mess about with past code, can you first please try the
>> current R-patched (see the FAQ) as I will shortly not have R-2.2.0 on
>> Solaris but rather R-patched (I already have R-devel in 4 flavours).
>>
>>>
>>> Thanks!
>>> David Dean
>
> OK, I'll try the current patched. I did locate the INSTALL script and added
> a -x flag to see what is going on. (This is in the stock 2.2.0):
>
> + do_exit_on_error
> remove_R_package_dir=yes
> + test -z
> /net/gsun374/app/openpkg/lib/R-2.2.0-64bit/lib/R/bin/INSTALL: test: argument
> expected
>
> I think this line in do_exit_on_error causes an abend if bundlepkg isn't
> set:
>
> If test -z ${bundlepkg}; do

It is in R-patched, and indeed it has quotes too.


-- 
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 mailing list