R help - Jul 1998 - No subject

Douglas Bates wrote:
>From 0.62 onward you should not have to create a symbolic link in
>/usr/local/bin.  It should be that you can run
> cd $RSOURCE
> ./configure --prefix=/usr/local
> make install
>and you will end up with the R script installed in /usr/local/bin and
>all the files needed to run R in /usr/local/lib/R.
>
>Can you tell what the prefix is set to after you run ./configure?
>
>BTW, the reason for these changes is to bring R more closely in line
>with the coding standards for GNU.  It is very helpful to be able to
>have the sources in one directory but make the installation in another
>directory, especially in a multi-architecture environment.
That is what I figured, which is why I did the make install, as described
in my earlier message,  but to no avail. According to the Makeconf file my
prefix is /usr/local. So now I have the actual R script in /usr/local/bin
and various bits and pieces in /usr/local/man/man1 and the R tree in
/usr/local/lib//R.

Peter Dalgaard wrote:
>Some thing is clearly going wrong during the reading of the init code.
>The question is "where?". It *is* a bit of a design problem in R
that
>that isn't readily seen.
>
>It could be in the reading of $RHOME/library/base/R/base which you
>might want to check looks reasonable. You could also try to run
>(from the top level of the R source tree) something like
>
>R> source('library/base/R/base',echo=T)
I have kept the 0.61.1 directory. Comparing /usr/local/lib/R/library/base/R
in 0.62.1
with the 0.61.1 equivalent, the Rprofile and base files seem pretty much
the same (though not exactly the same, obviously)


source('/usr/local/lib/R/library/base/R/base',echo=T) gives
Error: memory exhausted.



>
>which should rerun the init code, but with echoing turned on, so that
>if it stops prematurely, you'll know why.
>
>It is somewhat unlikely that the base file should be the culprit,
>though, because it only defines a bunch of functions without actually
>executing anything except "<-".
>
>Perhaps something in the system Rprofile is doing you in. Again, try
>sourcing it with echo on. In particular, might there be a problem with
>getenv() on your system?
source( '/usr/local/lib/R/library/base/R/Rprofile',echo=T) also gives
Error: memory exhausted.

This is quite good, because Rprofile is short enough to copy  in line by
line.

Everything seems ok until
.Library <- paste(getenv("RHOME"), "library", sep=
"/")
Error: memory exhausted

getenv("RHOME") gives "/usr/local/lib/R",  so getenv
does seem to be working.

The problem is with paste:
paste("the cat","sat on the mat")
Error, memory exhausted
>
>Just to be sure, also check for stray .RData and .Rprofile files. If
>the former exists and is too large, that would explain it all by
>itself...
There are no .RData and .Rprofile files in the directory I have been using.

For additional information:
gc() gives
Garbage collection ...
123580 cons cells free (61%)
1738k bytes of heap free (89%)

I am very grateful for your help and advice. Do you have any more
suggestions? Do you think there is a simple patch for paste?

All the best,
Mark



Mark A. Beaumont
Institute of Zoology
Zoological Society of London
Regent's Park
London NW1 4RY UK
Tel:  0171 449 6617
Fax: 0171 586 2870
m.beaumont at ucl.ac.uk


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Peter Dalgaard wrote:
>"Mark A. Beaumont" <M.Beaumont at ucl.ac.uk> writes:
>
>
>> The problem is with paste:
>> paste("the cat","sat on the mat")
>> Error, memory exhausted
>
>
>Ok. Let's first verify that the problem really is in paste: Try
>temporarily removing the expression that contains it from RProfile.
>
I have replaced the offending paste line in Rprofile with an assignment
to the appropriate string.

I then get the first prompt without a "memory exhausted" error, so it
does
look as though this use of paste was causing the initial problem.
>Next, let's first see if paste itself looks sane:
>
>> paste
>function (..., sep = " ", collapse = NULL)
>{
>            args <- list(...)
>            if (is.null(args))
>                    ""
>            else {
>                    for (i in 1:length(args)) args[[i]] <-
>as.character(args[[i]])
>                    .Internal(paste(args, sep, collapse))
>            }
>}
>
it looks fine
>if it does (well, except for the fact that the for loop could have
>been an lapply() call...), then try
>
>debug(paste)
>paste("the cat","sat on the mat")
This produces

	> debug(paste)
	> paste("the cat","sat on the mat")
	debugging in: paste("the cat", "sat on the mat")
	debug: args <- list(...)
	Browse[1]>
	debug: if (is.null(args)) "" else {
            	for (i in 1:length(args)) args[[i]] <- as.character(args[[i]])
            	.Internal(paste(args, sep, collapse))
	}
	Browse[1]>
	debug: for (i in 1:length(args)) args[[i]] <- as.character(args[[i]])
	Browse[1]>
	Error: memory exhausted

Following on from Albrecht Gephardt's message, I find
1:3 gives "error: memory exhausted", similarly with seq(1,5).

Thus, it would appear that the problem lies somewhere in here. Though why
it should manifest itself in two different ways on our two systems is
rather mysterious. Could there be an obscure long/int problem somewhere?
People tend to forget that they have different lengths on an alpha.

>
>> There are no .RData and .Rprofile files in the directory I have been
using.
>
>And no .Rprofile in your home dir either, I take it?
Thats right.


All the best,
Mark

Mark A. Beaumont
Institute of Zoology
Zoological Society of London
Regent's Park
London NW1 4RY UK
Tel:  0171 449 6617
Fax: 0171 586 2870
m.beaumont at ucl.ac.uk


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._