[R-sig-hpc] Rmpi long vector support

[Simon Urbanek]

On Aug 8, 2013, at 1:44 PM, Jim Gattiker <j.gattiker at gmail.com> wrote:

> Aha, thanks again.  
> 
> Two notes:
> 1) serialize() didn't work over 2^31 until 3.0.1 (so it failed in my 3.0.0, with a message that led me to believe that there was still a 32-bit limit on length).

Ok, thanks, I was checking with 3.0.1 so I didn't notice.


> 2) more puzzling: Using vectors over 2^31 (2^31+2 ?!) seems to be broken in Rstudio, even though it works in R command-line. It looks like the functions that monitor the workspace are not ready yet.
> 
> Anyway, the important thing is, this message:
>     "Error: long vectors not supported yet: ../include/Rinlinefuns.h:100"
> is from Rstudio, not R. 
> 

That's possible - Rstudio overrides R functions which has caused problems before. If in doubt, test with R.

Cheers,
Simon




>       --j
> 
> 
> 
> On Thu, Aug 8, 2013 at 10:38 AM, Simon Urbanek <simon.urbanek at r-project.org> wrote:
> 
> On Aug 8, 2013, at 10:58 AM, Jim Gattiker <j.gattiker at gmail.com> wrote:
> 
> > Simon, I don't understand this either: It seems that R is commonly indexed
> > by 32-bit pointers, and mine certainly is on my Mac.
> 
> Nope, since R 3.0.0 we use 64-bit binaries on Mac OS X:
> 
> $ R
> R version 3.0.1 Patched (2013-07-15 r63328) -- "Good Sport"
> Copyright (C) 2013 The R Foundation for Statistical Computing
> Platform: x86_64-apple-darwin10.8.0 (64-bit)
> [...]
> > .Machine$sizeof.pointer
> [1] 8
> 
> 
> > I'll get the message
> > above also, as did the OP; that was the direct problem. How do I get "52
> > bit" R?
> >
> 
> http://cran.r-project.org/bin/macosx/
> 
> Cheers,
> Simon
> 
> 
> >    --j
> >
> > On Wed, Aug 7, 2013 at 7:42 PM, Hao Yu <hyu at stats.uwo.ca> wrote:
> >
> >> ....
> >> However, on a win7 64 with R 3.0.1 (16GB ram), I got
> >>> n=6e8
> >>> x=integer(n)
> >>> a=serialize(x,NULL)
> >> Error: long vectors not supported yet: ../include/Rinlinefuns.h:100
> >>
> >
> >       [[alternative HTML version deleted]]
> >
> > _______________________________________________
> > R-sig-hpc mailing list
> > R-sig-hpc at r-project.org
> > https://stat.ethz.ch/mailman/listinfo/r-sig-hpc
> >
> >
> 
>