[datatable-help] Long vectors in R

Steve Lianoglou mailinglist.honeypot at gmail.com
Wed Apr 18 19:55:30 CEST 2012


Hi,

On Wed, Apr 18, 2012 at 10:22 AM, Matthew Dowle <mdowle at mdowle.plus.com> wrote:
>
> Oh. How can you tell that (that it's for 2.16 and a year away)?

I thought I had read it in one of the R lists I was on, but I can't
seem to find where that might have been now, so don't take my word on
that.

The other source of evidence, though, is that you see comments in the
svn log for R-trunk about long vector support, but not in
R-2-15-branch ... which is just circumstantial evidence, to be sure.

So, take my comment on long vector support only landing in 2.16 is
just a guess/bet and not a statement of fact.

-steve



>
> For completeness, it'll be a bigger difference (and for more people) using
> matrix (and objects using matrix) than data.table I'd guess? Since a
> matrix is really one long vector, a single matrix can then be greater than
> 16Gb (currently n*m < 2^31-1, iiuc). In data.table (and data.frame) each
> column can already up to 16Gb, and the data.table/frame as large as RAM.
> The 2 billion row limit can still bite though and the change is very
> welcome.
>
> Matthew
>
>
>> Hi,
>>
>> Yeah, long vectors are awesome and a welcomed (future) addition to R.
>> For whatever it's worth, I'm pretty sure that this (R-devel) is for
>> R-2.16.x (and not 2.15.1), which is still a year away.
>>
>> -steve
>>
>> On Wed, Apr 18, 2012 at 5:15 AM, Matthew Dowle <mdowle at mdowle.plus.com>
>> wrote:
>>>
>>> Dear datatablers,
>>>
>>> Looks like support for long vectors is being added to R :
>>>
>>> http://stat.ethz.ch/R-manual/R-devel/doc/html/NEWS.html
>>>
>>> Since data.table hooks into R's internals, it is already failing on
>>> r-devel CRAN checks (as Prof Ripley warning us a while back).
>>>
>>> So, I'll see what I can do, but advance warning that we might need to
>>> soon
>>> depend on R 2.15.1 (or whichever version of R long vectors makes it
>>> into).
>>> However, I'd say there was an 80% chance it'll be compile time switches
>>> and we can continue to depend on R 2.12.0 in future versions of
>>> data.table.
>>>
>>> Matthew
>>>
>>>
>>>
>>> _______________________________________________
>>> datatable-help mailing list
>>> datatable-help at lists.r-forge.r-project.org
>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help
>>
>>
>>
>> --
>> Steve Lianoglou
>> Graduate Student: Computational Systems Biology
>>  | Memorial Sloan-Kettering Cancer Center
>>  | Weill Medical College of Cornell University
>> Contact Info: http://cbio.mskcc.org/~lianos/contact
>>
>
>



-- 
Steve Lianoglou
Graduate Student: Computational Systems Biology
 | Memorial Sloan-Kettering Cancer Center
 | Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact


More information about the datatable-help mailing list