0

I am trying to use the embedded R feature in Windows 7. I start the server with 

mserver5 --set embedded_r=true

I do not see the message # MonetDB/R module loaded though.

Then I wanted to test the example rapi00 provided here: 

CREATE FUNCTION rapi00() RETURNS TABLE (d INTEGER) LANGUAGE R {
    seq(1,10);
};

runs well. However, when I execute 

SELECT d FROM rapi00() AS r WHERE d > 5;

I get the following error message 

TypeException:user.s10_1[2]:'rapi.eval' undefined in: _4:bat[:oid,:int] := rapi.
eval(_5:ptr, _6:str)
program contains errors

I wonder if the this feature is implemented at all in Windows (it runs well in mac os). Can someone please help?

teucer
  • 6,060
  • 2
  • 26
  • 36
  • The Installation section says you need to compile monetdb with certain flags: `./configure --enable-rintegration=yes`. And if successful you get a message and if not, then "check that the R binary is in your $PATH, and that R was configured with the --enable-R-shlib=yes flag. You may also be required to set the $LD_LIBRARY_PATH variable to include the path to libr.so before starting MonetDB." – IRTFM Jun 25 '15 at 22:54
  • @BondedDust I am aware of this. However, when I installed on a mac I did not have to compile. My understanding is that this is now integrated in monetdb. – teucer Jun 26 '15 at 06:48

1 Answers1

2

Embedded R for MonetDB is not available on Windows. I do not plan to ever make it available there. In my opinion, running a MonetDB server on Windows does not make too much sense because of performance limitations on that platform.

What we do have for Windows now is MonetDBLite for R, where we run MonetDB inside R.

Hannes Mühleisen
  • 2,542
  • 11
  • 13
  • Hannes, I cannot tell how enthusiastic I am about this possibility. I am happy to help! I think this, potentially combined with "zero copy integration", would be a killer feature. So far my main frustration with R has always been its big data limitations. Now I can see a clean solution that would solve that once for all. PS: I also think that this would be the way to go forward for Julia's DataFrame package. – teucer Jun 26 '15 at 14:54
  • @teucer hannes is working his tail off to get `MonetDBLite` stable. subscribe to his twitter or, if you to help, contribute to his work- https://github.com/hannesmuehleisen?tab=activity – Anthony Damico Dec 14 '15 at 13:30