2008/6/25 laurent oget <[EMAIL PROTECTED]>:
> Wise Rpy fellows,
>
> I just came back to using R after a few years and am happy to see that the
> RPy project is alive and well, and hope I will be able to provide some help
> in the future.
>
> For now I have a few questions about rpy2:
> -is there a document describing the motivation behind the rewrite and the
> difference of architecture?
There is no document. There were off-list emails, donated code, the
need for a major code clean up (a lot of workarounds no longer
necessary with more recent versions of R and Python), the same issues
coming up over and over again on the mailing-list.
In few bullet points, the rewrite is trying to:
a- give a low-level interface (close to R's C interface)
b- give a higher level interface built on a-
(the doc already mentions a- and b- in the Overview).
c- type a little more R objects (like have only R function-alikes
callable from Python, etc...)
(there are lot of pending issues, good reasons for some of the choices
being made, but probably not for other choices... but the most
complete technical documentation for both rpy and rpy2 remains the
source).
The development methodology adopted is probably "cowboy-coding", with
a keen eye for unit tests.
The "early alpha" is a signal that some development is going on, that
it is somewhat working and can be tried out, and inspected, and
reviewed.
> I am pretty excited to read there will no longer
> be conversion modes in rpy2 but the section 4.7 of the documentation is not
> giving me much light on what replaces the conversion modes
The doc is (still) incomplete.
The source code is the primary source of information (and it will
reveal that not everything is
ironed out yet).
> -what is the compelling reason for forcing R>2.7? Is there a significant
> change in the C API between 2.6.2 and 2.7 ?
R's C-level function "do_slot" only exists since R-2.7.0 and is used
in rpy2.rinterface.
>
> I also have a more pedestrian RPY question:
> if m is a matrix in R, how would i get m[,2] in python?
That's a good question.
In rpy it *should* be (I have not tried it):
rpy.r.get("[")(m, True, 2)
In rpy2:
What is in section 4.4.2 (svn revision 563) applies to all R objects
for which the function "[" is defined.
>>> import rpy2
>>> import rpy2.robjects as ro
>>> m = ro.r.matrix(range(10), 5, 2)
>>> print(m)
[,1] [,2]
[1,] 0 5
[2,] 1 6
[3,] 2 7
[4,] 3 8
[5,] 4 9
>>> m.subset(True, 2) # Note the use of R's recycling rule instead of "missing"
>>> parameters
list(5L, 6L, 7L, 8L, 9L)
>>>
I have contemplated the idea of having the method 'subset' defined above
in place of the python method '__getitem__'... but I ultimately found
it error-prone.
Features like R indexes starting at one, and Python indexes starting
at zero for example
made me prefer two separate method to access items: the Python one (that behaves
the way a Python programmer is used to), and the R one.
My current dissatisfaction is with the name 'subset' (as it reminds of
the R-base function
subset, and it is annoyingly longer that the "[" operator __getitem__
benefits from).
L.
> Thanks,
>
> Laurent
>
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> _______________________________________________
> rpy-list mailing list
> rpy-list@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rpy-list
>
>
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
rpy-list mailing list
rpy-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rpy-list
