[RsR] estimators based on random samples... - should be random
Martin Maechler
m@ech|er @end|ng |rom @t@t@m@th@ethz@ch
Wed May 10 11:21:10 CEST 2006
>>>>> "ValenT" == Valentin Todorov <valentin.todorov using chello.at>
>>>>> on Wed, 10 May 2006 09:46:58 +0200 writes:
>> C++ complicates things slightly
ValenT> you mean under R, not in general?
yes. It slightly complicates things for interfacing to any S dialect,
including R and S-plus.
>> and typically is a tad (10% ?) slower than C.
ValenT> and assembler is even faster!
:-)
>> R's C API is a C and not C++ API anyway.
>> I think it's only worth using C++ for some kind of problems
>> (e.g. when a class structure can make the program much more readable),
ValenT> and this is almost always the case - a class structure makes a program
ValenT> always more readable
I'm not going to debate further here; of course what
"readable" means exactly depends very much on the reader,
and in particular on what kind of reading one is used to...
>> and I doubt that the covMcd() or ltsReg() algorithms belong to
>> this class.
ValenT> not until they are translated 1-1 from Fortran. I doubt also that covMcd/Lts
ValenT> are readable at all now and I expect the readability to decrease (with much
ValenT> more than 10%) when converted authomatically to C.
I agree that readability would decrease slightly *) --
but if it's "not readable at all now", i.e., 0% readable,
it cannot get much lower by a decrease :-) :-)
Martin
*) for that reason I had developed my perl script to
post-process the result; and I usually also invest a few hours
of manual postprocessing such that in the end, I believe that
readability would be about the same.
More information about the R-SIG-Robust
mailing list