Hi,
The RMath plug-in has a fairly efficient routine for marshalling
AFL
vector data to SafeArrays in COM. This happens both on the way in
and back out of RMath and so is hidden from the end user, making the
code required very compact and simple.
Of course, it may be
possible to write code in AFL to do this
manually (as pointed out in the
manual), and it may be that this can
be improved further by having a play
around with the VBArray function
in AFL to assist with this (if you don't
want to use the plug in).
Re potential upgrade, again this has not been
decided yet*, but
basically yes that is the idea.
PS
* If
only because our intitial goal (i.e. access to R WITHIN AB) has
been
achieved, and this step would complicate matters in some
aspects, like
require the user to know how to operate R in its native
environment.
--- In amibroker@xxxxxxxxxps.com,
"loveyourenemynow"
<loveyourenemynow@...>
wrote:
>
> Hello,
>
> and thank you for your
answer.
> I refer to the fact that in order to retrieve an array from R,
this
> has to be done element by element using the com server, since AB
is
> not accepting the R data type.
> I wrote some AFL functions
to do this, but I was wondering if the
> plugin was somehow able to
actually transfer data directly (which
> would be much faster).
>
But if it using the com server I guess this is the only way.
>
>
Regarding the new implementation,do u mean that some AFL variable
is
> constantly update with its R "sister", without the need to
tranfer
> data back and forth?
>
> Thanks
>
Ly
>
>
>
> --- In amibroker@xxxxxxxxxps.com,
"vlanschot" <vlanschot@> wrote:
> >
> > Hi Ly,
>
>
> > First, which "problems" do you mean (or do you refer to
endnote
II in
> > the RMath manual)?
> >
> >
To answer your question: yes it does. And in terms of
transferring
> > data (e.g. vectors) to R via the plug-in, I have not encountered
any
> > limitation. In other words, I personally have no need
for
any "work-
> > around".
> >
> > However,
the current plug-in does not (yet) allow full two-way
> >
communication between R and AB, i.e. putting something in R which
>
> immediately becomes available in AB. We are looking (early stage)
at
> > possibly integrating another tool which would allow you to move
back
> > and forth like that between R and AB which, for
example, would
bypass
> > the need for the debugger as you would
see your results
immediately
> > within R.
> >
>
> Hope this answers your question, if not you'll need to be more
>
> specific I'm afraid.
> >
> > PS
> >
>
> --- In amibroker@xxxxxxxxxps.com,
"loveyourenemynow"
> > <loveyourenemynow@> wrote:
>
> >
> > > Hello Patrick,
> > >
> >
> I would like to ask you if your plugin uses the R com server to
>
> > transfer data between R and AB.
> > > I was doing that
and wrote some simple functions to transfer
> > vectors,
>
> > and I encountered the problems you mention in your
documentation
> > file,
> > > but with some workaround it was
working.
> > >
> > > I wonder if you are able to
directly transfer vectors or you
use
> > some
> > >
similar workaround.
> > >
> > >
> > >
Thanks
> > >
> > > Ly
> > > --- In amibroker@xxxxxxxxxps.com,
"vlanschot" <vlanschot@>
wrote:
> > > >
> >
> > Hello,
> > > >
> > > > This mail is
to inform you that I have just uploaded the zip-
file
> > >
> RPlugIn. It contains two files. The first is the file
> >
RMathAFL.dll,
> > > > the RMath plug-in for AB. The second is
the Word-document
R_Plug-
> > > > in_Amibroker.doc,
the accompanying manual. The manual briefly
> > > > describes
the functionality of the RMath plug-in for
Amibroker
> > which
> > > > I make freely available to all AB-users. Its purpose
is to
allow
> > you,
> > > > the AB-user, to
efficiently interact with R, the open-source
and
> > > >
freeware statistical/ mathematical package based on the S-
> >
language
> > > > (i.e. S-Plus). For more details on R , please
visit the
official
> > > > website: http://www.r-project.org/.
Here you will also find
> > manuals
> > > > and
other contributed papers on R. Specifically, see this web-
> > page:
> > > > http://www.stats.bris.ac.uk/R/.
>
> > >
> > > > It is impossible to discuss the full
scale of the R-
functionality
> > > > that the plug-in
makes available to the AB-user. It simply is
> > > > massive,
and even I have only explored a tiny bit of it.
Instead,
> > the
> > > > manual will describe a small subjective selection of
the
> > > > possibilities available, with the aim to get you
going. I
have no
> > > > pretension that my examples are
of any use to you. What I do
> > hope,
> > > >
however, is that they can get you going, and that this plug-
in
>
> will
> > > > enhance your trading/analysis skills, like
it has mine.
> > > >
> > > > The plug-in has
been kindly developed by a programmer-friend
for
> > > >
which I am very grateful (and no, it is not Tomasz). There
will
>
> be NO
> > > > technical support for it, and (in all
likelihood) no
upgrades, if
> > > > only because the
functionality embedded is:
> > > > - very extensive and
flexible in terms of accessing R's
core
> > > >
functionality (I would say almost unlimited, particularly if
you
>
> > > write your own R-functions separately in R, which
subsequently
> > can
> > > > then be called from
AFL);
> > > > - unlikely to be (negatively) impacted by changes
in R
itself;
> > > >
> > > > In the manual
you will read about my motivations to have this
> > plug-in
>
> > > developed, the main one being to thank, first and foremost,
> > Tomasz
> > > > Janeczko for developing AB into
what it is has become: an
> > industry-
> > > >
standard investment and trading platform. It has enabled me
to
>
> > > transform my investment thinking into practical applications,
> > making
> > > > some money along the way.
Extended thanks go to the AB-
community
> > in
> > >
> general. I am very grateful to the guidance, sample code, and
>
> other
> > > > help (including support from Marcin) I have
received over the
> > past
> > > > few
years.
> > > >
> > > > PS
> > >
>
> > >
> >
>