Finance::YahooQuote 0.24
Having espoused rule number one in regression testing in the post about yesterday's bug fix upload 0.23, we can now add rule number zero: Do not introduce a new error by omitting the trailing semicolon. I guess it shows that I don't really program in Perl anymore.

Anyway, a new version 0.24 of Finance::YahooQuote which addresses the issue that required upload 0.23 yesterday is now in the Debian queue and on CPAN and my local yahooquote page. This time it may even work. A big thanks to the CPAN Testers for getting me reports on this one too.

/computers/linux/debian/packages | permanent link

Finance::YahooQuote 0.23
Rule number one in regression testing is to not depend on volatile data. Which I seem to have violated in file t/02simple.t in the Perl package Finance::YahooQuote.

Which lead the automated Perl test scripts to remind me for a few days now that the full company name for symbol IBM no longer corresponded to what I had encoded. Not really a bug, but a failure in tests anyway.

So without further ado: a new version 0.23 of Finance::YahooQuote which addressed this issue is now in the Debian queue and on CPAN and my local yahooquote page.

/computers/linux/debian/packages | permanent link

Rcpp 0.7.3
A quick nine days after release 0.7.2 of Rcpp, our R / C++ interface classes, Romain and I are happy to roll out a new version 0.7.3. It has been uploaded to CRAN and Debian, and mirrors should have the new versions shortly. As before, my local page is also available for downloads and some more details.

This release combines a number of under-the-hood fixes and enhancements with one bug fix:

• The Rcpp:::LdFlags() helper function to dynamically provide linker options for packages using Rcpp now defaults to static linking on OS X as well. For installation from source dynamic linking always worked, but not for binary installation (as e.g. from CRAN). As on the other platforms, this default can be overridden. Thanks to the phylobase team for patient help in tracking this down.
• Accessing various types via [] should now be faster due to some enhancements in the internal representations.
• configure now has a command-line option (as well as an environment variable) to select support for the draft of the upcoming C++0x standard.
• A new function Rcpp.package.skeleton(), modelled after package.skeleton() in R itself, helps to set up a new package with support for using Rcpp.
• A number of other minor tweaks and improvements...

As always, full details are in the ChangeLog on the Rcpp page which also leads to the downloads, the browseable doxygen docs and zip files of doxygen output for the standard formats. Questions, comments etc should go to the rcpp-devel mailing list off the R-Forge page

/computers/linux/debian/packages | permanent link

RQuantLib 0.3.2 released
A new version of RQuantLib (a package combining the quantitative analytics of QuantLib with the R statistical computing environment and language) is now out at CRAN and in Debian (where it depends on the 1.0.0 beta of QuantLib that is currently in the NEW queue with its new library version). This RQuantLib release works with either the current release 0.9.9 as well as with the just-released first beta of QuantLib 1.0.0.

This versions brings a few cleanups due to minor Rcpp changes (in essence: we now define the macro R_NO_REMAP before including R's headers and this separate non-namespaced functions like error() or length() out into prefixed-versions Rf_error() and Rf_length() which is a good thing).

It also adds a number of calendaring and holiday utilities that Khanh just added: tests for weekend, holiday, endOfMonth as well dayCount, date advancement and year fraction functions commonly used in fixed income.

Full changelog details, examples and more details about this package are at my RQuantLib page.

/computers/linux/debian/packages | permanent link

Rcpp 0.7.2
Not even two weeks after the Rcpp 0.7.1 release, Romain and I have a new one to present: Rcpp 0.7.2. It has been uploaded to CRAN and Debian, and the respective package management systems should carry them around in the next few hours. As always, the local page is also available for download too.

A lot of the momentum for the new API is continuing, thanks in large part to Romain. A number of new classes have been added, and existing ones have been enhanced. There are more unit tests than ever, and more documentation. We have better build support (with g++ version detection so that we can add some C++0x support where available) and a new examples sub-directory.

We did take one toy away, though. The Doxygen-generated docs were getting so big that we decided to keep them out of the source tarball. (And arguably, they are also too volatile.) We still have the browseable html docs as well as the pdf version (now at over 300 pages!). And we added zip archives of the docs in html, latex, and man format for download.

As always, full details are in the ChangeLog on the Rcpp page. Questions, comments etc: bring them to the rcpp-devel mailing list off the R-Forge page

/computers/linux/debian/packages | permanent link

RInside release 0.2.1
The shiny new 0.2.1 release of RInside, a set of convenience classes to facilitate embedding of R inside of C++ applications, just went out to CRAN; sources are also at my RInside page

This is a maintenance release building on the recent 0.2.0 release which added Windows support (provided you use the Rtools toolchain for Windows). In this release, we changed the startup initialization so that interactive() comes out FALSE (just as we had done for littler just yesterday) and with that no longer call Rf_KillAllDevices() from the destructor as we may not have had devices in the first place. A few minor things were tweaked around the code organisation and build process, see the ChangeLog for details.

The new release should hit CRAN mirrors tomorrow, and is (as always) available from my machine too.

/computers/linux/debian/packages | permanent link

littler 0.1.3
A new littler release (now at 0.1.3) just went out of the door this evening.

littler provides r (pronounced littler), a shebang / scripting / quick eval / pipelining front-end to the the R language and system.

This version adds a few minor behind-the-scenes improvements:

• interactive() now evaluates to false as you'd expect in a non-interactive scripting front-end. To restore the previous behaviour, new switches -i or --interactive have been added.
• Some of the 'cleanup' functionality described in Section 8.1.2 on 'Setting R callbacks' from the R Extension manual have been adopted.
• Example scripts install.r and update.r received an update based on lessons learned from the R 2.10.0 roll-out and package rebuilding.
• A few build issues were improved, a minor manual page formatting bug was fixed.

As usual, our code is available via our svn archive or from tarballs off my littler page and the local directory here. A fresh package is in Debian's incoming queue and will hit mirrors shortly.

/computers/linux/debian/packages | permanent link

Rcpp 0.7.1
Two weeks after the Rcpp 0.7.0 release, Romain and I are happy to announce release 0.7.1 of Rcpp. It is currently in the incoming section of CRAN and has been accepted into Debian. Mirrors will catch up over the next few days, in the meantime the local page is available for download too.

A lot has changed under the hood since 0.7.0, and this is the first release that really reflects many of Romain's additions. Some of the changes are

• A new base class Rcpp::RObject that replaces RcppSexp (which is still provided for compatibility); it provides basic R object handling and other new classes derive from it.
• Rcpp::RObject has real simple wrappers for object creation and a SEXP operator for transfer back to R that make simple interfaces even easier.
• New classes Rcpp::Evaluator and Rcpp::Environment for expression evaluation and R environment access, respectively.
• A new class Rcpp::XPtr for external pointer access and management.
• Enhanced exception handling: exception can be trapped at the R even outside of try/catch blocks, see Romain's blog post for more.
• Namespace support with the addition of a Rcpp namespace; we will be incremental in phasing this in keeping compatibility with the old interface
• Unit test for most all of the above via use of the RUnit package, and several new examples.
• Inline support has been removed and replaced with a Depends: on inline (>= 0.3.4) as our patch is now part of the current inline package as mentioned <here.

/computers/linux/debian/packages | permanent link

inline 0.3.4 released
Oleg has updated the inline package to version 0.3.4 which is now on CRAN. It is includes my patch for both Rcpp support as well as extended header / library options for PKG_CPPFLAGS, PKG_CXXFLAGS, and PKG_LIBS which I had mentioned recently here and here. There is also a brand-new and shiny R-Forge project for it.

A new release of Rcpp will probably follow in due course as we can now depend on this new inline version; Romain has also put nice new code in around exceptions and we may add some more on external pointers. Good times ...

/computers/linux/debian/packages | permanent link

Beancounter minor bug fix release 0.8.9
Another small update to beancounter and just like the previous release due to changes in Perl's Date::Manip. It seems that Date_Calc now really wants a reference to store error codes, so we supply one. I also changed the FX code to use OandA ad there are often issues with Yahoo's FX data. The version 0.8.9 is now in Debian, at CPAN and on my beancounter page here.

/computers/linux/debian/packages | permanent link

RInside release 0.2.0
RInside, a set of convenience classes to facilitate embedding of R inside of C++ applications, was updated to release 0.2.0 over the weekend. Sources as well as binary packages are at CRAN; sources are also at my RInside page

The biggest news is that we now support builds on Windows -- if and only if you use the R toolchain -- more on that below. Since I had first released RInside people had emailed about support on Windows. However, Richard Holbrey deserves special mention as he actually sat down and tried to build it. Over a few email exchanges we made decent progress, but stalled. More recently, I sat down and had another go and lo and behold, it works. This required a few #ifdef statements here and there, as well as a function setenv() which Windows does not have -- Richard kindly helped here and borrowed most of the code from R itself.

Now, to make this plain: you need gcc (as this is how R itself is built). This gcc compiler is the native compiler on Linux and OS X. But on Windows this requires installing the Rtools compiled by Duncan Murdoch as detailed in the The Windows Toolset appendix to the R Installation manual. Do not, I repeat, do not report an error if you fail to build this with another compiler as this is not supported.

That said, if you have the proper tools, things are peachy. Install the package, change into the examples directory installed with it and say make (or on Windows: make -f Makefile.win). That's it -- now run the examples (which on Windows will most likely require setting the environment variable R_HOME). Now you can run simple examples like the one from the previous blog post or any of the other ones.

Happy embedding!

/computers/linux/debian/packages | permanent link

Rcpp and inline example
Following up on yesterday's Rcpp 0.7.0 release, here is a quick example of why the new feastures can be so useful.

The following R code defines a character variable gslrng. This variable contains a short C++ code segment, which is then transformed by the function cfunction into a function of two arguments assigned to funx:

    ## now use Rcpp to pass down a parameter for the seed, and a vector size
    gslrng <- '
       int seed = RcppSexp(s).asInt();
       int len = RcppSexp(n).asInt();

       gsl_rng *r;
       gsl_rng_env_setup();
       std::vector<double> v(len);

       r = gsl_rng_alloc (gsl_rng_default);

       gsl_rng_set (r, (unsigned long) seed);
       for (int i=0; i<len; i++) {
           v[i] = gsl_rng_get (r);
       }
       gsl_rng_free(r);

       return RcppSexp(v).asSexp();
    '

    ## turn into a function that R can call
    ## compileargs redundant on Debian/Ubuntu as gsl headers are found anyway
    funx <- cfunction(signature(s="numeric", n="numeric"),
                      gslrng,
                      includes="#include <gsl/gsl_rng.h>",
                      Rcpp=TRUE,
                      cppargs="-I/usr/include",
                      libargs="-lgsl -lgslcblas")
    print(funx(0, 5))

The signature argument to cfunction defines two variables s and n -- which the C++ function then reads in from R and converts to two integers seed and len. seed is used to initialize the random-number generator, and len draws are then taken and stored in the STL vector v which returned at the end.

As the R level, we now have a function of two arguments returning a vector of RNG draws of the given lenth and using the given seed.

Also note how we tell cfunction to add the GSL include line, specify that we want to compile and link against Rcpp and provide -I and -L arguments to compile and link with the GSL. (The include statement is not needed as the compiler would have found them in /usr/include anyway, but it shows how to set this if needed.)

Finally, we simply call our freshly compiled, linked and loaded C++ function with arguments zero for the seed and five for the length, and print the results vector returned to R from C++.

/computers/linux/debian/packages | permanent link

Rcpp 0.7.0
A new release of Rcpp awaits inclusion on CRAN and has also been sent as a .deb to Debian. It will hit mirrors over the next few days, in the meantime you can get it here as well.

This release has a couple new features :

• inline support: the cfunction from Oleg Sklyar's excellent inline package has been imported and adapted. This means simple C++ programs can be defined in an R character vector and passed to cfunction which will create a complete file that it then compiles, links and loads --- giving you access to compiled C++ code right from the R prompt without having to worry about compiler flags, linker options, ... Better still, we extended this to not only support Rcpp but any external library via addtional header / linker arguments that will be passed to R via the PKG_CPPFLAGS, PKG_CXXFLAGS and PKG_LIBS environment variables.
• this even works on Windoze (if you have the Rtools installed as detailed in the Windows Toolset appendix to the R Installation manual) in exactly the same way. And no folks, that still does NOT mean you can use Visual Whatever -- R really requires MinGW as the links in this parapgraph document very plainly. But if and when you have the tools, R's remarkable consistency across operating systems allows you to use Rcpp and inline in pretty much the same way.
• A handful of new examples for the inline support have been added.
• A new type RcppSexp for simple int, double or std::string scalars as well as vectors; this is particularly useful for the inline support.
• This also completes the source code reorginsation: every class now has its own header and implementation file
• Last but not least, the package has been relicensed from LGPL-2.1 (or later) to GPL 2 (or later).

/computers/linux/debian/packages | permanent link

RQuantLib 0.3.1 released
A new version of RQuantLib is now out at CRAN and in Debian. It updates the package to the 0.9.9 release of QuantLib and is mostly a maintenance release with a few minor fixes along the way.

Full changelog details, examples and more details about this package are at my RQuantLib page.

/computers/linux/debian/packages | permanent link

digest 0.4.2
digest version 0.4.2 is now on CRAN. This new version features a fix by Henrik Bengtsson who also joined the R-Forge-hosted digest project.

/computers/linux/debian/packages | permanent link

Rcpp 0.6.8
As memtioned when I announced release 0.6.7 of Rcpp two days ago, a new version 0.6.8 was due because of my oversight of also adapting Makevars.win (used on that other OS) to the new file layout.

So 0.6.8 went onto CRAN and into Debian earlier in the day. Beside the aforementioned fix, I also split off a more class headers and implementations into their own files but changed no actual functionality.

As always, more details are at the Rcpp page.

/computers/linux/debian/packages | permanent link

Rcpp 0.6.7
Version 0.6.7 of the Rcpp package---a set of C++ classes making it easier to glue C/C++ code to R---is now on CRAN and Debian. This is a fairly incremental release that contains just a few small changes that arose since the previous release in August.

One change is that a new type RcppList was added which allows us to build R 'list' types in C++. In particular, this makes it possible to have data structures of different types and dimensions. We had not previously accomodated this as our data structures where more regularised (i.e. think matrices or vectors of various types). The need for this came out of the work with the recently started RProtoBuf package which interfaces the Google ProtoBuf library. The incredible Romain François has joined the project and added a number of nice tricks; this will need another blogpost sooner rather than later. In the meantime, check out the RProtoBuf pages at R-Forge.

Another small change was the addition of a self-contained example of a function callback from C++ into R which wasn't all that well documented before; this was in response to a user request.

Finally, I started RcppList as a first step in reorganising the code a little better and split the class header and body off into separate files. This required changes to the infrastrucure files Makevars, cleanup, ... and I promptly forgot the Makevars.win Windoze variant which will require a 0.6.8 release real soon.

/computers/linux/debian/packages | permanent link

RPostgreSQL 0.1-6
A quick bug fix released of RPostgreSQL was uploaded to CRAN earlier today; it updates last week's 0.1-5 release. An editing error left one of the helper functions broken. I also updated a few of the test scripts to also allow for optional environment variable for the port at which the PostgreSQL server is listening. This is handy when you have parallel installations of different PostgreSQL versions which Debian makes really easy.

As always, all the details (including the full ChangeLog) are at the local RPostgreSQL page, the RPostgreSQL package page at CRAN and of course the Google Code repository for RPostgreSQL.

/computers/linux/debian/packages | permanent link

RPostgreSQL 0.1-5
A new version of RPostgreSQL just went out to CRAN and should propagate via the mirrors over the next few days.

This is a maintenance and bug fix release that addresses four of the seven issue at the Google Code page. Two more are really enhancement requests for which we encourage patches as we are unlikely to have to write them, and the last issue I have been unable to replicate.

This release has benefitted greatly from the generous help of Neil Tiffin who now looks after all things OS X and keeps a good eye on the Google Code site (where I had, ahem. overlooked the existence of these open issue tickets for a few months), Joe Conway who we can count on for hard-core PostgreSQL issues as well as Jeff Horner for general R, DBI and database smarts.

Some more details (including the full ChangeLog) are at the local RPostgreSQL page, the RPostgreSQL package page at CRAN and of course the Google Code repository for RPostgreSQL.

/computers/linux/debian/packages | permanent link

RInside release 0.1.1, and a fresh example
Last week's 0.1.0 release of RInside, and the first to have been published on CRAN, still had some issues with builds and use on OS X. Thanks to testing and fixes by Jan de Leeuw, Jeff Horner and particularly Simon Urbanked, things are said to be better now with the new release 0.1.1 which went onto CRAN yesterday. So no new features, but fixes to the main Makefile as well as the Makefile for the examples directory, some minor fixes and editing for the examples. I also added a file THANKS to show some appreciation for the various patches and fixes I have been receiving -- they are appreciated!

However, today I committed a new example to SVN archive at R-Forge. It is based on this thread on r-devel. Abhijit Bera tries to do this in C, but to me his questions provide rather clear motivation for showing how much simpler things can be via C++ and the Rcpp classes along with RInside. Using a small example, the task was to pass a weight vector to a portfolio solver from the Rmetrics package fPortfolio and to then access the computed solution. The original poster struggled with access from C to the S4 classes used by fPortfolio and could not set the weights. But when using RInside, we simply pass a C++ vector of weights down to R, solve the problem and pass a solution vector back using the handy evaluation of R expressions:

// -*- mode: C++; c-indent-level: 4; c-basic-offset: 4;  tab-width: 8; -*-
//
// Another simple example inspired by an r-devel mail by Abhijit Bera
//
// Copyright (C) 2009 Dirk Eddelbuettel and GPL'ed

#include "RInside.h"                    // for the embedded R via RInside
#include "Rcpp.h"                       // for the R / Cpp interface used for transfer
#include <iomanip>

int main(int argc, char *argv[]) {

    try {
        RInside R(argc, argv);          // create an embedded R instance
        SEXP ans;

        std::string txt = "suppressMessages(library(fPortfolio))";
        if (R.parseEvalQ(txt))          // load library, no return value
            throw std::runtime_error("R cannot evaluate '" + txt + "'");

        txt = "lppData <- 100 * LPP2005.RET[, 1:6]; "
	  "ewSpec <- portfolioSpec(); "
	  "nAssets <- ncol(lppData); ";
        if (R.parseEval(txt, ans))      // prepare problem
            throw std::runtime_error("R cannot evaluate '" + txt + "'");

	const double dvec[6] = { 0.1, 0.1, 0.1, 0.1, 0.3, 0.3 }; // choose any weights you want
	const std::vector<double> w(dvec, &dvec[6]);

	R.assign( w, "weightsvec");	// assign STL vector to R's 'weightsvec' variable

	txt = "setWeights(ewSpec) <- weightsvec";
        if (R.parseEvalQ(txt))		// evaluate assignment
            throw std::runtime_error("R cannot evaluate '" + txt + "'");

	txt = "ewPortfolio <- feasiblePortfolio(data = lppData, spec = ewSpec, constraints = \"LongOnly\"); "
	  "print(ewPortfolio); "
	  "vec <- getCovRiskBudgets(ewPortfolio@portfolio)";
        if (R.parseEval(txt, ans))      // assign covRiskBudget weights to ans
            throw std::runtime_error("R cannot evaluate '" + txt + "'");
	RcppVector<double> V(ans);      // convert SEXP variable to an RcppMatrix

	R.parseEval("names(vec)", ans);	// assign columns names to ans
	RcppStringVector names(ans);

	for (int i=0; i<names.size(); i++) {
	  std::cout << std::setw(16) << names(i) << "\t"
		    << std::setw(11) << V(i) << "\n";
        }

    } catch(std::exception& ex) {
        std::cerr << "Exception caught: " << ex.what() << std::endl;
    } catch(...) {
        std::cerr << "Unknown exception caught" << std::endl;
    }

    exit(0);
}

/computers/linux/debian/packages | permanent link

digest 0.4.0 and 0.4.1
After almost two years, I just made a small update to the digest package for R by adding sha-256 as another supported hashing function. While at this, a few minor edits where made here or there. Version 0.4.0 also included a file src/Makefile.win but as that actually prevents a successful build on the Windows platform I promptly removed it for a (more-or-less brown-bag) release 0.4.1. Thanks to Uwe Ligges and his automated win-builder service that permits easy unattended build tests on that platform.

The new package will appear shortly on its CRAN page; the R-Forge page has access to the Subversion repo etc and otherwise there's always my local my local digest page.

Oh, and 0.3.1 was never blogged about as it was just a maintenance release that added a single cast and an edit to the DESCRIPTION file.

/computers/linux/debian/packages | permanent link

RInside release 0.1.0 -- and now on CRAN
Version 0.1.0 of RInside, my C++ wrapper classes which facilitate embedding R into your own C++ application, has been released and is now an official CRAN package.

This release improves on the build process and should work on any sane Unix-alike operating system. A few more examples were added or extended. Details and the full ChangeLog are on my RInside page, and there is now a RInside page on CRAN as well.

/computers/linux/debian/packages | permanent link

RQuantLib 0.3.0 Windows build snag
Yesterday's upload of RQuantLib 0.3.0 contained one minor oversight: I had failed to update src/Makefile.win to the new and enlarged set of source files. The proper fix (of using wildcards and implicit rules) was simple, and Uwe Ligges kindly rebuilt RQuantLib on the Windows builder for CRAN using an updated Makefile so that binaries for Windows will be available. This fix will also be in the next upload. Should you need it in the meantime, here is a link to the new version directly from SVN.

/computers/linux/debian/packages | permanent link

RQuantLib 0.3.0 released
Earlier this evening, I rolled up a new version of RQuantLib. It has been pushed to CRAN and Debian, and source and binary version should appear on the respective mirror networks in due course.

This version, the first in a new '0.3.*' release series, contains all the work that Khanh Nguyen did during his Google Summer of Code participation which I had mentioned a while back and for which I acted as mentor.

And Khanh did great. We now have a lot more Fixed Income functionality; I include the full ChangeLog entry below. I also added some simple calendaring support and the odd fix here or there while mentoring. All of this had been available while technically 'in progress' thanks to support provided by the R-Forge project hosting where SVN checkouts as well as nightly binaries are provided. A few more details are on the RQuantLib page and I hope to add a few more examples.

As promised, the ChangeLog entry:

2009-09-05  Dirk Eddelbuettel  

	* DESCRIPTION: Release 0.3.0 reflecting all the excellent
	  Google Summer of Code 2009 work by Khanh Nguyen as well
	  as other small enhancements

	[ Changes by Khanh Nguyen below ]

	* R/bond.R: Added pricing functionality for various new instrument
	* R/discount.R: Idem
	* src/bonds.cpp: Idem
	* src/discount.cpp: Idem
	* src/utils.cpp: Idem

	* man/Bond.Rd: Added documentaiont for new functions
	* man/CallableBond.Rd: Idem
	* man/ConvertibleFixedCouponBond.Rd: Idem
	* man/ConvertibleFloatingCouponBond.Rd: Idem
	* man/ConvertibleZeroCouponBond.Rd: Idem
	* man/Enum.Rd: Idem
	* man/FittedBondCurve.Rd: Idem
	* man/FixedRateBond.Rd: Idem
	* man/FixedRateBondCurve.Rd: Idem
	* man/FixedRateBondPriceByYield.Rd: Idem
	* man/FixedRateBondYield.Rd: Idem
	* man/FloatingRateBond.Rd: Idem
	* man/ZeroCouponBond.Rd: Idem
	* man/ZeroPriceByYield.Rd: Idem
	* man/ZeroYield.Rd: Idem

	* rests/RQuantLib.R: Added tests for new functions
	* rests/RQuantLib.Rout.save: Added tests ouput for new functions

	[ Changes by Dirk Eddelbuettel below ]

	* man/BondUtilities.Rd: Added documentation for new function

	* R/calendars.R: Add support to access QuantLib calendars from R
	* src/calendars.cpp Idem
	* man/Calendars.Rd: Idem

	* src/bonds.cpp: Small C++ fixes to suppres g++ warnings
	* INDEX: Updated via 'R CMD build --force'
	* inst/QuantLib-License.txt: Updated to version from QL 0.9.7

/computers/linux/debian/packages | permanent link

random 0.2.1
A maintenance fix was applied to random: following the changes in release 0.2.0 a few months ago that reflected changes at the random.org site, we needed to make one more change in randomSequence. Thanks for Lynn Amon for spotting this.

The new package is available here and should appear on CRAN shortly.

/computers/linux/debian/packages | permanent link

Rcpp 0.6.6
Version 0.6.6 of Rcpp [ a set of C++ classes making it easier to glue C/C++ code to R ] went off to CRAN and Debian earlier as another small maintenance release.

While the diffstat output that will appear in cranberries may look impressive, it stems chiefly from updating the Doxygen documentation of the C++ classes. Only one small feature was added: one can now test via exists() whether a named parameter exists in an instance of the RcppParams class.

/computers/linux/debian/packages | permanent link

New RInside release
I just rolled up a new release of RInside, my C++ wrapper classes which facilitate embedding R into your own C++ application.

This releases owes a big Thank you! to Miguel Lechón who not only noticed errant behaviour and occassional segfaults with overly long commands sent to the embedded R, but even traced it to an oversight of mine in a simple memory buffer class and provided the one-line fix to reset a pointer! Much appreciated, especially as I got his mail two days before I talked about Rcpp and RInside at UseR! 2009 . The other changes are mostly cleanups, additions of two more test examples (to replicate the bug report) and some minor additions of a few new assign() functions to pass data between C++ and the embedded R session.

/computers/linux/debian/packages | permanent link

Rcpp 0.6.5
A minor new maintenance release 0.6.5 of Rcpp just went off to CRAN and Debian. This version corrects a small oversight for the OS X build, and adds the LGPL as file COPYING to the sources.

/computers/linux/debian/packages | permanent link

RQuantLib 0.2.11
The changes in Rcpp that I blogged about a few days ago required a few small changes in RQuantLib. Not really much more that prefixing std:: in a number of variable declarations and a few member function calls -- so this is definitely a minor maintenance release. New source and binary packages have already been pushed to CRAN and Debian.

/computers/linux/debian/packages | permanent link

Rcpp 0.6.4
A new maintenance version of Rcpp (now at 0.6.4) was just pushed to CRAN and has been uploaded to Debian. Rcpp is a set of utility classes that provide interfaces for transferring the major R data types to C++ and back which makes it easier to extend R with dynamically loadable code written in C or C++.

This version changes how use the std namespace: all usage is now properly prefixed. Likewise, we now define R_NO_REMAP to not let R define utility functions as length() or error() which occassionally creates trouble with other include files -- so now use the more explicit forms Rf_length(), Rf_error() etc. Also, starting from this release, C++ class documentation created by Doxygen is included; it can also be seen from my box as both browsable html and a pdf file. Lastly, this version also adds a minor correction to the Windows build (spotted by Uwe and Simon).

/computers/linux/debian/packages | permanent link

New project: RInside
A few days ago, I started a new project called RInside by uploading a few files to a new SVN repo at R-Forge.

RInside makes it easy to embed R into your own C++ application by hiding the nitty gritty of initializing an R interpreter behind a simple abstraction. More information is at a (currently pretty simple) RInside page, and you may want to look at the related Rcpp and possibly littler projects. The former is helpful for data exchange, and the latter provided my first real use of R embedding which in some ways also lead to RInside.

/computers/linux/debian/packages | permanent link

littler 0.1.2
Version 0.1.2 of r (pronounced littler) was just rolled up.

This version adds two new command-line switches:

• -t selects per-session temporary directories in the same way as R does (with thanks to Paul Gilbert for the suggestion), and
• -q skips autoloading of default libraries at startup for another small yet noticeable gain in startup speed (with thanks to Simon Urbanek).

As usual, our code in our svn archive, on my r page, and in the local directory here. A fresh package is in Debian's incoming queue, and Jeff's littler page at Vanderbilt should reflect the new release soon too.

/computers/linux/debian/packages | permanent link

Rcpp 0.6.3
I just pushed Rcpp 0.6.3 out to CRAN and Debian.

This version adds a fix to the OS X installation (thanks to Simon Urbanek), adds some 'view-only' classes for R vectors, matrices and string vectors (kindly suggested/provided by David Reiss) as well two shorter helper functions to derive compilation and linker flags for packages using Rcpp.

/computers/linux/debian/packages | permanent link

RQuantLib 0.2.10
Earlier this month and following the release of QuantLib 0.9.7, I updated RQuantLib to version 0.2.10. For once, there were no changes required to keep up with QuantLib. Rather, changes were internal as Rcpp had been spun off into its own package. As Rcpp is now in Debian, RQuantLib itself was also updated in Debian following the earlier upload to the CRAN network.

/computers/linux/debian/packages | permanent link

Rcpp now in Debian
Rcpp is an interface package that makes it easier to add C++ code to GNU R. It had started as a part of my RQuantLib project but has now come into its own as blogged here and here. As of today, Rcpp is now also available as a Debian package.

/computers/linux/debian/packages | permanent link

Rcpp relaunched with versions 0.6.0 and 0.6.1
I just announced Rcpp 0.6.0 and 0.6.1 on the low-volume R-packages list. Rcpp provides C++ classes that greatly facilitate interfacing C or C++ code in R packages using the .Call() interface provided by R.

Rcpp provides matching C++ classes for a large number of basic R data types. Hence, a package author can keep his data in normal R data structure without having to worry about translation or transfer to C++. At the same time, the data structures can be accessed as easily at the C++ level, and used in the normal manner.

The mapping of data types works in both directions. It is as straightforward to pass data from R to C++, as it is it return data from C++ to R.

Rcpp was initially written by Dominick Samperi to in the context of the RQuantLib package and later released on its own, but had not seen any releases in twenty-four months. I have substantially expanded the documentation, simplified the build structure yet made it easier to use Rcpp from other packages, and started to add some new classes (notably microsecond time types). Rcpp is supported on Windows, Linux and Mac OS X (with special thanks to Simon for some extended help).

More information for Rcpp can be found at the package homepage, the R-forge repository or the package CRAN page.

/computers/linux/debian/packages | permanent link

RDieHarder 0.1.0 released
I just rolled up version 0.1.0 of RDieHarder, an R package providing an interface between GNU R and the DieHarder battery of tests for random number generators developed by Robert G. Brown. See the the RDieHarder page for some introductory material and links to the talk at UseR! 2007.

Version 0.1.0 extends the functionality of the dieharder function quite substantially and catches up to a number of recent changes in DieHarder. In particular:

• dieharder() generator selection changes along the same line as in the DieHarder release: ids 1 to 200 are reserved for GNU GSL geneators, ids 201 to 400 for Dieharder, 401 to 500 for GNU R, 501 to 600 are hardware-based and ids over 600 are for user-contributed generators
• dieharder() now supports new arguments 'inputfile' (for file_input and file_input_raw) and 'ntuple' (for tests with variable bit length)
• dieharder() now also supports the 'rgb', 'sts' and 'user' tests
• dieharder() now returns multiple Kuiper KS p-values for those tests that generate multiple p-values
• dieharderGenerators() now returns a data.frame with two columns 'names' and 'ids' and generators can be selected via either a a name (as e.g. 'mt19937') or a numeric id (e.g. 16)
• dieharderTests() was added and also returns a data.frame with names and ids permitting a similar selection via test name or via test id.
• Some misc. code organisation, a cleanup removing more files, updated vignette output files, and the actual test sources updated to DieHarder 2.8.1

This new version should show up at CRAN and its mirrors in due course, in the meantime sources are also RDieHarder page.

/computers/linux/debian/packages | permanent link

littler 0.1.1 released
The new release 0.1.1 of r (pronounced littler) was just rolled up.

The only new feature is due to a suggestion by Paul Gilbert: r now reports the value of the optional status variable when calling q() at the end of a script:

$ r -e'q(status=42)'; echo $?
42


As usual, our code in our svn archive, on my r page, and in the local directory here. A fresh package is in Debian's incoming queue, and Jeff's littler page at Vanderbilt should reflect the new release soon too.

/computers/linux/debian/packages | permanent link

RQuantLib 0.2.9
As version 0.9.6 of QuantLib, which was released a couple of days ag, is now in Debian, I just uploaded an updated version of RQuantLib. Only minor API changes to src/curves.cpp were needed. This new version 0.2.9 is currently in the queue at R's master CRAN host and should hit the CRAN mirrors shortly; likewise the Debian package has been uploaded and should also propagate to Debian mirrors in due course. As usual, source are also available locally on my site. Lastly, RQuantLib is hosted on R-Forge and potential contributors are encouraged to register at R-Forge and to get in touch -- this is a great way to learn how to combine C++ and R.

/computers/linux/debian/packages | permanent link

Accelerated R in Debian
A few months ago, Stephen Milborrow started releasing a patched version of R that performs just-in-time compilation -- see his Ra page for some details and further pointers.

In a nutshell, Ra provides a modified R engine so that code preceded by all jit(1) function call, using his jit package from the CRAN archive, will run faster due to just-in-time compilation of loops and arithmetic expressions.

Ra offers to pick the low-hanging fruit for users as loops can be a bottleneck. Of course, as shown in Stephen's case study, using appropriate vectorised expression will often be faster still. That said, for a certain class of problems, Ra should offer a decent speed boost.

Debian users can now just say

    sudo apt-get install r-base-core-ra r-cran-jit

Lastly, version 1.1.0 of Ra was released by Stephen yesterday and is now also in Debian unstable.

/computers/linux/debian/packages | permanent link

smtm bug fix release 1.6.10
A new version of smtm just went to Debian and CPAN. Perl 5.10 required a small change in how we test whether certain arrays do, or do not, contain elements. No other changes were made.

/computers/linux/debian/packages | permanent link

getopt support for littler
Practically ever since Jeff and I released littler to add easy scripting for R, questions popped up about how to support getopt-alike command-line parsing.

And as of today, a new package r-cran-getopt is in Debian. It provides Allen Day's recently released package getopt from CRAN which provides a new function getopt.

Given a suitable data structure that provides long and short-form command-line option names, whether arguments are mandatory, optional or not required (as for flags), and a data-type, getopt munges the command-line arguments supplied by the user and fills a new variable opt accordingly. If a fifth columns with help text is provided, a usage string can be generated as well.

Thanks to Allen for writing getopt, for accepting a quick two-line patch extending support from Rscript to littler, and for fixing one or two minor bugs. Thanks also to the Debian ftpmasters for adding r-cran-getopt within a few days.

/computers/linux/debian/packages | permanent link

smtm maintenance release 1.6.9.1
A new version of smtm went to Debian yesterday. And after some pondering, I also just uploaded it to CPAN even though it only contains minor Debian packaging fixes rather than genuine code changes. In order to signal that the core Perl code hasn't changed, I simply added a new minor to move the version from 1.6.9 to 1.6.9.1.

/computers/linux/debian/packages | permanent link

PGApack 1.1: Almost as good as new
PGAPack is rather nice and fairly small library for 'parallel' optimisation via generic algorithms using the MPI message passing protocol. PGAPack 1.0 was written by David Levine while doing graduate work during the mid-1990s at Argonne Labs / University of Chicago.

PGAPack has also been in or around Debian for a rather long time, but it suffered from benign neglect in the last few years. Some of this came to the fore in this bugreport which lead to my offer to the then-maintainer Andreas to help on the relicencing request. After all, Argonne Labs is just a few miles from where I live, and I had already spent a little bit of time polishing and upgrading the package for my own exploratory use.

So I called Rusty Lust, head of the Mathematics and Computer Science section at Argonne to try to sort this out. He was sympathetic and put me in email contact with David Levine. As we are all somewhat busy, this dragged on for a little longer than we thought --- but as of today, about and a half years later, we have a new and shiny PGAPack 1.1 release, or around twelve years after the initial 1.0 version came out.

I have done a fair amount of polishing: there are now two library packages for serial use (i.e. for debugging) as well as parallel use via MPI. We use Open MPI where available and LAM where not. All open Debian bugs have been addressed. One minor issue in the postscript documentation remains as David can no longer locate his LaTeX sources; I may just have to extract the text and re-latex this from scratch to update it. One day.

Anyway, for full reference, the changelog entry is below. The package is currently in the NEW queue (as the new sub-package require manual inspection and approval) but should hit mirrors in a couple of days.

My thanks to the two previous Debian maintainers; to Rusty Lusk for helping with the from the end MSC department at Argonne Labs and for suggesting the rather liberal and easy MPICH2 license (and he happens to be one of the MPICH2 authors); and of course to David Levine for writing PGAPack in the first place, for agreeing to relicense it and giving valuable feedback on my repackaging of what is now version 1.1 on the MCS ftp server at Argonne --- this library has held up really well over the years; let's hope it will find more good use going forward.

pgapack (1.1-1) unstable; urgency=low

  * Really good news:  The MCS divsion of Argonne National Laboratories has
    agreed to relicense pgapack using the MPICH2 license. So pgapack
    is now Free Software and can move into Debian's main archive!
  
    Our thanks go to David Levine and Rusty Lusk to make this possible.
  
  * New maintainer, following Andreas' offer dated 2006-10-04 in #379388

  * debian/control: Change section to math		(Closes: #379388)
  
  * Added new brinary packages libpgapack-mpi1 and libpgapack-serial1
  * The MPI package is configured using Open MPI where available and LAM
    where not. 
  * debian/control: Changed Build-Depends: to use OpenMPI where available, 
    and LAM otherwise.
 
  * Finally acknowledges old NMUs 		(Closes: #379168,#359549)

  * source/integer.c: Apply patch for one-off error 	(Closes: #333381)

  * source/report.c:  Do not unconditionally print at generation 1
  
  * debian/rules: Remove a bashism 			(Closes: #379168)
  * debian/rules: Install examples directly 		(Closes: #134331)
  * debian/control: libpgapack-lam1 Depends on lam4 	(Closes: #60376)
  * debian/rules: Rewritten using debhelper
  * debian/control: Added Build-Depends: section for debhelper

  * No longer install mpi.h in /usr/include		(Closes: #404027)
    
  * debian/control: Updated Standards-Version: to current version

  * man/man1/PGAGetCharacterAllele.1: fix whatis entry 		(lintian)
  
 -- Dirk Eddelbuettel   Mon, 10 Mar 2008 18:03:34 -0500

/computers/linux/debian/packages | permanent link

littler 0.1.0 released
I've just rolled up release 0.1.0 of r (pronounced littler). A few changes made it into this release:

• A new option -l (with long form --packages) to load packages into the R session. This is useful when you have data-processing one-liners as it does away with the explicit library(foo) inside your actual R expression. For added bonus, library() is wrapped in a suppressMessage() call.
• The --no-restore option is now passed to the embedded R session.
• Argument handling was improved / corrected for a corner case.
• Output shown by --help was improved.
• The manual page and README were updated accordingly.
• The datasets package will now be autoloaded.
• Two minor fixes went into bootstrap, the wrapper around autotools and friends.
• The cache-clearing part of the R package updater example update.r was improved.
• We added some comparisons to Rscript in the timing tests.

As usual, our code in our svn archive, on my r page, and in the local directory here. A fresh package is in Debian's incoming queue, and Jeff's littler page at Vanderbilt should reflect the new release soon too.

/computers/linux/debian/packages | permanent link

RQuantLib 0.2.8
Version 0.9.0 of QuantLib was released a couple of days ago on Christmas Eve; as usual, Debian packages for QuantLib were uploaded right away following a few earlier pre-releases.

As QuantLib is approaching its 1.0 release, a few API changes requires updates to basically all of RQuantLib's C++ source files. Luckily most changes were minor. At the same time, we also generalised the Binary (aks Digital) option pricer to allow for a 'binType' argument (with values 'cash', 'asset' or 'gap' for CashOrNothing, AssetOrNothing or Gap digitals) as well as an 'excType' argument to switch between European and American exercise. Dominick made a small change to the DiscountCurve object to seamlessly pass a switch variable indicating whether we have 'flat' curves or not.

Another change was the addition of formal unit tests using the RUnit from CRAN (which we happened to have added to Debian recently in the wave of new RMetrics packages). We use the scheme initially proposed by Gregor Gojanc and extended by Martin Maechler for RMetrics that allows the unit tests to be run from both the source and the installed package which is nice. As QuantLib itself has a massive amount of unit tests in its code; I am hoping to add more and more of those into RQuantLib itself as we add more functionality.

On that front, more exciting news: RQuantLib is now hosted on R-Forge. Potential contributors are encouraged to register at R-Forge and to get in touch -- this is a great way to learn about combining C++ and R.

To wrap up, the new version 0.2.8 is currently in the queue at R's master CRAN host and should hit the CRAN mirrors shortly; likewise the Debian package has been uploaded and should also propagate to Debian mirrors in due course. As usual, source are also available locally on my site.

/computers/linux/debian/packages | permanent link

Several new Rmetrics packages
Debian has provided Rmetrics packages for financial engineering and computational finance since the Rmetrics release for R 1.9.0 in the summer of 2004. Over the years, Rmetrics has gotten more granular and changed from a handful of packages to two handfuls --- and the most recent release extended this trend even further to almost two dozen packages as shown in this chart.

Rmetrics now comprosises over twenty individual packages. Eleven new packages were added in the 260.72 release for R 2.6.0, and they required eight other new packages from CRAN. While I would have preferred a more spread-out approach than the shotgun approach of having to introduce all these new packages at once (which took the last four weeks), I am in support of the reorganisation which should make maintenance more easy going forward.

So to get all of these packages onto a Debian box, a quick sudo apt-get install r-cran-rmetrics is all it takes. Currently supported only in the always-fresh unstable flavour, but hopefully soon in testing too.

A big Thank You goes to the Debian FTPmasters. Of the 20-some packages that I added to Debian during this Rmetrics expansion, many were added within a day or two.

Lastly, thanks also to Florian Hahne, Robert Gentleman and Elijah Wright for much appreciated help with R's Rgraphviz and graph packages to create the chart above. It only takes a handful of lines to create the basic graph, and another few lines for the colours and titles. The code is available on request, of course, but you need the current development versions of the BioConductor packages Rgraphviz and graph (which are not in Debian yet).

/computers/linux/debian/packages | permanent link

Beancounter minor bug fix release 0.8.8
For some odd reason, the trusted Perl module Date::Manip now wants the "approx" argument to format date differences. Adding this in the five calls across beancounter and BeanCounter.pm is the main change in version 0.8.8 which is now in Debian's incoming area, uploaded to CPAN and onto the beancounter page here.

/computers/linux/debian/packages | permanent link

Overdue smtm bug fix releases 1.6.9
Some time ago, Yahoo! Finance made some changes at their backend for the provisioning of png images which are also displayed by smtm. Version 1.6.9, which just went to Debian's master site and will go to CPAN in a moment, finally updates the code for that. My apologies for taking so long, and thanks to Michael Kurlin, Clinton R Lambe, Ian Seow and possibly others whose mail I haven't kept who sent me patches and heads-ups.

/computers/linux/debian/packages | permanent link

New OpenMPI packages
Debian had OpenMPI package since early last year when Florian Ragwitz made some initial stabs at packaging. The package has seen a number of NMU and patches since then, but was generally getting cobwebs ... which was too bad because OpenMPI seems to have some wind behind its sails upstream. Unfortunately, little of that got packaged for Debian.

After some discussions on and around the debian-science list, a new maintainer group was formed on Alioth under the pkg-openmpi name. Tilman Koschnick (who had already helped Florian with patches), Manuel Prinz, Sylvestre Ledru and myself have gotten things in good enough shape in reasonably short time. And I have just uploaded a lintian-clean package set openmpi_1.2.3-0 to Debian, where it is expected to sit in the NEW queue for a little bit before moving on to the archive proper. The changelog entry (which will appear here eventually) shows twelve bugs closed.

Our plan is to provide a stable and well maintained MPI implementation for Debian. OpenMPI is the designated successor to LAM, and apart from MPICH2, everybody seems to have thrown their weight behind OpenMPI. So we will try to work with the other MPI maintainers to come up with sensible setups, alternatives priorities and the likes. If you are interested in MPI and would like to help, come join us at the Alioth project pkg-openmpi.

Last but not least, thanks to Florian for the initial packaging, and to Clint Adams, Mark Hymers, Andreas Barth, and Steve Langasek (twice even) for NMUs.

/computers/linux/debian/packages | permanent link

random 0.1.2
Following yesterday's minor maintenance release of random, a small brown bag fix release just went out to CRAN. Kurt Hornik, diligent as usual, spotted that the vignette would not build, and that has been corrected. The code itself is unchanged.

/computers/linux/debian/packages | permanent link

digest 0.3.0
A new version of digest has just been sent to CRAN. Thanks to excellent contributions by Simon Urbanek and Henrik Bengtsson, some internals of the code have been improved. The output produced by digest should now be invariant to R version number changes.

/computers/linux/debian/packages | permanent link

random 0.1.1
A minor maintenance release of random was just uploaded to CRAN. The only change is a correction on regular expression string: R 2.5.0 now warns that percentage signs should not be escaped. Not other code changes were made.

/computers/linux/debian/packages | permanent link

littler 0.0.11
I rolled up 'little r' (pronounced littler) version 0.1.11 earlier today. This release only includes a robustification of the main Makefile (so that e.g. BSDers can build with non-GNU make) and a small fix by Jeff to 'update.r', a handy example script.

As usual, littler can be found in the GoogleCode svn archive, on my r page and in the local directory, and soon on Jeff's littler page at Vanderbilt. The Debian package has been uploaded as well (and has been built again the new R version 2.5.0 that was released yesterday).

/computers/linux/debian/packages | permanent link

RQuantLib 0.2.6
Version 0.4.0 of QuantLib was released a few days ago (and Debian packages for QuantLib are waiting to be added by the ftpmaster).

This required some minor changes by Dominick in the Bermudan pricer, and we made some small updates in other place. All in all just a regular maintenance release. The new version 0.2.6 has now been uploaded to both R's master CRAN host and Debian, and is also available locally here.

/computers/linux/debian/packages | permanent link

Bug fix release of Finance::YahooQuote
Following up on the patch mentioned earlier, a new bug-fix release 0.22 of Finance::YahooQuote has been uploaded to Debian, CPAN and my yahooquote pages here. I also updated the Freshmeat record.

/computers/linux/debian/packages | permanent link

Yahoo! URL changes and Finance::YahooQuote, and hence Beancounter, breakage
Looks like Yahoo! Finance changed their server layout.

Data scraping via tools like my Finance::YahooQuote, and those that use it such as my Beancounter toolkit, are therefore inoperational. As a first quick fix, apply the following patch to YahooQuote.pm:

--- YahooQuote.pm.orig  2005-07-17 13:10:20.000000000 -0500
+++ YahooQuote.pm       2007-02-22 18:50:07.000000000 -0600
@@ -34,7 +34,7 @@
 $VERSION = '0.21';

 ## these variables govern what type of quote the modules is retrieving
-$QURLbase = "http://quote.yahoo.com/d?f=";
+$QURLbase = "http://download.finance.yahoo.com/d/quotes.csvr?e=.csv&f=";
 $QURLformat = "snl1d1t1c1p2va2bapomwerr1dyj1x";        # default up to 0.19
 $QURLextended = "s7t8e7e8e9r6r7r5b4p6p5j4m3m4";        # new in 0.20
 $QURLrealtime = "b2b3k2k1c6m2j3"; # also new in 0.20

/computers/linux/debian/packages | permanent link

littler 0.0.10 released
Another bug fix release -- this time triggered by Detlef's problems to build r (pronounced littler) on SUSE/OpenSUSE. Jeff also added better rng initialization fixing problems I noticed with tempfile(), added checks for the upcoming R 2.5.0, and made tests in tests/ use a path of ../r. Lastly, we added a short segment on the upcoming, err, shall I say 'cousin' Rscript that will be released with R 2.5.0.

As usual, everything is in our svn archive, on my r page, and in the local directory here. A fresh package is in Debian's incoming queue, and Jeff's littler page at Vanderbilt should reflect the new release soon too.

/computers/linux/debian/packages | permanent link

littler 0.0.9 released
Following a bug report earlier today, Jeff committed a quick fix to our svn archive. A few minutes ago, I wrapped it up as version 0.0.9 of r (pronounced littler), put it into the local directory and page here and sent it to Debian's incoming queue. It should also hit Jeff's littler page at Vanderbilt in due course.

The principal change is an added initialization to prevent the reported segfault, as well as a small improvement to the update.r script (which makes it easy to keep your CRAN packages up-to-date). I use littler all-day-long at work, and have found it rather reliable and easy, including during upgrades of the underlying R version.

/computers/linux/debian/packages | permanent link

Version 0.2.3 of digest
A new version of digest has been sent to CRAN and should appear on the CRAN mirrors in a few days; in the meantime you can fetch it from here too.

This version is a simple maintenance release that adds a call of path.expand() around the optional file argument so that things like ~ get expanded by R. My thanks to Jarek Tuszynski and Tom Bloemberg for the suggestion.

/computers/linux/debian/packages | permanent link