-----Original Message----- From: boost-users-bounces@lists.boost.org [mailto:boost-users- bounces@lists.boost.org] On Behalf Of Steven Watanabe Sent: August-29-11 11:42 AM To: boost-users@lists.boost.org Subject: Re: [Boost-users] why aren't I getting multithreaded versions of
Thanks Steve, Alas, no joy. I followed your direction, and saw b2 building or revising a few targets, but the result of make in QuantLib's directory (version 1.1 with boost 1.47.0 ), is: libtool: link: g++ -g -O2 -o .libs/quantlib-test-suite.exe quantlibtestsuite.o americanoption.o array.o asianoptions.o assetswap.o autocovariances.o barrieroption.o basketoption.o batesmodel.o bermudanswaption.o blackdeltacalculator.o bonds.o brownianbridge.o calendars.o capfloor.o capflooredcoupon.o cashflows.o cdo.o cdsoption.o chooseroption.o cliquetoption.o cms.o commodityunitofmeasure.o compoundoption.o convertiblebonds.o covariance.o creditdefaultswap.o curvestates.o dates.o daycounters.o defaultprobabilitycurves.o digitalcoupon.o digitaloption.o distributions.o dividendoption.o europeanoption.o everestoption.o exchangerate.o extendedtrees.o factorial.o fastfouriertransform.o fdheston.o fdmlinearop.o forwardoption.o gaussianquadratures.o gjrgarchmodel.o hestonmodel.o himalayaoption.o hybridhestonhullwhiteprocess.o inflation.o inflationcapfloor.o inflationcapflooredcoupon.o inflationvolatility.o instruments.o integrals.o interestrates.o interpolations.o jumpdiffusion.o libormarketmodel.o libormarketmodelprocess.o linearleastsquaresregression.o lookbackoptions.o lowdiscrepancysequences.o margrabeoption.o marketmodel.o marketmodel_cms.o marketmodel_smm.o marketmodel_smmcapletalphacalibration.o marketmodel_smmcapletcalibration.o marketmodel_smmcaplethomocalibration.o matrices.o mclongstaffschwartzengine.o mersennetwister.o money.o nthtodefault.o operators.o optimizers.o optionletstripper.o overnightindexedswap.o pagodaoption.o pathgenerator.o period.o piecewiseyieldcurve.o quantooption.o quotes.o rangeaccrual.o riskstats.o rngtraits.o rounding.o sampledcurve.o shortratemodels.o solvers.o stats.o surface.o swap.o swapforwardmappings.o swaption.o swaptionvolatilitycube.o swaptionvolatilitymatrix.o termstructures.o timeseries.o transformedgrid.o tqreigendecomposition.o tracing.o utilities.o variancegamma.o varianceoption.o varianceswaps.o volatilitymodels.o -lboost_unit_test_framework-mt ./.libs/libUnitMain.a ../ql/.libs/libQuantLib.a /usr/lib/gcc/i686-pc-cygwin/4.3.4/libstdc++.dll.a -L/usr/lib/gcc/i686-pc-cygwin/4.3.4 ./.libs/libUnitMain.a(libUnitMain_la-main.o): In function `main': /home/Ted/QuantLib-1.1/test-suite/main.cpp:11: undefined reference to `boost::unit_test::unit_test_main(bool (*)(), int, char**)' ./.libs/libUnitMain.a(libUnitMain_la-main.o): In function `_Z13init_functionv': /home/Ted/QuantLib-1.1/test-suite/main.cpp:7: undefined reference to `boost::unit_test::framework::master_test_suite()' collect2: ld returned 1 exit status make[1]: *** [quantlib-test-suite.exe] Error 1 make[1]: Leaving directory `/home/Ted/QuantLib-1.1/test-suite' make: *** [all-recursive] Error 1 It is looking for a file ' libboost_unit_test_framework-mt.a', which still does not exist. I assume ' boost::unit_test::framework::master_test_suite' is in that library. The boost documentation I looked at indicates I ought to see libraries ending in '-mt' if the library supports multithreading. What else ought I try? the
libraries that need to be built?
AMDG
On 08/29/2011 08:12 AM, Ted Byers wrote:
This is in cygwin, using gcc.
initially I used the following (I don't use python and don't have MPI, but wanted everything else):
$ ./bootstrap.sh --with-libraries=chrono,date_time,exception,filesystem,graph,graph_par allel, iostreams,math,program_options,random,regex,serialization,signals,syst em,tes t,thread,wave
./b2 install
I then tried building QuantLib, which makes extensive use of boost, including the test framework. But it failed because it tried linking to a multihtreaded build of the test frameqwork.
<snip>
You are getting a multithreaded library. The problem is that the name is not what QuantLib apparently expects. To get the old naming convention, use --layout=tagged.
In Christ, Steven Watanabe