Menu
▾
▴
Re: [Quantlib-dev] QuantLib-1.8 | vc14 | AdaptiveRungeKutta testsuite error
|
From: Ivan A. C. <iva...@gm...> - 2016-09-07 14:30:54
|
Yes, release x64 version crashes with the "program stopped working" dialog. Maybe, because of the QL_FAIL (repro doesn't catch anything). I used boost 1.61.0 if that matters. On Sep 7, 2016 5:42 PM, "Peter Caspers" <pca...@gm...> wrote: > Hi, > > I can not reproduce this (I am using vc12 though), with your code and > x=3.15 I get > > y3=-0.00840725 exact3=-0.00840725 > > What do you mean by this example code “crashes”? Do you get a wrong result > or is it “really” crashing? > > Regards > Peter > > On 07 Sep 2016, at 13:44, Ivan A. Cherkasov <iva...@gm...> > wrote: > > I was able to reproduce the test case failure in a standalone repro > (crashes in release x64 only), please see attached along with the compiler > and linker commands. > > It fails on: > x = 3.15 > rk_real(ode3_, y30, 0.0, x) > On Sep 6, 2016 1:58 PM, "Ivan A. Cherkasov" <iva...@gm...> > wrote: > >> Hello, Luigi. >> >> Wasn't able to figure it out so far. I'm building with >> configuration:Release and Platform:x64. >> On Sep 6, 2016 1:47 PM, <qua...@li...> >> wrote: >> >>> Send QuantLib-dev mailing list submissions to >>> qua...@li... >>> >>> To subscribe or unsubscribe via the World Wide Web, visit >>> https://lists.sourceforge.net/lists/listinfo/quantlib-dev >>> or, via email, send a message with subject or body 'help' to >>> qua...@li... >>> >>> You can reach the person managing the list at >>> qua...@li... >>> >>> When replying, please edit your Subject line so it is more specific >>> than "Re: Contents of QuantLib-dev digest..." >>> >>> >>> Today's Topics: >>> >>> 1. Re: [Quantlib-users] QuantLib SWIG C# - Calling Conventions >>> (cdecl/stdcall) issue (Fabrice Lecuyer) >>> 2. Building QuantLib for C# - Is stdcall needed? (Fabrice Lecuyer) >>> 3. Potential Bug (Thompson Mark) >>> 4. Re: QuantLib-1.8 | vc14 | AdaptiveRungeKutta testsuite error >>> (Luigi Ballabio) >>> >>> >>> ---------------------------------------------------------------------- >>> >>> Message: 1 >>> Date: Thu, 11 Aug 2016 18:28:44 +1000 >>> From: Fabrice Lecuyer <fab...@gm...> >>> Subject: Re: [Quantlib-dev] [Quantlib-users] QuantLib SWIG C# - >>> Calling Conventions (cdecl/stdcall) issue >>> To: Luigi Ballabio <lui...@gm...> >>> Cc: qua...@li..., >>> qua...@li... >>> Message-ID: >>> <CAOEeWn5YK072TgpBFA=8wCV+PS7jN1arDqw1iOeLZXsEUvZqpw@ >>> mail.gmail.com> >>> Content-Type: text/plain; charset="utf-8" >>> >>> Hi Luigi, >>> >>> Thanks. I've made a pull request on Github. Note I've kept it as minimal >>> as >>> possible, I haven't updated the examples as I do not see any use for it >>> (plus getting some warnings due to the main functions calling convention >>> mismatches), and made the solution build configuration only build >>> QuantLib, >>> not the examples. >>> >>> I'd still be interested in discussing with anyone who is developing in C# >>> to confirm this is how they circumvented the problem. >>> >>> Regards, >>> Fabrice >>> >>> On 1 August 2016 at 18:41, Luigi Ballabio <lui...@gm...> >>> wrote: >>> >>> > Hello Fabrice, >>> > yes, adding the configurations would be nice, thanks. >>> > >>> > Luigi >>> > >>> > >>> > On Sat, Jul 2, 2016 at 12:07 PM Fabrice Lecuyer < >>> fab...@gm...> >>> > wrote: >>> > >>> >> Hi all, >>> >> >>> >> >>> >> (email already sent as a non member, apologies if it creates a >>> duplicate) >>> >> >>> >> >>> >> I?ve been building QuantLib & QuantLib SWIG ?as is? with VS14 for use >>> >> with C# and came across an initial error when running in debug, where >>> >> visual studio breaks giving a PInvoke stack imbalance error, as >>> described >>> >> in this old thread: >>> >> >>> >> https://sourceforge.net/p/quantlib/mailman/quantlib- >>> >> users/thread/33276539.post%40talk.nabble.com/#msg28797382 >>> >> >>> >> I first tried the solution described in this thread (modifying the >>> SWIG >>> >> generated file to add calling convention = cdecl) and all seemed ok >>> for a >>> >> while. >>> >> >>> >> The next problem appeared when QuantLib C++ was throwing an error, >>> the C# >>> >> wasn?t catching it and instead crashing with a memory imbalance error. >>> >> Again, this is described in the thread below, but this time without a >>> >> solution. >>> >> >>> >> In order to catch the thrown C++ error, I?ve had to revert to the >>> >> original SWIG generated C# source file, and instead build QuantLib C++ >>> >> using the stdcall calling convention (default in VS14 proj files is >>> cdecl). >>> >> >>> >> Now everything is good, but I just wanted to ask around to other C# >>> users >>> >> if this is a problem they have had, if they solved it this way that >>> I?m >>> >> safe to assume I won?t find another issue of the kind down the line, >>> or if >>> >> I?m unnecessarily complicating things. >>> >> >>> >> If this is indeed the best solution, would adding build configurations >>> >> for stdcall be considered? Happy to provide the updated vcproj files >>> if yes. >>> >> >>> >> >>> >> >>> >> Regards, >>> >> >>> >> Fabrice. >>> >> ------------------------------------------------------------ >>> >> ------------------ >>> >> Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in >>> San >>> >> Francisco, CA to explore cutting-edge tech and listen to tech >>> luminaries >>> >> present their vision of the future. This family event has something >>> for >>> >> everyone, including kids. Get more information and register today. >>> >> http://sdm.link/attshape_________________________________ >>> ______________ >>> >> QuantLib-users mailing list >>> >> Qua...@li... >>> >> https://lists.sourceforge.net/lists/listinfo/quantlib-users >>> >> >>> > >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> >>> ------------------------------ >>> >>> Message: 2 >>> Date: Tue, 23 Aug 2016 22:04:39 +1000 >>> From: Fabrice Lecuyer <fab...@gm...> >>> Subject: [Quantlib-dev] Building QuantLib for C# - Is stdcall needed? >>> To: qua...@li..., >>> qua...@li... >>> Message-ID: >>> <CAOEeWn4RdrQsQchOTO44xQsrz2XhyNwf_YttUetafQexdqBVWA@mail. >>> gmail.com> >>> Content-Type: text/plain; charset="utf-8" >>> >>> Hi all, >>> >>> >>> I'm trying to get a confirmation that building and using QuantLib in C# >>> requires QuantLib and potentially the Boost libraries built using the >>> *stdcall >>> *calling convention due to SWIG generated files forcing stdcall and >>> various >>> spurious issues when mismatching the calling conventions. >>> >>> Could any user or dev of QuantLib-SWIG for C# confirm they are doing so, >>> share if they are circumventing the issue(s) another way, or even mention >>> if they are not seeing any issue at all when building QuantLib >>> "normally"? >>> >>> Any help would be much appreciated. >>> >>> Regards, >>> Fabrice >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> >>> ------------------------------ >>> >>> Message: 3 >>> Date: Sun, 28 Aug 2016 11:31:10 +0000 >>> From: "Thompson Mark" <ma...@st...> >>> Subject: [Quantlib-dev] Potential Bug >>> To: "qua...@li..." >>> <qua...@li...> >>> Message-ID: <D1F...@st...> >>> Content-Type: text/plain; charset="utf-8" >>> >>> Dear Quantlib Maintainers, >>> >>> I?m an analyst trying to set QuantLib with Anaconda Python up on Mac OSX >>> El Capitan 10.11.6. >>> >>> 0. After some strange ac file errors compiling, and a compile with >>> pointer errors under testing, I finally got the lastest version of QuantLib >>> to compile with ?brew?. >>> 1. brew currently defaults to 1.7 I got brew to compile 1.8 by modifying >>> the ruby formula. >>> 2. The quantlib test-suite reports all tests passed >>> >>> >>> ? >>> A. I downloaded Quantlib-SWIG 1.8 and compiled and installed it. >>> B. I got a strange symbol when trying to import the library in python: >>> >>> ImportError: dlopen(/Users/mark/anaconda/li >>> b/python3.5/site-packages/QuantLib/_QuantLib.cpython-35m-darwin.so >>> <http://quantlib.cpython-35m-darwin.so/>, 2): Symbol not found: >>> __ZN8QuantLib10DateParser14parseFormattedERKSsS2_ >>> Referenced from: /Users/mark/anaconda/lib/pytho >>> n3.5/site-packages/QuantLib/_QuantLib.cpython-35m-darwin.so >>> <http://quantlib.cpython-35m-darwin.so/> >>> Expected in: flat namespace >>> in /Users/mark/anaconda/lib/python3.5/site-packages/QuantLib/_Q >>> uantLib.cpython-35m-darwin.so <http://quantlib.cpython-35m-darwin.so/> >>> >>> C. I get a similar error in when running the tests included in the >>> Python director of QuantLib-SWIG-1.8. >>> >>> >>> D. I read that Removed DateParser::split method (deprecated in version >>> 1.6). is this related? >>> >>> >>> I?d be happy to write a detailed OSX guide for 1.8 as the instructions, >>> but would need help solving this last issue which I now think has to do >>> with this DateParser change. >>> >>> Any hints or suggestions? >>> >>> >>> >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> >>> ------------------------------ >>> >>> Message: 4 >>> Date: Tue, 06 Sep 2016 09:44:55 +0000 >>> From: Luigi Ballabio <lui...@gm...> >>> Subject: Re: [Quantlib-dev] QuantLib-1.8 | vc14 | AdaptiveRungeKutta >>> testsuite error >>> To: Peter Caspers <pca...@gm...>, "Cherkasov, Ivan" >>> <Iva...@ot...> >>> Cc: "qua...@li..." >>> <qua...@li...> >>> Message-ID: >>> <CAJkxnzcdPu-Bxdhokz5mV7-YHjthAuTGojf5VqRu_j4NkBMB0Q@ >>> mail.gmail.com> >>> Content-Type: text/plain; charset="utf-8" >>> >>> Hello Ivan, >>> could you make any check on this? >>> >>> Thanks, >>> Luigi >>> >>> On Thu, Jul 28, 2016 at 3:17 PM Peter Caspers <pca...@gm...> >>> wrote: >>> >>> > Hi Ivan, this is not OK. Can you first check the following: You seem to >>> > run the 64bit executable, but the platform displayed in the test suite >>> is >>> > Win32. Something seems to have got mixed here. Regards, Peter >>> > >>> > On 28 Jul 2016, at 06:01, Cherkasov, Ivan <Iva...@ot... >>> > >>> > wrote: >>> > >>> > Hello. >>> > >>> > I get two failing test cases for AdaptiveRungeKutta in vc14: >>> > >>> > 1> Testing adaptive Runge Kutta... >>> > 1>unknown location : fatal error : in "QuantLib test suite/ode >>> > tests/QuantLib::detail::quantlib_test_case(&OdeTest::testAda >>> ptiveRungeKutta)": >>> > class QuantLib::Error: Stepsize (3.1414) underflow in >>> > AdaptiveRungeKutta::rkqs >>> > 1> c:\tmp\ql18_vc2015\quantlib-1.8\test-suite\utilities.hpp(74): last >>> > checkpoint >>> > 1> Testing matrix exponential based on ode... >>> > 1>unknown location : fatal error : in "QuantLib test suite/ode >>> > tests/QuantLib::detail::quantlib_test_case(&OdeTest::testMat >>> rixExponential)": >>> > class QuantLib::Error: Too many steps (10000) in AdaptiveRungeKutta >>> > >>> > Testsuite output is attached. >>> > >>> > Please advise if this is OK. >>> > >>> > Thank you. >>> > >>> > ------------------------------ >>> > ?????????? ?????????? ????????????? ?????? ??? ???????? ??? >>> ???????????? >>> > ????, ? ???????? ??????????, ? ????? ????????? ???????????????? >>> ????????. >>> > ????? ?????, ????????, ??????????????? ??? ?????? ????????????? >>> ?????????? >>> > ?????? ??? ????????? ????? ??????????? ?????????? ?????????. ???? ?? >>> > ???????? ?????? ????????? ?? ??????, ??????? ???????? ?? ???? >>> ??????????? >>> > ????? ????????? ? ??????? ????????? ?? ????? ???????????? ? ????. >>> > >>> > This e-mail message and any attachment are for the sole use of the >>> > intended recipient(s) and may contain confidential and privileged >>> > information. Any unauthorized review, use, disclosure or distribution >>> is >>> > prohibited. If you are not the intended recipient, please contact the >>> > sender by reply and destroy all copies of the original message and any >>> > attachments thereto. >>> > Unless specifically stated otherwise, this communication is for >>> > information purposes only and may not be considered as advice, is not >>> an >>> > offer or solicitation with respect to any transaction or an >>> advertisement, >>> > and does not form a fiduciary relationship. For details of the >>> > disseminating legal entity and terms relating to risks, other >>> conditions, >>> > and your reliance on information contained herein, see >>> > http://www.capital.otkritie.com/en/disclaimer/. If you cannot access >>> the >>> > link, please notify the sender by reply message and you will be sent >>> the >>> > full contents of this notice. Otkritie Capital International Limited, >>> > authorized and regulated by the Financial Conduct Authority (FCA). >>> > >>> > <testsuite.txt> >>> > ------------------------------------------------------------ >>> ------------------ >>> > >>> > >>> > _______________________________________________ >>> > QuantLib-dev mailing list >>> > Qua...@li... >>> > https://lists.sourceforge.net/lists/listinfo/quantlib-dev >>> > >>> > >>> > ------------------------------------------------------------ >>> ------------------ >>> > _______________________________________________ >>> > QuantLib-dev mailing list >>> > Qua...@li... >>> > https://lists.sourceforge.net/lists/listinfo/quantlib-dev >>> > >>> -------------- next part -------------- >>> An HTML attachment was scrubbed... >>> >>> ------------------------------ >>> >>> ------------------------------------------------------------ >>> ------------------ >>> >>> >>> ------------------------------ >>> >>> _______________________________________________ >>> QuantLib-dev mailing list >>> Qua...@li... >>> https://lists.sourceforge.net/lists/listinfo/quantlib-dev >>> >>> >>> End of QuantLib-dev Digest, Vol 123, Issue 1 >>> ******************************************** >>> >> <repro_rk.cpp><CL.command.1.tlog><link.command.1.tlog>---- > -------------------------------------------------------------------------- > _______________________________________________ > QuantLib-dev mailing list > Qua...@li... > https://lists.sourceforge.net/lists/listinfo/quantlib-dev > > > |
