From: Joseph W. <jo...@gn...> - 2007-01-21 04:32:42
|
Sorry for the slowness, I'm juggling seven or eight things right now. Can you compile in debug mode and send a stack trace? If that doesn't work run it with a -dump_tags option. Swig shouldn't segfa= ult=20 at all, and there must be some unix/windows issue. My guess is that there = is=20 some pointer that isn't being initialized, and windows is less forgiving of= =20 this than unix. I'll try running swig under valgrind next week to see if I= =20 can track down something. Has anyone gotten this to run under windows? =E5=9C=A8 Tuesday 16 January 2007 13:19=EF=BC=8CDouglas Pearson =E5=86=99= =E9=81=93=EF=BC=9A > Thanks for the response Joseph. > > Unfortunately I'm still getting no joy. Running make doesn't work out of > the box and it looks like there's an assumption of doing some sort of > autoconf step -- which isn't easy on Windows. > > So I just tried building things manually. If I cd to > Swigwin-1.3.31\Examples\r\simple > > and execute (as Doc/Manual/R.html suggests): > ..\..\..\swig.exe -r -o example.c example.i > > I get an unhandled exception 0xc0000005. > > We've been using Swig just fine with other languages for a while (Java, C= #, > Tcl) but my understanding of Swig is still pretty basic. I usually just > follow the recipe and things work. > > Presumably this works fine for other folks so any idea what could be > different for me? Do I need to have something else built first or have R > installed in some known place? > > Doug > > > -----Original Message----- > From: Joseph Wang [mailto:jo...@gn...] > Sent: Friday, January 12, 2007 6:26 PM > To: bi...@su... > Cc: swi...@li...; William S Fulton > Subject: Re: [Fwd: [Swig-user] Swig and R on Windows] > > > I'm using swig-1.3.31 on Windows and trying to interface to the > > language "R". > > > > Following the sample steps for doing this on: > > http://www.omegahat.org/RSWIG/GettingStarted.html > > The version that is on omegahat is for a very old version that is not the > same as the one in the distribution. You should be able to go into > Examples/r in the distributions and use "make" to make those files work or > go into Examples/test-suite/r and do a "make check" there. > > > The comments do point out that R support is "experimental" but is it > > meant to be broken like this or am I doing something wrong? > > It should run against all of the files in the test-suite without breaking. > > > Doug > > -- > -------------------------------------------------------------------------= =2D- >- --- > Joseph Wang Ph.D. - jo...@gn... > China Derivatives Researcher and Software Developer - QuantLib > http://en.wikiversity.org/wiki/User:Roadrunner =2D-=20 =2D------------------------------------------------------------------------= =2D----- Joseph Wang Ph.D. - jo...@gn... =20 China Derivatives Researcher and Software Developer - QuantLib http://en.wikiversity.org/wiki/User:Roadrunner |