Missing Class Export in R

[AndLLA]

Setup:

• fresh installation of QuantLib 1.27 + QuantLib-SWIG 1.27
• R 4.2.1
• swig 4.02 (as distributed with - Ubuntu 22.04)

Error (after trying to QuantLib::DiscountCurve): Error in getClass(Class, where = topenv(parent.frame())) : “_p_ext__shared_ptrT_InterpolatedDiscountCurveTLogLineart_t” is not a defined class

Sample code for reproducing the bug:

library(QuantLib)

bToday = QuantLib::Date(31, "August", 2022)
QuantLib::Settings_setEvaluationDate(self = QuantLib::Settings_instance(), d = bToday)
print(Settings_instance()$getEvaluationDate())

bufDates = QuantLib::DateVector()
QuantLib::DateVector_append(self = bufDates, x = Date(31, "August", 2022))
QuantLib::DateVector_append(self = bufDates, x = Date(30, "September", 2022))
QuantLib::DateVector_size(self = bufDates)

bCurve = QuantLib::DiscountCurve(dates = bufDates , discounts = c(1.0, 1.0), dayCounter = Actual360())

in discountcurve.i the necessary export seem to be present:

%template(DiscountCurve) InterpolatedDiscountCurve<LogLinear>;

other functions seem to work correctly.

[boring-cyborg[bot]]

Thanks for posting! It might take a while before we look at your issue, so don't worry if there seems to be no feedback. We'll get to it.

[AndLLA]

some additional observations:

• compiling QL for python in the same environment produces a working module when testing these cases
• it seems that in R if a templated class is exported using %template of a shared_ptr we get broken code (see log below for

> bCurve = QuantLib::DiscountCurve(dates = bufDates , discounts = c(1.0, 1.0), dayCounter = Actual360())
Error in getClass(Class, where = topenv(parent.frame())) : 
  “_p_ext__shared_ptrT_InterpolatedDiscountCurveTLogLineart_t” is not a defined class

> bCurve = QuantLib::MonotonicLogCubicDiscountCurve(dates = bufDates , discounts = c(1.0, 1.0), dayCounter = Actual360())
Error in getClass(Class, where = topenv(parent.frame())) : 
  “_p_ext__shared_ptrT_InterpolatedDiscountCurveTMonotonicLogCubict_t” is not a defined class

> bCurve = QuantLib::NaturalCubicDiscountCurve__SWIG_1(dates = bufDates , discounts = c(1.0, 1.0), dayCounter = Actual360(), calendar = TARGET())
Error in getClass(Class, where = topenv(parent.frame())) : 
  “_p_ext__shared_ptrT_InterpolatedDiscountCurveTSplineCubict_t” is not a defined class

> bCurve = QuantLib::ZeroCurve__SWIG_3(dates = bufDates , yields =  c(0.01, 0.01), dayCounter = Actual360(), calendar = TARGET())
Error in getClass(Class, where = topenv(parent.frame())) : 
  “_p_ext__shared_ptrT_InterpolatedZeroCurveTLineart_t” is not a defined class

> bCurve = QuantLib::ForwardCurve__SWIG_1(dates = bufDates, forwards =  c(0.01, 0.01), dayCounter = Actual360(), calendar = TARGET())
Error in getClass(Class, where = topenv(parent.frame())) : 
  “_p_ext__shared_ptrT_InterpolatedForwardCurveTBackwardFlatt_t” is not a defined class

did it ever work in R ?

is this a config problem on QL side or on SWIG side?

[AndLLA]

opened an issue also on swig side, as it seems it's maybe more related to swig: https://github.com/swig/swig/issues/2372

[lballabio]

It does seem more related to swig, but we'll need a self-contained example for them to have a chance at diagnosing the problem. Let me know if I can help with that.

Unfortunately the CI build (see https://github.com/lballabio/QuantLib-SWIG/blob/master/.build/r.build) just tries compiling the R module. It would be great if it also tried running a few examples (e.g. those in R/demo) but I don't know R so I have no idea how to do that. I'm open to suggestions.

[AndLLA]

Hallo, I guess I managed to replicate the bug in a easy way for the SWIG team, maybe you can "up-vote" the bug :) ?

I can try to stitch together something for the R CI build, but right now I fear it would just keep on failing, because basically any template class does not work. I looked at the code swig produces, it seems some king of weird naming issue, but I'm no swig expert.

Thanks

[AndLLA]

Hallo, I guess I fixed it in swig with https://github.com/swig/swig/pull/2386

[lballabio]

Thanks — let's see what happens.