Re: [Quantlib-users] QuantLib parameters/enums from a string

[David S. <da...@el...>]

Thanks Peter, I’ll take a look!

I am in the fortunate position of being in control of all parts of the system, so I should be able to maintain consistency.

I think I will probably go for an object repository indexed by a string. The QuantLibXL Excel interface has handy enumeration functions, so I can probably generate a lot of the code from that.

Best

David

> On 9 Apr 2021, at 08:55, Peter Caspers <pca...@gm...> wrote:
> 
> Yes, we have functions to convert strings into QuantLib / QuantExt
> objects and back, see
> 
> https://github.com/OpenSourceRisk/Engine/blob/master/OREData/ored/utilities/parsers.cpp
> 
> However, we don't aim at a complete coverage of all existing C++
> classes nor do we strive to support aliases from all possible systems
> in the world. Also, for certain objects like calendars, currencies,
> also some index types, we move more and more to dynamically created
> objects, i.e. we set up the objects from the contents of an xml file.
> That is of course not possible in cases where the objects involve
> complicated computations like ActAct ICMA day counter, but for the
> Chinese Calendar it's easier than to update C++ and depend on  the
> release cycle.
> 
> Thanks, Peter
> 
>> On Thu, 8 Apr 2021 at 15:59, Francois Botha <ig...@gm...> wrote:
>> 
>> I'm guessing here, but does OpenRiskEngine not do something like that? Maybe Peter can confirm. If so, maybe David can leverage off the work that has been done there.
>> 
>> regards
>> Francois Botha
>> 
>> 
>>> On Thu, 8 Apr 2021 at 15:40, Luigi Ballabio <lui...@gm...> wrote:
>>> 
>>> Hello David,
>>>    I'd prefer not to do that in the library.  It would mean not only maintaining the factory, but also the whole set of different day-counter names that people might have in their existing DB because they're getting data from Reuters or Bloomberg or whatever provider they're using.
>>> 
>>> Luigi
>>> 
>>> 
>>> On Wed, Apr 7, 2021 at 7:40 PM <da...@el...> wrote:
>>>> 
>>>> Hi,
>>>> 
>>>> This may be something that has been asked before, but I am trying to rebuild various DayCounters (and indeed other Enums/classes) from a string-based name in C++. My goal is to store the configurations needed to build a QL bond in an SQL database (eg if I want to build a Belgian spline curve, from a defined list of bond ISINs).
>>>> 
>>>> In Python, with its run-time type discovery, I can save the string representation of the class or enum into the database, and get the object back using eval(). However that isn’t much help for when/if I migrate the code to C++. I’d much prefer to use the string-based name, as in QuantLibXL (see QuantLibXL: Enumerations ) which has the added benefit that I can use the same definitions in Excel.
>>>> 
>>>> 
>>>> 
>>>> It is simple enough to write a DayCounterFactory class in C++, build a map of {name,instance of QL class), and get back a DayCounter object of the correct type. However, I run the risk of being out of sync with any changes made in QuantLib (eg new enums).
>>>> 
>>>> Is there any chance of putting this factory into QuantLib itself, and hence maintaining it along with the rest of the project? This may have been asked before, and rejected for sound conceptual reasons of which I am unaware!
>>>> 
>>>> 
>>>> 
>>>> Best wishes,
>>>> 
>>>> 
>>>> 
>>>> David Sansom
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> 
>>>> _______________________________________________
>>>> QuantLib-users mailing list
>>>> Qua...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/quantlib-users
>>> 
>>> _______________________________________________
>>> QuantLib-users mailing list
>>> Qua...@li...
>>> https://lists.sourceforge.net/lists/listinfo/quantlib-users
>> 
>> _______________________________________________
>> QuantLib-users mailing list
>> Qua...@li...
>> https://lists.sourceforge.net/lists/listinfo/quantlib-users
> 
> 
> _______________________________________________
> QuantLib-users mailing list
> Qua...@li...
> https://lists.sourceforge.net/lists/listinfo/quantlib-users