[Rquantlib-devel] RQuantLib fixed-income discussion (Re: 1 Rquantlib-devel moderator request(s) waiting)

[Michele Salvadore]

On 8 October 2014 13:52, Dirk Eddelbuettel <edd at debian.org> wrote:

>
> On 8 October 2014 at 10:53, Michele Salvadore wrote:
> | Hi Dick,
>
> Dirk, with an r. It happens :)
>

So sorry!!! apologies for the pun, was definitely not intended.

>
> | Eventually I was able to get it up.
>
> Ok. If you have suggestions for making the documentation clearer let me
> know.
>
> | I've been working a bit on the bond interface as I needed to make it more
> | flexible and match more closely the structure on the QuantLib side.
> Briefly, I
> | changed the interface (just for FixedRateBond for the time being) to
> match the
> | most general case available on the quantlib side:
> | 1. provide a named list 'bond' with all parameters matching
> | the QuantLib::FixedRateBond ctor (with externally constructed schedule),
> apart
> | from the schedule (constructed separately) and the coupon rates vector
> (also
> | constructed separately).
> | 2. provide a named list 'schedule' with all parameters matching the
> | QuantLib::Schedule ctor.
> | 3. provide a named list 'calc' with parameters required for some
> additional
> | calculations (for example duration)
> | 4. Either a yield or a discountCurve can be provided (if both are not
> available
> | or defined I stop()). Depending on which is given, we include a price
> engine or
> | not and calculate parameters accordingly (in case of yield, as there is
> no
> | price engine, I return NA for NPV). In both cases, the same type of
> object is
> | returned to R, a FixedRateBond.
> | 5. I added some additional field to the returned object: the
> settlementDate
> | used in the calculation (not necessarily obvious from the input
> parameters) and
> | the duration.
> | 6. I extended the supported calendars by using the getCalendar function
> already
> | available in calendars.cpp (there was a different getCalendar function
> declared
> | in the header which didn't have an implementation and I substituted it
> with the
> | correct declaration).
> |
> | I was wondering what you think of these changes. I would like to
> contribute
> | them back to the package, but I am worried about the fact the the need to
> | extend the interface has broken backward compatibility for
> FixedRateBond. I was
>
> I generally like backwards compatibility.
>

Me too, as long as it is possible.

>
> This may be an exception where it is worth breaking things.
>

Yes, I belive the current API is limited and not flexible enough for
pricing bonds.

>
> | initially thinking of adding a new function with a different name, but I
> | believe that would be very confusing for users. My believe is that the
> | FixedRateBond should match as closely as possible the
> QuantLib::FixedRateBond
> | (even the name says that :)).
> | If you think it would be worthile, I would implement similar changes to
> the
> | other bond functions in the package and contribute it back.
> |
> | Looking forward to hearing your opinion on this,
>
> Can you provide a 'pull request' ?  That is the GitHub way of clearly
> showing
> the differences.  If you know (some) git, creating a branch and a pull
> request is easy.  If you don't know git, it may be worth learning.  We can
> also fall back to 'diff and patch'.
>

Yes, git is my versioning tool of choice, I will clean things up a bit and
provide a pull request.

Michele

>
> This discussion should be on the mailing list, though, so I am reverting it
> there. Please follow-up on list, if possible.
>
> Dirk
>
> | Michele
> |
> |
> | On 1 October 2014 18:32, Dirk Eddelbuettel <edd at debian.org> wrote:
> |
> |
> |     Hi again,
> |
> |     On 1 October 2014 at 18:18, Michele Salvadore wrote:
> |     | Hi Dirk,
> |     |
> |     | sorry to bother you, I was wondering if you had the chance to take
> a look
> |     at
> |     | this issue (multiple definitions at linkage). It looks like
> something to
> |     do
> |
> |     I have not.
> |
> |     RQuantLib builds at CRAN incl on the win-builder service.
> |
> |     | with proper header inclusion, but I'm not sure what's going on.
> Btw, I
> |     was able
> |     | to get all other issues/warning sorted (including openmp and so
> on), this
> |     is
> |     | the last bit left before being able to work on it.
> |
> |     Can you construct a small reproducible example and post it (within
> the 40kb
> |     limit, ie do not attach or include lots) to the list?
> |
> |     Dirk
> |
> |     | Thanks a lot, apologies for bothering!
> |     |
> |     | Michele
> |     |
> |     | On 26 September 2014 22:10, Dirk Eddelbuettel <edd at debian.org>
> wrote:
> |     |
> |     |
> |     |     On 26 September 2014 at 20:48, Michele Salvadore wrote:
> |     |     | Sorry about that dirk, will keep it shorter in the future :)
> i was
> |     out of
> |     |     Office today and saw emails just now.
> |     |
> |     |     No worries.
> |     |
> |     |     What is a little stranger is that I can see the post now in
> the list
> |     |     archives, yet I never seem to have gotten a copy to my account.
> |     |
> |     |     We'll follow up in a bit.
> |     |
> |     |     Dirk
> |     |
> |     |     --
> |     |     http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org
> |     |
> |     |
> |
> |     --
> |     http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org
> |
> |
>
> --
> http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rquantlib-devel/attachments/20141008/75a08078/attachment.html>