These are chat archives for symengine/symengine

14th
May 2016
Isuru Fernando
@isuruf
May 14 2016 09:56 UTC
@irislq, would it be possible for you to send a new PR for changing the structure of UnivariateExprPolynomial and UnivariatePolynomialwithout @chenchfort's fft changes?
I'd like to get this in to master within the next couple of days
Srajan Garg
@srajangarg
May 14 2016 10:53 UTC
@isuruf what's the reason that map_uint_mpzshouldn't be used as the container?
Isuru Fernando
@isuruf
May 14 2016 10:54 UTC
Commented on the PR
The point of a templated function is to write common code. There's nothing common about map_uint_mpz and piranha::polynomial. How can they be used to write code that is common for both?
Srajan Garg
@srajangarg
May 14 2016 10:56 UTC
So, the UnivariateIntPolynomial class should be used?
But then we won't have stuff common between these classes too right?
Isuru Fernando
@isuruf
May 14 2016 11:00 UTC
I'm not sure about how to use the UnivariateIntPolynomial. Let's see how UnivariateBase would work
When designing the API, it's better to think about the general stuff early
UnivariateExprPolynomial, piranha::polynomial and flint::fmpq_polyxx all have arithmetic operators overloaded right?
So, if we are writing a algorithm for polynomials, we can write it in the UnivariateBase class itself so that each container can be used everywhere
Srajan Garg
@srajangarg
May 14 2016 11:05 UTC
Right
Does this mean we need operator overloads first
Only then a common interface can be formed
Isuru Fernando
@isuruf
May 14 2016 11:07 UTC
Yes.
See this commit by @irislq, chenchfort/symengine@dc0f39c
Srajan Garg
@srajangarg
May 14 2016 11:12 UTC
Will have a look
Charles Chen
@chenchfort
May 14 2016 13:13 UTC
@isuruf I have submitted the PR, #940, on behalf of her. Please review.
Isuru Fernando
@isuruf
May 14 2016 13:24 UTC
Thanks
I
I'll review
Iris Lui
@irislq
May 14 2016 19:15 UTC
@isuruf @chenchfort I made the changes you asked