Re: [boost] Gsoc chrono::date Library

Subject: Re: [boost] Gsoc chrono::date Library
From: Vicente J. Botet Escriba (vicente.botet_at_[hidden])
Date: 2014-02-08 11:35:16

Le 07/02/14 22:39, Arun Sori a écrit :
> Hey All,
Hi Arun,
> I am interested in contributing to the chrono::date library project.I
> have read the paper given in the link and need some ideas on how to
> start. I am a new contributor so please guide me to the initial phase
> and spare no details.
Great.
>
> As a start the paper seems to propose a new library based on boost::date.
Humm based on std::chrono, isn't it? Could you point to which paper you 
are talking of.
> Should I first familiarize with the boost::date code ( try to compile
> it. Read the documentations etc.).
Yes this is a good starting point. It would be also interesting to 
google for discussion about the subject.
Note that I have tried it since last GSoC (2013/05?) so maybe there are 
things that are not working now.

> The project also has a prototype
> available. Are we trying to build the library further upon it ? What
> extra do we need to built in this project ?
The paper from H.Hinnant as well as the current Boost.Chrono/Date 
prototype take a slightly different direction. H.H. paper has a single 
date class that takes in account absolute and relative dates. My 
prototype separates absolute dates and relative dates. In addition my 
prototype Date is much more a concept than a class and there could be 
several models of this concept. The idea is to implement several design 
and see which one performs better. The project would conclude with which 
design is more appropriated for a given usage.

Lastly H. Hinnant has added on his web page some date algorithms that 
would be good to adapt to boost and make my prototype use them.
> What are some mandatory prerequisites for the project ?
>
>
Have a good understanding of the problem domain and the performance that 
are required to such a Date library. Knowledge of the papers related to 
a possible Date library for the standard.
Be familiar with the prototype design and code.
Be confident with the Boost development environment. Moving the SVN 
repository to an initial Git repository following the new modular 
structure and been able to run the regressions test would be a good 
starting point.

Best,
Vicente

P.S. Please contact me directly if you need more details.