[prev in list] [next in list] [prev in thread] [next in thread] 

List:       koffice-devel
Subject:    Re: Got accepted for Google's Summer of Code, yay ! :D
From:       Tomas Mecir <mecirt () gmail ! com>
Date:       2005-06-26 12:55:51
Message-ID: 492258b105062605553ba9aaab () mail ! gmail ! com
[Download RAW message or body]

On 6/25/05, Inge Wallin <inge@lysator.liu.se> wrote:
> Congratulations!!
Thanks ! :)

> Can we read your proposal anywhere?

Sure, in the attachment of this mail :D

/ Tomas

["proposal.txt" (text/plain)]

Synopsis: Improve computation engine of KSpread

Goals:
This project will focus on improving the computational engine, making
it comparable with engines used in other spreadsheets, hopefully even better.
With these improvements, KSpread should become useable for more serious
computations, hence increasing possible user base.
I have been already contributing to KSpread for some time, and I'd like
to base this work on things that I already wrote in the past.

Project Details:
I would like to implement the engine using the following architecture:
- all basic operations would be encapsulated in a dedicated set of classes
- this approach will allow us to transparently use various data types, without
  having to write special functions for each of them
- the engine would use the formula parser, which already exists in the source,
  but is not used yet
- there would be a dedicated class for managing and calling functions
- this class would be able to handle automated parameter conversion and loop
  constructs
- loop constructs will allow easy implementation of array formulas, which are
  important for more serious computing
- by placing loops to the formula manager, the functions themselves will not
  have to know anything about array formulas
- this will also allow very easy implementation of basic matrix manipulation
- also, all existing functions will be converted to use the new function
  manager, and they will also make use of the transparent datatypes described
  above
- next, displaying of arrays/matrices will need to be added - I already have
  an idea how this could be done easily
- if time remains, I would like to make use of the aforementioned datatype
  abstraction, by introducing arbitrary precision support (that means,
  allowing the user to specify precision higher than long double). This
  will use some specialized library such as GnuMP.

Project Schedule:
I expect the project to take approximately 2 monthes.

Motivation:
I have two different motivations. First, I am interested in spreadsheets, and
I would like to improve the KDE's one. Second, I like to experiment with new
ideas, and this project offers a lot of interesting possibilities.




_______________________________________________
koffice-devel mailing list
koffice-devel@kde.org
https://mail.kde.org/mailman/listinfo/koffice-devel


[prev in list] [next in list] [prev in thread] [next in thread]