Re: handle graphics

Bill Denney <[email protected]> Tue, 14 Feb 2006 16:40:15 -0500 (EST)
Newsgroups gmane.comp.gnu.octave.maintainers,gmane.comp.gnu.octave.graphics
Message-ID <[email protected]>
On Tue, 14 Feb 2006, Michael Schmid wrote:

> In matlab, there can be changed each line in a plot. I mean really each 
> line.. also the short lines in a legend can be moved... it is an 
> exhausting work, but it can be done! Each special plot can be programmed 
> in m-files, although a lot of the functions aren't written in m-code ( I 
> think in R14 there is a lot of Java code ... very poor 
> performance..!!!). The first question, if anything in a plot can be done 
> with m-files coding... is there a need to understand the underlying 
> system (if it is written in c++) ?

For users, there is no need to understand the underlying system, but for 
the intermediate programmers (like myself), I like to understand the 
underlying system.  I've done some rather heavy work on matlab graphics 
(helping with the plot2svg utility: 
http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=7401), 
and I didn't need to understand the underlying system because I can do it 
in m files.  I know that I will help

> Out there are two sorts of people, users and programmer... the users 
> aren't interested in "how is it done". They want to know "what can I do 
> with it". The programmers, should know what c/c++ is! Even if I'm better 
> in m-file coding as in c++, I would write the "handle graphics system" 
> in c++.

You leave out the class of the power users who aren't quite system 
programmers but are able to work on significant parts of the code.  I 
think that you discount their number and impact.

> 6 years ago, training and simulation of neural networks were faster than 
> today ... today is everything written in m-files, even the 
> Levenberg-Marquardt algorithm. To analyse this m-files is to difficult 
> and need to much time.. so I write now Levenberg-Marquardt in C++

I believe that the general position is that if code is likely to be used 
often (like in a nested for loop) and the speed increases are significant 
then C++ is warranted, but if it's something that's likely to be used 
sparingly then m files are easier to edit/write.

> If a person say "I would help, if it isn't in C/C++", is this person 
> able to understand this m-files??? I don't think so ...

I'm a relatively regular contributor to octave and I fall in the category 
of "I understand m-files, but not C/C++".

Bill

-- 
"I want to be Robin to Bush's Batman."
   -- Dan Quayle