[Scilab-users] Matlab vs Scilab perf

[Samuel Gougeon]

Hello Ricardo,

Le 02/03/2017 à 19:33, Ricardo Fabbri a écrit :
> Speaking from experience:
>
> It is worth mentioning that in many ways performance is not critical
> for a "lab" language like Scilab or Matlab. It is just an extremely
> simple language to test concepts and algorithms at a very small scale
> of granularity. The real crucial factor for Scilab or Matlab is the
> GUI for exploring data and developing algorithms interactively. Once
> you have a working solution, you'll fit it inside a bigger and more
> relevant
> system by porting promptly to a language like C++ for scalability and speed.
>
> Just use Scilab for what its worth, don't obsess with speed, even
> though it is important.

I agree: Matlab, Scilab, Octave, IDL, GDL.. Yorick etc were and are 
still first made for prototyping, not for speed. This is why they were 
back to interpreted -- and so slower -- languages.
But they should not have any handicapping snail instructions for common 
frequently used ones like scf(). Loosing time gained with an easy 
language avoiding to declare each object type and to compile and link 
the whole thing each time that we change a semicolon in the code... 
would be meaningless.

So yes, i definitely agree: it is a matter or ergonomic for GUIs but 
also and first for the language (regular namings, rational order of 
arguments, etc) that make it easy to learn and use. This is somewhat why 
i never really got inside python (and sciPy). I felt its syntax not 
straightforward to learn, even for simple things. Same thing for R. And 
it's always the same feeling when i try to compare results from these 
various languages -- including javascript --, for instance to extend 
Scilab or debug it, seeing usages or conventions the most used 
elsewhere. Each time, it is quite harder for me to find and understand 
the relevant syntax with Python and R, while for instance i find 
javascript more intuitive.
This is why i don't know much about Python. When Pierre writes "Python 
outshines Scilab/Octave/Matlab in many other areas.", i would be 
interested to know more about that.

I agree with you, Pierre, about the whole environment. This is why, on 
this aspect, we could have a specific discussion about which modules -- 
with which components -- should rather compose "Scilab core", and which 
features could be distributed rather on ATOMS. I am not convinced that 
the current "Scilab core" composition is optimal. IMO, it is quite 
unbalanced by historical or particular influences.

As a conclusion, i think that introducing non-optimal syntaxes or 
duplicates etc in the language hurts much more than introducing a quite 
slow algorithm. Simply because the algorithm can be changed later 
without breaking anything, while introducing badly designed syntaxes or 
usages is much harder to manage afterwards, and impedes much more 
learning, using and maintaining the language.

Best regards
Samuel