[Scilab-Dev] Two SEPs

[javier ignacio carrero mantilla]

Well, I understand your analysis, but it precisely points to the fact 
that such change in Scilab would be major and required to rethink the 
whole process of variable creation (adding linking). In other words it 
requires to rewrite the core of Scilab, that's why I wrote it as a 
proposal, as I'm merely an amateur and I don't have the knowledge or the 
resources to do so.

El 2010-12-16 03:20 a.m., Bruno JOFRET escribió:
> Hi there,
>
> Having a look at SEP_47, i should add some inputs about scilab behaviour.
> Scilab a is numerical computation software. It's quite the opposite of 
> a "symbolic/formal computation software".
>
> The difference stay in the way "operations" are stored.
> Symbolic :
> c = a + b
> the software will remember c is the result of a and b addition.
>
> Numerical :
> c = a + b
> The software will remember only the value of c. And not how c has been 
> computed.
>
> In other words, we can not "link" variables as you want it to be 
> because the information is already lost. We are only manipulating values.
>
> The only solution you have is to manage special symbolic variables.
> Then override each operator, aso, aso...
> To put it in a nutshell write a dedicated symbolic module with 
> dedicated type and operations :-S
>
> Hope this help.
>
>
> On 12/16/2010 01:19 AM, javier ignacio carrero mantilla wrote:
>> Dear Sylvestre:
>> Thanks for your comments, I will check SEP_47. About the linking
>> variables, I think it should be incorporated to the Scilab core because
>> the new behavior associated with linked objects cannot be described in
>> terms of the Scilab's operators in their current form. But anyway
>> backwards compatibility (i.e. the future "linking" Scilab will be able
>> to interpret the current scripts and variables) is guaranteed as a
>> non-linked variable is a particular case of the linking process.
>>
>> Sincerely yours
>> Javier I. Carrero
>>
>>
>>
>> El 2010-12-15 10:46 a.m., Sylvestre Ledru escribió:
>>> Hello,
>>>
>>> Looks like pretty cool :)
>>>
>>> The SEP about the Scilab environnement, we are planning to implement it
>>> in Scilab 5.4.0. See SEP SEP_47_Scilab_Desktop
>>> However, your idea to be able to save it from the Scilab language is
>>> pretty good.
>>>
>>> About the "linking variable", is it interesting to have this into the
>>> Scilab distribution itself?
>>> I wonder if a module would not be better.
>>>
>>> Anyway, many thanks!
>>> Sylvestre
>>>
>>>
>>> Le mercredi 15 décembre 2010 à 10:28 -0500, javier ignacio carrero
>>> mantilla a écrit :
>>>> Hi
>>>> In this message I have included two enhancement proposals for Scilab,
>>>> related with the environment and the linking of variables.
>>>>
>>>> Sincerely yours
>>>> Javier I. Carrero
>>>
>>
>>
>
>