[Scilab-Dev] vector^scalar allowed again in Scilab 6?

[Samuel Gougeon]

Hello Pierre-Aimé,

Le 10/08/2016 17:56, Pierre-Aimé Agnel a écrit :
>
> Hello,
>
> It was changed at resolution of 
> http://bugzilla.scilab.org/show_bug.cgi?id=14316
>
> I am not sure though whether enforcing the 'vector ^ scalar' error 
> should be reintroduced. My 2-cents: there are many operations that 
> have the same behaviour that their dot-operation counterpart when used 
> with a scalar.
>
> I think the operation should remain available, but I can also enforce 
> the obsolescence in the release for 6.0.0.
>

Thanks for your answer.
I understand that the reversion of the vector^scalar change was done 
when fixing #14316,
but as far as i understand, things are not connected. This reversion was 
not mandatory to fix #14346.

IMO, what was planed for Scilab 6 was great.

I know that this syntax vector^scalar instead of vector.^scalar was very 
widespread. But instead of parsing vector^scalar as vector.^scalar as in 
Scilab <6 or always yielding an error, why not making Scilab 6 testing 
whether the overload %s_p_s() (or other%@_p_@() overload according to 
operands types) exists. Then, users wanting to work with the old 
equivocal behavior could define deff("r=%s_p_s(a,b)","r=a.^b") in their 
library (or startup file) instead of changing all their code. Or/and 
include their own warning in their overload whether they wish to 
actually upgrade their code. And that's it.

Best regards
Samuel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/dev/attachments/20160810/520a375c/attachment.htm>