Re: Suggestion

[Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr>]

On Thu, Oct 11, 2007 at 05:24:43PM +0200, Martin Larsen wrote:
> 
> Hi
> 
> I have a suggestion for a new feature, but first I will mention a few problems with GP/PARI CALCULATOR Version 2.4.1 (alpha) on Windows.
> 
> * a script loaded at startup will not have it's local()'s recognized (they become global)

Could you give an example ? local() are only supported inside functions
in GP 2.4.1.

> * a script with syntax error loaded with \r will prevent any new attempt of loading it.
> 
> * flag==4 in vecsort produces an error

A GP 2.4.1 specific bug which was a pretty buggy development release,
alas! Try out GP 2.3.2 which is much more solid.

> And now a little funny example:
> 
> > k=1;vector(6,i,[k,k=k*10%7][1])
> %1 = [32, 3, 32, 6, 32, 5]

This is a well-know bug with the memory handling: GP forget to copy
the first k in the vector. This is fixed in the forthcoming GP 2.4.2
I recorded it as bug #686.

> > k=1;vector(6,i,[k*1,k=k*10%7][1])
> %2 = [1, 3, 2, 6, 4, 5]
> 
> I think this is a nice way if you don't want the result from the last expression(?)
> Only 2. example works as expected, but is the method safe?

Actually yes, but you could just do
k=1;vector(6,i,l=k;k=k*10%7;l)
which is safe.
 
> Now 2 suggestions:
> 
> > vector(6,i,1)
> %3 = [1, 1, 1, 1, 1, 1]
> 
> I think it would be nice to be allowed to exclude unused locals in vector and matrix.

This would clash  with the current interface design: there is no
polymorphism.

> And last - Wouldn't it be nice if we had a standard-script that
> followed Pari with lots of extra functions, where functions could be
> loaded as last resort on demand?

What we need is standardised way to write and publish library of GP
functions. Would you like to try ?

Cheers,
Bill.