Re: PKG: miscellanoues

[Olaf Till]

On Mon, Apr 15, 2013 at 02:05:16AM +0100, Carnë Draug wrote:
> On 13 April 2013 19:12, Juan Pablo Carbajal <address@hidden> wrote:
> > I think the functions *poly (as in legendre, chebyshev, etc...) in the
> > miscellaneous package (1.2.0) should be merged with their not
> > completely equivalent specfun package siblings or, even easier just
> > move them over cause there is no naming conflict.
> >
> > What do you think?
> 
> I have another thought about the miscellaneous package. Removing it
> completely. Miscellaneous functions is the type of thing that belongs
> to Agora. Or in core if they're useful.
> 
> Here's my suggestion:
> 
> 1 - good useful pieces of code - move to core (publish() is part of matlab 
> core)
> 2 - functions that fit in another package - move to that package
> 3 - the rest - move to Agora as individual functions or small bundles
> (I think some could actually be turned into sections of the Octave
> cookbook)
> 
> Carnë

There is at least one difficulty with this: read_options.m, an option
passing system by Etienne Grossmann. It is used by his optimization
functions within optim and by his vrml package. Since recenently,
publish.m in miscellaneous uses it, too. Most other optimization
functions, in particular some in Octave core, have adopted the
optimset mechanism instead. Moving read_options to Octave core would
require a discussion on whether a different option passing mechanism
is reasonable, and if yes, on possible changes and refinements of this
mechanism. My personal opinion is that though optimset (et al.) might
not be ideal, it is no use discussing it since it is Matlabs mechanism
and also already established in Octave, and it would be an unnecessary
complexity to maintain a second option passing mechanism. But 3
different packages currently depend on read_options.

Olaf

-- 
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net

signature.asc
Description: Digital signature