Re: fit function norm indices.

From: John Houck <houck_at_email.domain.hidden>
Date: Fri, 24 Dec 2004 06:15:01 -0500
On Mon, Dec 13, 2004 at 14:02 -0500, John E. Davis wrote:
> David P. Huenemoerder <dph_at_email.domain.hidden> >  isis> define dummy_fit(){ return 1; }
> >  isis> add_slang_function("dummy", ["a","b","c"], [0:2] ); 
> >  Norm indices must be numbered from 1 <= n <= N
> [...]
> >I think the help file should state that the indices are NOT array
> >indices, but the logical parameter indices, starting at 1:
> Alternatively, the interface could use the 0-based array indices for
> the specification of the normalization parameters, as you originally
> assumed.  In fact, I think that this is better anyway since using the
> "parameter indices" as given by list_par assumes something about how
> isis determines those values.  IMO, it is better not to expose this
> internal convention as part of the interface.

The indices don't assume anything about how isis determines
those values -- they're simply 1-based indices.  However,
because S-Lang arrays are 0-based and because the mechanism for
setting default parameter values uses 0-based indexing, I agree
that it makes more sense to use 0-based indexing here.

The next release (isis-1.2.5) will use this behavior.

Received on Fri Dec 24 2004 - 06:15:01 EST

This archive was generated by hypermail 2.2.0 : Thu Mar 15 2007 - 08:45:50 EDT