On Tue, Feb 28, 2012 at 10:51 AM, Riccardo (Jack) Lucchetti <r.lucchetti@univpm.it> wrote:
On Tue, 28 Feb 2012, Lee Adkins wrote:

Hi all,
Using lists is a great time saver in gretl.   I like using lists a
lot. It's still a bit hard to use them in the GUI, though. So.... I have
two feature requests.

 1. I'd like to be able to choose a list in the specify model dialog
 boxes in the GUI.

Oh yes.

 Also, it would be nice to have a button in the specify model dialogs
 to take you to the define list dialog.

Better still: select variables from the main window via shift-click, ctrl-click or whatever, then right-click and choose "make this a list"

Excellent idea!

 2. I'd like the ability to specify variables to be included in the

 modtest --breusch-pagan test statement. (nothing to do with lists, but what
 the heck).  Something like

       modtest z1 z2 --breusch-pagan

       or even better

       modtest zlist --breusch-pagan

I like this: however, there's a syntax problem; not unsolvable, but a bit annoying. At present, modtest has one optional argument (a scalar), mainly used in a time-series context. Allowing for a list makes syntax parsing a bit more complex. But then, a nice opportunity opens: how about replacing "add" and "omit" with "modtest <list> --add" and "modtest <list> --omit", respctively? Clearly, this would be massively backward-incompatible, so obviously it would be a 2.0 thing; still, in the meantime we could keep the two.

I like this one too.  It fixes the add/omit 'problem' -- as in what to do with them.  I wonder if a more flexible version of modtest that could take more inputs would help?  To maintain backward compatibility it would have to have a new name -- definitely a version 2.0 thing, though.  On the other hand, it would be confusing trying to figure out which to use... so maybe not.... 


Riccardo (Jack) Lucchetti
Dipartimento di Economia
Universitą Politecnica delle Marche

Gretl-devel mailing list

Lee Adkins
Professor of Economics