Re: FW: tell us about what you'd like to do for PSPP

[Ben Pfaff]

"Ronald Szoc, PhD" <address@hidden> writes:

> Thank you... I downloaded the code (but not from Savannah, which has the
> most recent, so I may do that again).... John suggested trying out some
> procedures and finding bugs, and then quashing them. That is one way to get
> familiar with the code. 

I'd really recommend using the Savannah CVS.  Anything else will
be very old and buggy.

> I also noticed that some of the notes in the code say things such as
> "spearman's rho is not computing correctly" or something similar... 

Some of that is fixed in CVS.  Some is not (particularly in
CROSSTABS, which needs to be rewritten anyway).

> What I would like to do initially, is to get very familiar with the code. If
> PSPP is still a console app, then I think we need to discuss how to make it
> a GUI app. (I myself took a long time to get to GUI, and when I tell people
> - when I'm trouble-shooting - "I want to get at the command line", most of
> them now say: "Why????"... )... 

You're not the first to suggest a GUI.  A couple of people have
come up with prototypes but I don't know if any of them did
anything--I think at least one was just pretty sets of menus that
weren't "hooked up" to actions.

You're welcome to build a GUI if you're interested.  At one point
I had very specific ideas on how a GUI should be done but I'm
willing to settle for anything now.

If you build a GUI, I hope you can coordinate with anyone else
who is interested.

> One of the things that SPSS does (I suspect, don't know for sure) is to
> generate a series of command line statements based on GUI menu choices (
> strange but true, and kind of an elegant solution to a problem, no?), and
> then processes that command line. 

I know that's the way older versions of SPSS and SPSS/PC+
worked.  I haven't played with any recent versions of SPSS though
so I don't know if it still works that way.

That would be fine.  It's actually a nice way to go in some
respects because it means you can easily record an analysis and
then replay it later on different data.
-- 
Ben Pfaff 
email: address@hidden
web: http://benpfaff.org