[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

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

From: Ronald Szoc, PhD
Subject: FW: tell us about what you'd like to do for PSPP
Date: Sun, 2 Jan 2005 20:29:30 -0500

I am sending this again, because the first time, I just responded to Ben's
e-mail, and not "to all". So I am responding to the list to spread the info
around. And (like a lot of programmers), I've changed a few things for the
sake of clarification.... 


-----Original Message-----
From: Ronald Szoc, PhD [mailto:address@hidden 
Sent: Sunday, January 02, 2005 4:48 PM
To: 'address@hidden'
Subject: RE: tell us about what you'd like to do for PSPP


I sent an e-mail via another facility within the savannah site to you, but I
guess you didn't get that one. 

A little bit of background on me: 

        I've been programming since 1968 (when I was a young lad) in a
variety of languages and operating environments: Fortran, IITRAN (Don't
ask), and PL/I (The best language ever, IMHO - but we won't go there), on
the IBM 360/370 under OS/MFT, OS/MVS, DEC PDP 8 Fortran and Basic, GE635
(Honeywell) in Fortran, Data General (Fortran), some specialized computers
at Bell Labs for analysis of waveform spectra, and most recently in the
IBM-PC Compatible world. (I topped out as CIO at a firm, and then left to
pursue other interests). In the PC world, I've programmed a lot in C, a
small amount in C++ (meaning creating objects and such, as contrasted with
just using a C++ compiler like most seem to ),  and a little in MASM. I've
created systems containing about 750,000 lines of code spread out among
about 50 subsystems in financial applications, as well as smaller systems to
do data transformations, etc. 

I also have a PhD in Psychology, and have been an active researcher, and
have taught classes at the undergraduate and graduate level in statistics
and research methodology. Most recently I gave a three hour workshop on
Logistic and Multinomial regression at the American Evaluation Associate
meeting this past November. 

In the past, I've given workshops on SPSS, and have used SPSS, SAS, OSIRIS,
BMD, and SOUPAC. (Don't worry if you have not heard of some of them... They
all had their day). 

In the good old days, when SPSS passed out the source code as part of the
software site license (this around the late 60's and early 70's), I examined
the source code, and then took a SAS routine (SAS also passed out the source
code) that calculated 10 different measures of correlation, converted it
from PL/I to Fortran, and then embedded it within SPSS so that it was
callable just as any other procedure. I saw first hand some of the tricks
that the first developers of SPSS did (I still have the manual from those
days, in fact). For example, why are the variable names restricted to 8
characters? Well, in those days, Fortran didn't have character variables as
such. So, the developers defined arrays of double floating point variables,
and filled them with 8 byte labels. Clever, no! Comparisons were a snap
because you could do a linear  search among floating point variables using
just the "=" operator. And how did they parse the rest of the command (the
part that was on columns of 16 through 80 of the punch card)? By overlay
defining (think unions and structs) an array of Logical*1 variables, and
then looking at them one at a time. Still very clever, IMHO. 

Anyway, these days I use SPSS 11.0 for the PC.

Given that, I would like to help in any way possible. For example, one note
that I saw (I believe on was that you wanted some help on
translating the formulae in the SPSS documentation into programmable
algorithms. I can help there. I can also help in writing the code, and in
designing (or perhaps re-designing) the PSPP architecture. 

For example, I saw another thread where there was a discussion of developing
a "library" of functions that encapsulate certain common statistical
operations (e.g., taking the sum of squares and the squares of sums), in
order to minimize code bloat. I could help there. 

And obviously, I would not use any SPSS code that I saw, because I hardly
remember any of it (it was about 75,000 lines of Fortran and about 10,000
lines of Assembly language). 

So, given my experience, where would you see me as being of most help? 

By the way, you can also google my name in its various forms (Ron Szoc,
Ronald Szoc, R Szoc, etc) and you will get an idea of my creditability in
applying your bozo filter. 

It just seems that with 4 people doing most of the work, and I assume that
everyone has a day job, progress must be slow. Another body would help
quicken the pace I would suspect. 

There is one area where I am deficient: developing under Linux. I downloaded
the PSPP source code, and noticed some file name extensions that I do not
understand, and certain written statements don't make much sense to me. Such
as: "Just use the ./configure call and develop a version for your particular
environment... " (That's from memory, so it may not be an accurate quote,
but I sure don't know what ./configure does because it doesn't exist in the
Microsoft/Windows world)...

On the other hand, I could do development for the Wintel world, if that is
of any interest for you. Or keep it within Linux... 

Anyway, hope to hear from you soon... 


-----Original Message-----
From: Ben Pfaff [mailto:address@hidden On Behalf Of Ben Pfaff
Sent: Sunday, January 02, 2005 3:29 PM
To: address@hidden
Cc: address@hidden
Subject: tell us about what you'd like to do for PSPP

Hi.  I see that you've applied for membership in the PSPP project at
savannah, but I haven't seen any email from you describing what you're
planning to contribute.  Can you tell us a little about it?



Ben Pfaff
email: address@hidden

reply via email to

[Prev in Thread] Current Thread [Next in Thread]