st: re: different result everytime - something strange

[Kit Baum <baum@bc.edu>]

< >

Re my last posting, apologies to Philippe: he did indeed mention - 
adoupdate- in his posting, which I read WBOC*. Let me just clarify  
what I said about that routine to indicate why it did not ultimately  
solve the problem. If you do -findit glcurve- you find six references  
to the SJ and STB-published forms of this routine, most recently  
updated in SJ 7:2. You also find references to the SSC version of the  
routine (dated 20070216) and a version presented at the 5th UK SUG  
meetings (which was, as it happens, a very long time ago as this  
year's meeting will be #14).

The general problem, as people smarter than me have concluded, is  
insoluble by artificial intelligence. There is no constraint that  
causes a particular SSC package to be coincident with the contents of  
a particular SJ package. For instance, we may have package -zorch-  
which contains zorch.ado, zorch.hlp, foo.ado and foo.hlp accompanying  
an article in the 2007 Stata Journal. The SSC archive may contain two  
packages -- zorch, with only the zorch files, and foo, with only the  
foo files. The zorch package on SSC dates from 2006; the foo package  
was updated in 2008 (hypothetically). What should -adoupdate- do?  
Programmers may argue that this could be solved with makefiles, but  
that requires a great deal of information available to such a  
facility (for instance, there is a modification date on each SSC  
package, but not on each included file).

For -glcurve-, as the SSC package dates from 2007 and the SJ package  
dates from 2007, there is no substitute for manual examination of the  
headers of each ado-file, in the hopes that the authors have given a  
version number that allows you to conclude that one is a later  
version (or, perhaps, that they are the same version).

Kit Baum, Boston College Economics and DIW Berlin
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html

*: Without Benefit of Caffeine
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/