I too have just tried the ODBC tools and noticed problems when importing
data (from MS Access). I also found the padded string problem (and strange
error messages when I run out of memory due to it).
I have also found that when I load a table it often has errors -- a
quick -cf- command on the loaded table generally does not agree with a
version of the same table imported through Stat/Transfer. There are
problems with missing values (there are almost no missing values in the odbc
loaded table), variables disagreeing on just a few values, and variables
becoming completely corrupted (99% of the data turned to missing and the
remainder all 0). When I have repeated the same transfer, the problems
change and a different variable may be similarly corrupted.
I'm not sure of the source of these problems, but I would be quite careful
using any data loaded through the ODBC utilities for now.
Michael Blasnik
[email protected]
----- Original Message -----
From: "Grochulski, W. Daniel" <[email protected]>
To: <[email protected]>
Sent: Tuesday, January 21, 2003 12:52 PM
Subject: st: Stata 8 - ODBC utility
> Dear Listers,
>
> While happily navigating my freshly installed Stata 8 (running under
> Windows 2000) I was stopped by seemingly odd behavior of the ODBC utility
> (one of the goodies in Stata 8). Loading tables via -odbc- is just a snap,
> there is, however, something wrong with the way the string variables are
> handled - my Stata consistently adds a 127 blank characters pad to any
> downloaded string variable, so i.e. 20 characters string is loaded and
> reported as having 147 characters (str147). This happens with both
> SQL-server and MS Access databases - below I reproduce results of the load
> from "Northwind" database to be compared with printouts given in Reference
> Manual for -odbc-. Since all these 'strxxx' strings are longer than 80
> characters Intercooled Stata can handle, this dataset is exploding into
> user's face when trying to -list-, -drop- or -keep- (Stata crashes and
> exits). One can of course extract the original strings by generating
> properly defined new string variables using substr function, but this is a
> bother (and you have to keep the 'oversized' string variables anyway).
>
> Any suggestions?
>
>
> Daniel
<snip>
*
* 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/
