Without SOME method of abbreviating variable names, it probably would not be
possible to use any of the list type data commands, like renaming groups of
variables.
That said, being able to turn off abbreviations in analysis commands (as
opposed to data generating commands) might be a good idea, since tracking
down a mistake presumes one is alert enough to notice that something is
amiss...
Bryan Sayer
Statistician, SSS Inc.
[email protected]
-----Original Message-----
From: Glen Waddell [mailto:[email protected]]
Sent: Tuesday, February 11, 2003 1:07 PM
To: [email protected]
Subject: st: RE: RE: Referring to a varname, leading to errors
I suppose the quick reply is, bummer.
There is certainly a place for abbreviated command names. My example
with respect to varnems, however, is intentionally simple. One must
recognize that this issue can lead to costly down time when mistakes are
made and must be tracked down. Abbreviated varnames just doesn't seem
worth it, and clicking on the varname doesn't help when running
do-files.
- Glen
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Nick Cox
Sent: Tuesday, February 11, 2003 1:33 AM
To: [email protected]
Subject: st: RE: Referring to a varname, leading to errors
Glwn Waddell
>
> This is one of those problems that only arises out of human error.
> Nonetheless, Stata is set up to allow this mistake to go unchecked.
> Consider the following example...
>
> Suppose a variable currently exists with the name "abcdefg" and one
> wants to condition on a different variable, abcde, such that one runs
> a command only for abcde == 1. If abcde has not been created
> (i.e. human
> error, the creation of this variable having been overlooked), the
> command "reg y1 x1 x2 if abcde == 1" would actually be run
> conditional
> on abcdefg == 1, not abcde == 1.
>
> I suppose the idea behind allowing shortened varnames is one of
> convenience. However, this seems like more trouble than it is worth.
> Is there a way to toggle between allowing such shortened
> names and not?
> A permanent fix would be even better.
>
We can argue over whether allowing abbreviated
varnames -- or indeed abbreviated command names --
was a design mistake, and when the feature bites
you, you certainly think it may have been.
However, this example does not seem that scary.
A habit of clicking on variable names in the Variables window
can help in this circumstance.
In general, Pandora -- or more precisely Bill Gould
-- opened the box many years ago, and I don't think that
act can be reversed. I don't think there is a completely
general way of setting Stata to acting if and only if
variable names supplied are the unabbreviated versions. There
are solutions to specific problems, e.g. you can write
a program to -drop- what you specify if and only if
what you name includes complete variable names.
-drop-ping something important by accident
can be an issue.
It's not your question, but I'm reminded that some years
ago someone suggested that Stata programmers should not
be allowed to abbreviate command names, because it made
programs more difficult for beginners to understand, so
that every program should only include full command names.
That didn't get any support from programmers.
Nick
[email protected]
*
* 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/
*
* 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/
*
* 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/
