Re: st: exact row in which error occurs

[Daniel Feenberg <feenberg@nber.org>]

On Thu, 24 Nov 2011, Nick Cox wrote:

> Stata doesn't work in this way, partly because do-files can and
> usually do call other do-files, ado-files, commands, functions, etc.,
> and a row number -- presumably meaning file line number -- would not
> have very much meaning in many instances, and/or would refer to a line
> in code that is proprietary and so hidden from you.


That would be called a "traceback" and I remember it in IBM Fortran G 
before GUI debugging displaced it. The traceback would look something 
like:

   illegal command at line 37 of subroutine SUB2

   called from line 21 of subroutine SUB1

   called from line 15 of MAIN

Other compilers gave additional information for each call, such as the 
value of scalar variables. These were very useful in debugging and had no 
runtime CPU overhead. (There was some memory used). The alternative with 
Stata is to -set trace on-, but sometimes that produces a prodigious 
amount of output. As it is, it is often time consuming to associate a 
generic error message such as "no variables found" with a particular (a)do 
file statement in a long program.

Daniel Feenberg
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/