We have a problem concerning the compilation of Mata-files. We have
implemented an application “lsq.mata” that consists of some new
Mata-functions and some self-defined Mata structures. The application is
invoked by an Ado-File called -sqom.ado-.
We have compiled lsq.mata and created the file “lsq.mlib”. When we execute the
program directly after compiling of it, it works properly. However, under the
conditions we describe in a minute, it does not. Under these conditions we
get an error massage telling that Stata cannot find one of the
Mata-functions that are contained in the library. Error message is either
hashing(): 3499 elemlist() not found
sqomref(): - function returned error
istmt>: - function returned error
r(3499);
or
<istmt>: 3499 sqomref() not found
r(3499);
where elemlist() is a constructor-function that Mata provides for
Mata-structures, and sqomref() is Mata-function defined by ourself in
lsq.mata. Unfortunately, we have not been able to find a regularity in
occurance of the upper or lower error message, so far. Most of the times it
is the upper (The second only appears on a Windows computer, and we are not
able to reproduce it anymore) .
However, the following listing contains some situations we have tried out:
0) After using lsq.mlib on computer A that was compiled on computer B.
Compiling lsq.mata on computer A temporarily solves the problem, but ...
1) After restarting Stata:
. do lsq.mata
. use xyz
. sqom // -> Program runs properly
. exit
(Restart Stata)
. sqom // -> Program does not run
2) After using -clear- as a command:
. use xyz
. do lsq.mata
. sqom // -> program runs properly
. clear
. use abc
. sqom // -> program does not run
or:
. use xyz
. do lsq.mata
. sqom // -> program runs properly
. clear
. use xyz
. sqom // -> program does not run
But:
. use xyz
. do lsq.mata
. sqom // -> program runs properly
. use abc, clear
. sqom // -> program runs properly
. use xyz, clear
. sqom // -> program runs properly
In all cases, the compilation of Mata-file with –do lsq.mata- solves the
problem and program works perfectly afterwards. It seems that the program
runs as expected after compilation as long as we do not "clear" Stata with a
the command clear, or by exiting it.
We do not see an explanation why compilation of “lsq.mata” solves our problem.
Should we continue to search for the cause of the error in our source code or
is there something that we are not aware of? Is it possible, that our problem
is connected with the process of compilation and creation of libraries or
interpretation of the compiled code by Stata?
We know that it is not easy to solve this problem without having access to
the source code. The source code is, however, a bit long, so that we don't
want to put it in this email. If someone likes to see the source code, please
contact us directly. However, maybe there is something trivial that we do not
know.
many regards
Uli and Magda
--
[email protected]
+49 (030) 25491-361
*
* 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/
