Page 1 of 1

Pre or post file PRE does not exist & Warning: getversion

Posted: Tue Mar 13, 2012 11:17 pm
by MartinOFlaherty
Hi,

I'm encountering two problems trying to make runmlwin work. I am using stata 11.2 and MLwiN 2.25, and have verified that runmlwin is up to date and fully installed.

As a general note, I'm specifying my MLwiN directory via the mlwinpath option on each estimation command rather than as a global macro.

The first problem arises for all models which I try to run - the model runs fine, but additionally returns the following message:
Warning: getversion plugin could not be loaded
I've followed the advice given in this topic: http://www.cmm.bristol.ac.uk/forum/view ... ?f=3&t=280 All of the tests suggested in that topic work exactly as described and indicate that the getversion plugin should work fine.

The second problem arises only for models estimated via MCMC - MLwiN returns the error message:
Error while obeying batch file C:\DOCUMENTS~\temp\ST_0600007e.tmp at line number 325: PREF: PRE

Pre or Post file PRE does not exist
I've followed the advice given here http://www.bristol.ac.uk/cmm/software/s ... rrors.html and checked that files PRE and POST are in the 'discrete' folder of my MLwiN directory (they are) and that the 'Fpath', 'pre' and 'post' options are specified correctly (they are)

Any suggestions? Help would be much appreciated.

Regards,

Martin

EDIT: I tried specifying my MLwiN directory via global macro and it fixed both problems, although I don't understand why exactly - will leave the topic here for anyone else who encounters the same difficulties. For anyone who (like me) wasn't sure how to use the global macro simply enter in stata's command window (editing to fit your MLwiN filepath obviously): global MLwiN_path C:\Program Files\MLwiN v2.25\mlwin.exe

Re: Pre or post file PRE does not exist & Warning: getversio

Posted: Wed Mar 14, 2012 12:56 pm
by ChrisCharlton
Thanks for reporting this, it appears to be a bug. The contents of the global macro and the command option should be stored in a local macro, and then this used within the command. It appears that in some places, such as discrete models, we are using the global macro directly. I will fix this for the next release.