On Apr 4, 10:48 am, Laszlo Nagy <[EMAIL PROTECTED]> wrote: > Hello, > > I have a win32 service written in Python that starts a plain > application, written in Python. > > The win32 service tries to launch the application in a while loop and > logs the return value of the os.system call. That's all. > > The application is a simple Python program that connects to an https > xml/rpc server, and works with the data retrieved from that server. It > is written as an application because it must be able to download updates > for itself. Here is how it works: > > a.) connect to server > b.) compare current version with the latest > c.) if there is a difference, then download all sources from the server, > delete all pyc files and exit; otherwise start processing > > I could not write a self-restarting server, and I did not want to > write/install two services for one problem. The win32 service is very > small and primitive so probably I will not need to update it. I think > the basic idea is good, but... > > When there is a client update available, my application updates itself > cleanly and exists. Then the service tries to restart the application, > but it cannot. os.system returns with OS error code -1. The pyc files > are NOT generated for the application. However, if I start the > application from the command line, then pyc files are created, and then > the service will also start the application immediatelly. The win32 > service is running as "Localsystem" so it is sure that it has write > permission on all files. > > I cannot log out the error from the application since it is not started. > The only error message I have is OSError -1, but it tells me nothing > about the nature of the error. > > Thanks, > > Laszlo
Have you tried the subprocess module for this rather than os.system? You might be able to pipe errors to a file with it. You might be able to use the traceback module for more verbose error catching too. Mike -- http://mail.python.org/mailman/listinfo/python-list
