In the process of building a new login method for the Login Manager product, I thought it would be a good idea to have the beast log various indications of state out to syslog for later examination. ZLogger's interface is nice and easy and it looked like all I had to do to get logging to go to syslog was to set an environment variable. Unfortunately it just doesn't work that easily. Setting ZSYSLOG to log via /dev/log causes zope to crash straight away, setting ZSYSLOG_SERVER causes it to crash after between 10 seconds and a minute, and setting STUPID_FILE_LOGGER does nothing at all. Wading into the code I discovered a few things, and made some other observations: 1. There are no less than three separate logging systems used in Zope - logging via ZServer, logging via ZLogger and some form of debug logging via DebugLogger. They all seem to have different APIs. 2. ZServer and ZLogger both use a copy each of Sam Rushing's syslog module, but while ZLogger calls it properly, ZServer simply isn't. It doesn't appear to pass it a port at all. 3. Further up the line, the code that creates the logging object seems to pass ZSYSLOG straight through as a server address. I expect that a pair of patches to ZServer and z2.py will be able to fix the problem, but I'm interested if anyone has found any other solutions. John