[Spambayes] Mulitple SpamBayes installs on a network

[Dan Burgin]

Right, the desktop runs Outlook with SpamBayes continuously, while the
laptop comes and goes - meaning that at times both machines would be running
Outlook with SpamBayes.

-----Original Message-----
From: Mark Hammond [mailto:mhammond at skippinet.com.au]
Sent: Thursday, October 09, 2003 6:36 PM
To: 'Ryan Malayter'; Dan Burgin; spambayes at python.org
Subject: RE: [Spambayes] Mulitple SpamBayes installs on a network


[Dan]
> Yes, it can be done... Everything SpamBayes uses is stored by 
> default in
> the "Application Data" section of the user profile on each individual
> machine. Stuff in the Application Data folder here gets copied to
> servers with the rest of the user profile so it can follow the user
> around the network.
> 
> If you use the SpamBayes INI file settings to move your DB to the
> "C:\Documents and Settings\username\Local Settings\Application Data"
> area of the user profile, you will get a different SpamBayes 
> DB on each
> machine, and you won't have to worry about them overwriting 
> one another.
> Files in this "Local Settings" section of the user profile do not roam
> to the network like the rest of the user profile in a Windows domain.

Everything you say is correct, but it may not solve the problem.
Specifically:

> > given this setup, can I install a copy of SpamBayes on a desktop
> > machine which runs Outlook Client (and SpamBayes) 24/7, yet still 
> > run my laptop (on the LAN, or via VPN) with a seperate copy of 
> > SpamBayes (allowing me to "train" emails that mis-filter) without 
> > messing up my original instance of SpamBayes on the desktop machine?

If I understand it, the desktop is *always* running Outlook, even while the
remote machine is.  This is likely to be a problem:
* If you stick with the default settings, the second version of Outlook
running will be unlikely to be able to open the database for writing -
therefore, training can not be done by the second instance.
* If you change the data directory, then both instances will be able to
train - but will have completely different databases.  Training done on the
remote machine will not be reflected back to the main machine's training
database.

Neither of these really seem suitable.

I believe we could change things to make this work, and indeed, it may fall
out of our investigations of the strange DB errors we are seeing - but at
this point, we would need some help to make it happen in the short term.

Mark.