[Spambayes] 2 Copies running at the same time?

[Tony Meyer]

> I was talking with someone and they wanted to run outlook on 
> a machine and have it on all of the time and have the 
> spambayes plugin loaded so that they didn't have to wait for 
> all of the spam to get moved when they started outlook. But I 
> believe that the score information is local to the machine.  
> This is using exchange by the way. The problem I see would be 
> that things that didn't get caught by the first machine would 
> be trained on the second machine and so training might 
> degrade on the first machine.
> 
> Just wondering what people think about the 2 copy situation?

It really depends on what sort of training you want to do.  As long as
you're only training on one machine, it should be ok - basically you
wouldn't even install SpamBayes on the second machine.

You can set the data_directory (where the training data is stored) to a new
location, like a shared drive, but we really don't recommend having more
than one process access the databases at one time - you're just asking for
the database to get corrupted.

Could the second machine just run Outlook all the time rather than the first
one?

Maybe what would work is to only have SpamBayes on the first machine, and
have it set to use 'incremental training' (Training tab in the Manager) on
the spam folder.  On the second machine, whenever you want to train as spam,
don't use the "Delete as spam" button (it wouldn't be there), just move the
message to the spam folder (you can do this sort of thing with OWA, for
example).  To train as ham, you move the message into one of the folders
that you're watching.  This should work ok, although it's perhaps less
convenient than the regular buttons.

> Another thing that would be nice in the future would be proxy 
> support for updates.

I presume you mean the "Check for new version" option - this does work
through http proxies, but not when you're using "automatic proxy
configuration".  I looked into this a while back (there's still an open
tracker), and it is unfortunately very complex.  The location holds a
javascript file, so we'd have to parse that (or give it to a javascript
engine to do so), but it's not even that simple, because there are special
functions available that aren't generally available to a javascript script.

Note that we're not alone in falling to handle this - a lot of applications
I have only work if I turn off automnatic configuration and specify the
proxy specifically (which is what I now do, since the automatic script here
does very little anyway).  Since all we need it for is checking for a new
version, there doesn't seem to be much point in spending a lot of time on
this.  A better solution, which we might do at some point, is just open up a
web page (that will say what the latest version is) if the connect fails.

At some point, I imagine that support for this will be added to Python, and
so we'll handle it then (automatically, probably).  Looking into the Mozilla
source might give clues, since I believe Firefox/Mozilla handle it fine.

=Tony Meyer

---
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes. This
way, you get everyone's help, and avoid a lack of replies when I'm busy.