[Spambayes] Whitelisting

[Mark Hammond]

> I am thinking about adding this feature myself, so I am just 
> trying to assess 
> 
> 1) If I am wrong thinking Address Book Whitelisting is a good thing.

Most of us share Tim's opinion - no objection, but also no real desire.  I
*may* have used it once or twice when I was expecting a "new" email that I
feared would be seen as spam by SpamBayes (as nothing like it had been
trained on yet).  This almost always turns out to be unfounded though -
SpamBayes gets it right most of the time.  The fact someone knows my name is
"Mark" can be a huge clue :)

> 2) If it would be hard to add to the curent codebase.

There are 3 main things that need to be done:
1) Determine the list to match against.
2) Design a UI that configures it (even if as simple as a checkbox!)
3) Match the addresses as they come in.

3 should be simple (assuming 1).  

1) is a little tricker - I see 2 basic options - match against address books
etc, or match against our own list we maintain.  Matching against address
books isn't as simple as it sounds in an Outlook/Exchange environment, as we
have the Global Address List etc on the servers.  "Address Books" are
different in an exchange environment than "Contacts" are in a pure Outlook
environment.  Performance implications may force us to cache or pre-read the
addresses we want to match against, rather than asking Outlook to match per
incoming mail, etc

The "easy" solution here would seem to be to keep our own list of addresses,
and provide an 'Import' feature, where we pre-populate the list of
addresses, but maintain the list ourself.

Which leaves us with (2).  Assuming we take the 'easy' option above, the UI
becomes harder.  We will need to provide a UI for managing the list itself
(add/delete/import names etc - presumably via the "SpamBayes Manager"
dialog).  We would probably also want an "Add user to Whitelist" option in
the "Spam" and "Spam Maybe" folders.  Or should it be "Recover from Spam,
and add user to whitelist"?  If so, how do we fit that on a button <wink>?

So, unfortunately, it gets pretty hard, pretty quick.  If we could assume
the UI was "edit the text file in your favourite text editor", it would be
alot simpler, but I fear we are already past the point where this is a
reasonable option given our users.

I hope this doesn't put you off though - I'd be happy to help with the
tricky stuff, as long as someone is chasing up all the edges and actually
driving it :)

> I took a look at the binary distribution and it is...  well, 
> binary :-)

Not if you look a little deeper :)  Most of the code exists as .pyc files in
the .zip file.  However, just run from the source-code if you want to fiddle
with it.

> Are you guys using a Win32 Python compiler to produce this output?

We are using py2exe - find it at sourceforge - but really it is just a very
smart "packaging" of the Python code, rather than a compiler as such.

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 3320 bytes
Desc: not available
Url : http://mail.python.org/pipermail/spambayes/attachments/20040413/8a9735c6/winmail.bin