Sharing spam data between machines

[frisket]

I have all my email in TBird using IMAP, and I use TBird on that identical set of accounts from several machines.
Is it possible to periodically merge the junk training data between the installations, so that they will all have an up-to-date set of data to filter junk with?

[tanstaafl]

http://kb.mozillazine.org/Files_and_fol ... hunderbird
http://kb.mozillazine.org/Junk_Mail_Controls

Copy the training.dat file from one profile to another, copying it over the existing file. Long term you might need to use the Bayes Junk Tool from http://bayesjunktool.mozdev.org/ to examine and modify the raw training data. Sometimes it helps to get rid of tokens that are just as likely to occur in spam and legitimate messages, especially if the training data file gets very large. I don't know of a replacement (it hasn't been updated since 2005) but last time I tried it a couple of years ago the tool still worked. AFAIK the format of the training.dat file has never changed.

[frisket]

Thank you!

I'd like to merge the files between machines, as they have all had Junk buttons clicked for different incoming mail for a long time.

Is the file format described in documentation anywhere or do I need to get the TBird source code? I don't suppose it's in some kind of known format, is it?

[tanstaafl]

The only documentation I can find is https://developer.mozilla.org/en-US/doc ... _filtering , which says its based on http://www.paulgraham.com/spam.html . Your best bet might be the source code for the bayes junk tool since you'd have to search a smaller set of files and the author claims every Java method has a comment of some kind.

http://www.mozdev.org/source/browse/bayesjunktool/

[frisket]

Looks like the Bayesian Junk Tool might just do the job anyway, s it has a Merge function.

Unfortunately it doesn't create well-formed XML, so it'll need postprocessing until they fix that (unescaped less-thans and ampersands).

Many thanks.

[frisket]

OK, the tools merges perfectly without the need to export, so all I need to do is script it. Many thanks.

The tool is hosted on mozdev, so I'll report the bugs in the generated XML on there.

Peter

[frisket]

http://kb.mozillazine.org/Files_and_fol ... hunderbird
http://kb.mozillazine.org/Junk_Mail_Controls

[...] Long term you might need to use the Bayes Junk Tool from http://bayesjunktool.mozdev.org/ ...

While this solved the immediate problem, I wanted to report a bug in the XML, but I cannot log into their bug page with my regular Bugzilla username and password (it claims it's invalid) and I can't create a new account OR get a password reset because it is not sending me the email. I've tried using three of my email addresses but I never receive anything, nor does anything appear in my spam filter. I am concluding it is currently broken, but I cannot find anywhere to report this. Suggestions?

[tanstaafl]

You need a different bugzilla account ("Log in to Bugzilla@Mozdev") since the bayesjunktool is not a Mozilla product. Its actively used, but doesn't seem to have recent bug reports for the BayesJunkTool. The most recent seems to be from 2006.

https://www.mozdev.org/bugs/show_bug.cg ... AndLogIn=1 has a link to create a Bugzilla@Mozdev account.

[frisket]

I have been trying to do that for a week on that link. It claimed I didn't have an account under my email address, but when I try to create one, it says it's sending me a message but nothing ever arrives. I have tried this with three addresses, all current, active, and working, so the error is on mozdev's end. Whom can I contact to fix it?

[tanstaafl]

I tried using a gmail address three hours ago and never got a message. The web page states: "If it doesn't arrive within a reasonable amount of time, you can contact the maintainer of this Bugzilla@Mozdev installation at webmaster@mozdev.org." I suggest you do that.

[frisket]

I missed that, thanks.

[frisket]

I tried using a gmail address three hours ago and never got a message. The web page states: "If it doesn't arrive within a reasonable amount of time, you can contact the maintainer of this Bugzilla@Mozdev installation at webmaster@mozdev.org." I suggest you do that.

They seem to have some serious problems:

Code: Select all

<webmaster@mozdev.org>: connect to mozdev.mozdev.org[140.211.166.81]:25:
        Connection refused

Does anyone know how to contact them?

[frisket]

It gets worse. I emailed feedback@mozdev.com and that bounced back too:

<feedback@mozdev.org>: connect to mozdev.mozdev.org[140.211.166.81]:25:
Connection refused

Basically, there's no-one at mozdev responding to anything. Does anyone have any idea how to contact them? Is there even anyone there any more?

Peter

[tanstaafl]

No idea. I just sent a email to webmaster@mozdev.org mentioning the problems and including a link to this thread. It hasn't bounced after 5 minutes.

[frisket]

Thanks very much.

My original mail to webmaster@mozdev.org was sent on 31 Jan, and the rejection message arrived on 5 Feb.

My later email to feedback@mozdev.org was sent on 6 Feb, and the rejection message arrived on 11 Feb,

P