Evolution does not filter out spam in Ubuntu 5.10
Bug #24806 reported by
Irios
This bug report is a duplicate of:
Bug #13341: antispam filter doesn't installed by default.
Edit
Remove
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
evolution (Ubuntu) |
Invalid
|
Medium
|
Sebastien Bacher |
Bug Description
After a fresh install of Breezy, spam filtering does not work in evolution.
It requires some googling to find out that spamasassin has to be manually
installed for spam filtering to work. Previously, I have wasted a long time
trying to train a spam filter which was just not there.
Even after installing spamasassing, spam filtering does not work, even after
quite a bit of training.
There are a few threads in the web forum about this problem, but I have not
found any bug filed. Hope it is easy to solve.
Thanks for your dedication.
To post a comment you must log in.
Spam filtering is indeed broken in almost every way imaginable in Evolution:
1. Spamassassin isn't installed by default, and isn't even available in main.
Evolution takes spamassassin for granted, and doesn't even give an error message
when it is missing.
2. Spamassassin integration is broken, even if installed: As far as I can tell,
only the "Mark selected message as Junk" button actually does anything. "Mark
selected message as not being Junk" seems to be a no-op.
This means that bayesian spam filtering can never work. Spamassassin, as it is
configured in ubuntu, requires training with at least 200 spam and 200 ham
messages before the bayesian filter is activated. Evolution only lets you train
on spam, not ham.
So, in order to get bayesian filtering working, you need to call "sa-learn
--ham" on at least 200 non-junk mails from the commandline.
3. Even if you do that, you still won't get bayesian junk mail filtering.
Spamassassin, again as configured in ubuntu, assigns a maximum score of 4.07 to
messages messages classified as junk by the bayesian filter (the BAYES_99 rule).
At the same time, it doesn't consider a message as spam unless the score is above 5.
So, to get a working bayesian filter you'll need to add something like "score /user_prefs.
BAYES_99 5" to ~/.spamassassin
4. And if you do all this, and actually get spam filtering to work, you'll
notice the next bug: Evolution silently marks every spam-filtered message as
read. This makes it nearly impossible to track what Evolution is doing with the
(hopefully) junk mail.
5. Spamassassin also uses tremendous amounts of memory, and is very, very slow.
This seems to be, at least partly, because Evolution launches 5 (five) spamd
processes. Or rather, it should call spamd with "--max-children 1", to avoid the
bloat.
6. And while I'm whining: junk mail filtering needs to be turned on in two
different places in Evolutions maze of prefs: once in "Receiving Options" for
the account, and once in "Mail Preferences->Junk".
All in all, Evolution isn't exactly user friendly...