http://bugzilla.spamassassin.org/show_bug.cgi?id=3483
------- Additional Comments From felicity@kluge.net 2004-06-07 10:53 -------
Subject: Re: New: bayes _journal not syncing to _toks
On Mon, Jun 07, 2004 at 10:36:40AM -0700, bugzilla-daemon@bugzilla.spamassassin.org wrote:
> it looks to me like is_scan_available() is called, returns 0 because $nn < 200,
> therefore it 'goto skip', which means it never gets to do
>
> $self->opportunistic_calls();
>
> which is where we need to get for a sync_due() to run.
hrm. suboptimal.
> maybe moving opportunistic_calls() higher would help.. do they really need to
> run at the bottom of scan() ?
I have a ticket open to move opportunistic calls out of scan completely.
make spamassassin, spamd, etc, call it specifically when it wants to.
gets the opportunistic part away from the check part.
But for right now, the two are tied, and since the opportunistic part
is part of Bayes, it makes sense to do it in scan().
as for doing it at the end vs the beginning of scan()... bug 3345 moved
it to the end of a successful scan instead of the first part of skip: ...
so what it looks like is that we need to move opportunistic_calls()
back to the skip section, but have it able to check whether or not the
DB is available. I have an idea. patch in a minute.
> you will never have bayes. everything will learn to the _journal and it will
> grow and grow and grow (unless you --force-exire of course). at least thats
> how i see it, someone tell me i'm wrong :)
no, that's right. pre-3345, there was no problem, but ...
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
------- Additional Comments From felicity@kluge.net 2004-06-07 10:53 -------
Subject: Re: New: bayes _journal not syncing to _toks
On Mon, Jun 07, 2004 at 10:36:40AM -0700, bugzilla-daemon@bugzilla.spamassassin.org wrote:
> it looks to me like is_scan_available() is called, returns 0 because $nn < 200,
> therefore it 'goto skip', which means it never gets to do
>
> $self->opportunistic_calls();
>
> which is where we need to get for a sync_due() to run.
hrm. suboptimal.
> maybe moving opportunistic_calls() higher would help.. do they really need to
> run at the bottom of scan() ?
I have a ticket open to move opportunistic calls out of scan completely.
make spamassassin, spamd, etc, call it specifically when it wants to.
gets the opportunistic part away from the check part.
But for right now, the two are tied, and since the opportunistic part
is part of Bayes, it makes sense to do it in scan().
as for doing it at the end vs the beginning of scan()... bug 3345 moved
it to the end of a successful scan instead of the first part of skip: ...
so what it looks like is that we need to move opportunistic_calls()
back to the skip section, but have it able to check whether or not the
DB is available. I have an idea. patch in a minute.
> you will never have bayes. everything will learn to the _journal and it will
> grow and grow and grow (unless you --force-exire of course). at least thats
> how i see it, someone tell me i'm wrong :)
no, that's right. pre-3345, there was no problem, but ...
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.